Human Factors in Design, Engineering, and Computing

Editors: Tareq Ahram, Waldemar Karwowski, Jay Kalra
Topics: Artificial Intelligence & Computing, Human Systems Interaction
ISBN: 979-8-950676-14-7
DOI: 10.54941/ahfe1008234
Table of Contents
AI-Enhanced Live-Streamed English Language Learning: Designing and Evaluating an Intelligent Platform for Chinese Learners
Knowledge Sharing Live Streaming (KSLS) on Chinese commercial platforms such as Douyin (TikTok) and BiliBili is a popular choice for English language learning, offering attributes such as accessibility, low cost, and large-scale reach. However, a fundamental paradox exists: while effective language learning requires active engagement, real-time feedback, and meaningful interaction, KSLS is inherently a passive, one-to-many format that struggles to support these essential pedagogical elements. With AI technologies, particularly large language models, there is an opportunity to explore if AI can bridge this gap. A focus group study was conducted with 18 participants experienced in live-streamed English lessons. Thematic analysis identified two primary pain points: (1) uncertainty about pronunciation accuracy due to the absence of immediate corrective feedback, and (2) limited contextual speaking practice, as self-organised groups frequently collapsed without structure or topic guidance. Informed by these findings, an AI-integrated platform was designed with two key innovations: an AI-driven pronunciation feedback module providing real-time visual and auditory assessments, and AI-led conversational practice in breakout rooms where AI assumes four differentiated roles (mentor, expert, partner, servant) along two dimensions of closeness and guidance derived from social actor theory. The design was evaluated through Wizard-of-Oz experiments. For the pronunciation module, a within-subject study found that 63% of learners preferred multimodal (visual + auditory) feedback and 78% preferred sentence-level immediate feedback. For the AI roles, a between-subject study showed that the mentor role achieved the highest overall satisfaction (M=4.3/5), while the partner role scored highest on comfort (M=4.4/5). A subsequent deployment study using AI-powered chatbots further confirmed significant differences across conditions in engagement (F=10.521, p<0.05), vocabulary acquisition (F=81.213, p<0.05), and speaking proficiency across fluency, grammar, lexicon, and pronunciation. This study contributes to human-computer interaction and technology-enhanced language learning by providing empirical evidence of the KSLS paradox in English education, a theoretically grounded design framework, and experimental validation of AI integration in live-streamed language learning.
Nan Jiang, Chengcheng Ma, Peter Childs
Open Access
Article
Conference Proceedings
"They Will Move": Confirmation Bias, Expectation Frames, and Premature Monitoring Drop Off in COLREG Encounters
Collision avoidance depends not only on detecting cues but also on interpreting them against expectations about other vessels’ intentions, compliance with rules, and local traffic norms. This paper examines confirmation bias in maritime watchkeeping: the tendency to discount cues that contradict an initial expectation and reduce monitoring of a developing hazard. We analyse Applied Cognitive Task Analysis interviews from five structured simulator scenarios with seven experienced mariners. In 439 coded excerpts, 14 expectation-driven breakdowns (‘Over Trust Assumptions’) were identified, nine co-occurring with situation-awareness failures, particularly misinterpreted cues and weak projection. Qualitative themes show how reliance on local knowledge, stereotyped vessel behaviour, and presumed International Regulations for Preventing Collisions at Sea (COLREG) compliance can lead to premature closure and delayed corrective action. We identify design directions supporting ‘frame checking’ through explicit assumption states, anomaly cues tied to expected COLREG behaviour, and continued monitoring in stand-on situations.
Felix - Marcel Petermann, Ole Andreas Alsos, Jooyoung Park, Erik Styhr Petersen
Open Access
Article
Conference Proceedings
Automation transparency communication does not improve supervisory control performance, but shapes operator trust
With increased reliance on automated systems in supervisory control environments, such as unmanned aerial vehicle (UAV) missions, it is critical to understand what factors influence the human operator’s ability to interact effectively with these systems. The present study was designed to examine whether messages about potential problems with the automation for certain UAV sensor feeds influenced human operator behaviors and trust approach between systems. Two hundred and twenty-one participants, consisting of both Naval aviation enlistees and officers, engaged in a multi-UAV supervisory control task for 28 minutes. They were asked to monitor automated target detection in three UAV sensor feeds, which varied in reliability unbeknownst to the participants, while completing other tasks. All participants received communication about potential problems with two of the sensor feeds after completing seven minutes of the supervisory control task. Subjective assessments of automation trust were administered for each sensor feed immediately before the transparency communication and again at the very end of the task. After the UAV supervisory control task, individual difference measures of cognitive ability were collected. Results revealed that all participants showed some awareness of differences in system performance, but higher cognitive ability was associated with a more distinct pattern of trust over time. These findings demonstrate the importance of considering operator characteristics when designing transparency interventions and supervisory control systems.
Kaylin Strong, Bailee Smith, Jill Shelton, Shannon Mcgarry, Joseph Coyne, Michael Kukenberger, Noelle Brown
Open Access
Article
Conference Proceedings
The influence of neuroticism on startle and surprise response: a flight simulator study
This paper investigates the influence of personality traits on startle and surprise (S&S) responses in a realistic flight simulation. Thirteen pilots completed an A320 landing scenario during which a thunderclap and lightning flash (S&S event), and subsequent system failures were introduced. Participants also completed questionnaires assessing the Big Five personality traits, anxiety, perceived stress, and mental workload. Flight performance was measured using aircraft altitude deviations following the thunderclap and lightning flash. The exploratory results suggest that neuroticism was the only personality trait significantly associated with poorer flight performance, with more neurotic pilots exhibiting greater deviations following the S&S event. In contrast, flight experience, state and trait anxiety, perceived stress, and mental workload were not significantly related to flight performance. The study also suggests that pilots displayed a personality profile characterized by higher conscientiousness and lower neuroticism compared with a sample from the general French population. Although exploratory due to the small sample size, our findings suggest that stable personality traits, particularly neuroticism, may, under certain circumstances, play a role in determining pilots’ resilience to unexpected events.
Mickael Causse, Maud Dupuy, Alexandre Duchevet, Nadine Matton, Flora Schwartz, Jean-paul Imbert, Jonathan Deniel
Open Access
Article
Conference Proceedings
EEG-Based Classification of Reward, Penalty, and Movement Events Across Four Video Games of Varying Complexity
Longitudinal studies have tested video games as tools to alleviate mental health disorders such as depression and anxiety, with promising results. However, to fully assess the efficacy of video games for treating and diagnose mental health disorders, it is necessary to understand how they affect cognitive events in real time during self-paced video game interaction, and how reproducible these events are across different types of video game environments. This study characterises ERPs to three interactive mechanics: reward, penalty, and movement, across four games of increasing complexity, and tests whether the mechanics can be decoded from single-events. Seven healthy male participants played four video games of different complexity for approximately ten minutes in randomised order while 64-channel EEG was recorded. ERPs were examined over the FCz electrode and single-events were then classified with an EEGNet convolutional network across three binary tasks in cross-subject and within-subject settings. Reward events elicited a fronto-central positivity in the 200–500 ms range in every game, consistent with the reward positivity. Penalty events elicited fronto-central deflections in a comparable window but were less consistent across games. Movement events elicited a negative shift around key-press onset over fronto-central sites. All three binary classification tasks exceeded chance. Penalty versus movement was the most separable (73.3% pooled cross-subject; 81.0% best single game) and reward versus penalty the least (64.3% pooled). Within-subject models with subject-specific analysis windows performed better than cross-subject models. Both the ERP and classification results indicate that game context and individual differences shape these responses.
Laura Martinez-Tejada, Dechathron Dasri, Yuta Yamamura, Natsue Yoshimura
Open Access
Article
Conference Proceedings
The Effects of Task Similarity and Time Pressure on Task Performance Following Interruptions in Complex Human-Computer Interaction Systems
Task interruptions are common in complex human-computer interaction environments, particularly when users operate under time pressure or when interrupting tasks are similar to the primary task. Such interruptions may impair task performance and increase subjective workload, yet the combined effects of task similarity and time pressure remain insufficiently understood. The present study investigated their joint effects using a 2 × 2 within-subjects experimental design. Task similarity (high vs. low) and time pressure (present vs. absent) were manipulated, while response time, task accuracy, and subjective workload were measured. Task similarity and time pressure both showed significant main effects on the response time of the interrupted primary task, and both significantly increased subjective workload. No significant interaction was found for response time or task accuracy, whereas a significant interaction between task similarity and time pressure was observed for subjective workload. These findings suggest that task similarity and time pressure influence interruption performance largely through independent main effects, while their combined effect is evident primarily in subjective workload. These findings provide implications for interruption management and task design in complex human-computer interaction environments, with the aim of improving user performance and interaction efficiency.
Yuzheng Li, Weining Fang, Zhe Sun, Yirui Han
Open Access
Article
Conference Proceedings
Trusting the Chain, Not the Black Box: A Provenance-Based Architecture for Human-Machine Teaming in Military Autonomy
Military adoption of AI-enabled autonomy depends on more than algorithmic performance. Commanders and operators must be able to determine when an autonomous recommendation is trustworthy, when it is stale or insufficiently corroborated, who or what generated the underlying data, what authority governs the recommended action, and how the action can be audited after execution. This paper presents a human-machine teaming trust model for military autonomy based on accountable provenance rather than opaque model confidence alone. The architecture is an accountable autonomy substrate for multi-domain manned-unmanned teaming in contested environments. Signed mission records, provenance-preserving data custody, target-quality decay, corroboration-gated confirmation, and policy-bound human authorization create a shared operational record between humans and autonomous systems. Large sensor, map, and reconstruction products remain off-ledger behind signed cryptographic commitments, so a military user can inspect the chain of evidence behind a recommendation without every platform carrying exquisite sensing, navigation, or compute. The human-factors contribution is a trust-calibration framework organized around five operator questions: Can I verify the source? Was it altered? Is it current enough for the mission? Is the machine acting within authorized policy and command intent? Can I audit what happened afterward? An operator-facing trust card renders the answers. Simulation results serve as a case study: a stress test with 300 air and ground platforms showed a compact signed index governing an off-ledger data cloud roughly 600 times its size while rejecting tampered data, repairing missing fragments, and prioritizing synchronization under bandwidth limits. The approach shifts the operator's burden from trusting a black-box output to verifying an accountable chain of evidence and authority.
Richard Stone, Michael Jenkins, Joseph Fernandez
Open Access
Article
Conference Proceedings
Design of a Simplified Humanoid Robot Face: Cultural Differences and Emotional Synchronization
Humanoid robots are expected to interact with people in everyday environments, where facial expressions serve as an important nonverbal interface for conveying emotions and social intentions. However, highly human-like appearances can cause discomfort, a phenomenon known as the Uncanny Valley. To address this issue, this study investigates simplified robot faces that prioritize readability, emotional comfort, and the artifact-like nature of robots over full human resemblance. We examined basic principles of facial expression design from three perspectives: simplified facial elements, cultural differences in emotion recognition, and emotional synchronization. Expressions were built from only two components, eyes and a mouth, each rendered as a simple geometric shape and systematically deformed by parameters such as openness and inclination. In a questionnaire experiment, university students in Denmark (n = 80) and Japan (n = 51) classified the perceived emotion of each expression using Ekman’s six basic emotions plus two neutral categories. Happiness, anger, sadness, and neutral expressions were assigned to similar parameter configurations in both countries, suggesting that simplified geometric expressions may be interpreted consistently across these cultures. In contrast, disgust, fear, and surprise showed cultural differences, indicating that these emotions depend more on context. These findings suggest that existing emotion categories remain usable for highly reduced faces. We also introduce the concept of emotional synchronization, a shared emotional state between individuals that appears as matching facial expressions when two people feel the same way. An exploratory experiment using romance manga examined a method for observing emotional synchronization . Detecting such subtle expressions could ultimately allow robots to express and respond to emotions more naturally, making human–robot communication easier.
Meina Tawaki, Ichi Kanaya, Keiko Yamamoto
Open Access
Article
Conference Proceedings
From Visual Observation to Auditory Interaction: A Human-Centered Tangible Music System Based on Physical Objects
Visual information plays a central role in many contemporary interactive experiences, while sound is often treated as a supplementary feedback channel. This study presents Contour Sound Chamber, an object-based tangible interactive music system that explores how physical object properties and bodily actions can be transformed into auditory experiences. The system integrates AI-based object recognition, spatial sensing, sound mapping, and tangible interaction. Objects are classified according to material and color temperature, forming four predefined categories. Material determines the basic timbre, color temperature influences tonality and instrumental layers, and contour data acquired through ToF scanning are quantized into rhythmic patterns. Users further participate in the generated music through four tangible interaction modes: striking, plucking, pressing, and stroking, corresponding to different instrumental sounds. A user study with 34 participants evaluated object recognition, interaction task performance, and subjective perception. The system achieved a 97.1% first-attempt recognition accuracy across 136 trials, while the interaction tasks achieved an overall mean efficiency of 90.1% and a completion rate of 96.3%. The subjective evaluation showed an overall Likert score of 4.60/5. Participants were able to perceive differences between sounds generated by different objects, recognize the influence of their actions on musical output, and perceive certain object characteristics through sound. Together, these findings provide initial support for a perceptible relationship between object properties, sound representation, and user interaction, highlighting the potential of sound as an active component of multisensory interaction.
Yishan Hu, Liwen Liu, Qihang Yu, Sishuo Liu, Yiqi Wang
Open Access
Article
Conference Proceedings
The Impact of Scarcity-Based Dark Patterns on User Performance, Perceived Darkness, and Workload in Subscription Services
This study examines how different types of scarcity cues, a common form of dark pattern, affect user interaction with subscription platform interfaces. Three interface versions were designed based on a well-known subscription-based video streaming website, each presenting a different scarcity condition: time scarcity, resource scarcity, and no scarcity. The experimental task required participants to subscribe to the service. Thirty participants were recruited through purposive sampling and assigned to one of the three conditions in a between-subjects design. Task performance, perceived interface darkness, and cognitive load were assessed using task completion time, the System Darkness Scale (SDS), and the NASA Raw Task Load Index (NASA-RTLX), supplemented by observations and interviews. Task performance, SDS, and NASA-RTLX data were analyzed using one-way analysis of variance (ANOVA). The results indicate the following: (1) Both the time scarcity and resource scarcity conditions yielded better task performance than the no scarcity condition; however, participants reported disliking interacting with such dark interfaces under scarcity conditions. (2) The resource scarcity interface was perceived as the darkest; consistent with interview responses, these dark interface elements heightened users' uncertainty regarding both quantity and time. (3) The resource scarcity interface required greater effort to operate than the no scarcity interface. Future research may investigate factors such as users' purchase intention and subjective preferences toward scarcity-based interfaces. These findings may serve as a reference for researchers, designers, and practitioners in related fields.
Yihung Hsu, Chien-Hsiung Chen
Open Access
Article
Conference Proceedings
A Behavioral Observation Tool for Diagnosing Human–AI Interaction Profiles in Engineering Design Education
The integration of generative Artificial Intelligence into engineering design education has created a practical challenge for instructors: how to recognize, in real time, how individual students are interacting with AI systems during laboratory sessions. While empirical research has identified three recurrent Human–AI interaction profiles — Cognitive Adapters, Functional Executors, and Skeptical Observers — existing diagnostic approaches rely on survey instruments and post-hoc analysis that are unsuitable for classroom use. This paper proposes a Behavioral Observation Tool (BOT) designed for use by instructors during laboratory sessions. The BOT consists of 17 behavioral indicators organized across five dimensions — goal formulation, output interpretation, iterative behavior, agency distribution, and reflective engagement — rated on a three-point scale. A weighted scoring system produces normalized profile scores that classify students into dominant, mixed, or indeterminate profiles. Profile-specific instructional guidelines, derived from an adaptive Human–AI interaction framework, translate classification outcomes into concrete teaching responses. An Excel-based scoring instrument accompanies the tool, automating calculation and classification without requiring statistical expertise. The BOT bridges the gap between empirical interaction research and everyday instructional practice, supporting reflective, differentiated, and human-centered AI integration in engineering education.
Stefano Filippi
Open Access
Article
Conference Proceedings
The Mediated Moon: Semantic Mediation Between Gesture and Generative Image
This paper investigates how vision-language model (VLM)-based semantic mediation can reorganize the relationship between bodily input and real-time generative output in interactive art. The study develops a scaled interactive prototype from Luminous Night, an existing moon-inspired spatial light artwork. The system combines MediaPipe hand tracking, a locally deployed Gemma 3 model, and StreamDiffusion-based real-time image generation. A participant’s hand contributes both numerical tracking data and visual input; the VLM generates four semantic keywords within an artist-defined semantic field, which then participate in conditioning generative imagery presented through palm projection and LED output. The paper uses semantic mediation to describe this intermediate transformation from bodily input to language and image, and proposes the designed gap as an artistic design proposition for a partially indeterminate interaction in which bodily input remains influential without completely specifying the result. Observations made during prototype development provide an initial account of how participants adjusted and repeated gestures while encountering changing system responses. The contribution lies in articulating semantic mediation as an interaction configuration for generative art and establishing a practical basis for examining the relationship among bodily influence, generative variation, and perceived control.
Shu-Yu Lin, Jiaming Day, Derlor Way
Open Access
Article
Conference Proceedings
A Sustainable Assistive Technology for Children with Visual Impairments
Access to printed educational materials remains a significant challenge for children with visual impairments, particularly during the early stages of Braille literacy and foreign-language learning. This paper presents LIBRA (LEGO Interactive Braille Assistant), an exploratory assistive prototype designed to transform short printed English texts into embossed Braille while providing complementary auditory feedback. The system combines LEGO Mindstorms with an external processing module to perform optical character recognition, text processing, Braille generation, and audio output. LIBRA was developed using the iPlus methodology and iterative prototyping, with requirements focused on tactile interaction, Braille representation, auditory feedback, and adjustable learning support for children aged 7 to 10. The prototype was evaluated in terms of Braille output, processing performance, interaction, and perceived usability. The evaluation involved 25 participants with training in usability and accessibility and experts in assistive technologies and visual impairment. The results showed an average SUS score of 81%, together with evidence of functional feasibility and positive perceptions of the tactile and auditory interaction. The study contributes an exploratory, low-cost approach for integrating printed-text recognition, Braille embossing, and auditory support into an educational assistive system, providing a basis for future validation with children with visual impairments.
Mayra Carrión, David Morales-Martínez, Marco Santórum, Lisbeth Romo, Karen Santórum
Open Access
Article
Conference Proceedings
Physiological Indicators of Trust in AI-Assisted Air Traffic Control: A Multimodal Eye-Tracking and EEG Study
The increasing use of artificial intelligence (AI) in air traffic management requires operators to maintain appropriate trust in automated decision-support systems. Excessive trust may lead to insufficient monitoring, whereas insufficient trust may result in system disuse and increased workload. This study investigates whether eye-tracking and electroencephalography (EEG) signals can serve as objective indicators of trust during a simulated AI-assisted air traffic control task. Nineteen participants completed a simulated conflict-detection and resolution task with an AI advisor operating at low, high, and perfect reliability levels. After data-quality screening, 2,536 valid trials were retained. Eye-tracking and EEG signals, behavioural responses, and subjective trust ratings were recorded, and physiological features were subsequently extracted for classification. Trust classification was performed using Logistic Regression, Support Vector Machine, Random Forest, XGBoost, and a one-dimensional Convolutional Neural Network. Random Forest achieved the best overall performance, with eye-tracking features providing the highest classification accuracy (F1 = 0.759 ± 0.059). Low-trust trials were associated with longer fixation durations, greater pupil dilation, and elevated or less stable beta-band EEG activity. These findings demonstrate the feasibility of using non-invasive physiological sensing for trust monitoring in safety-critical human–AI collaboration. The results further suggest that robust and interpretable eye-tracking features may provide a practical foundation for real-time trust-aware adaptive decision-support systems.
Wenqian Qi, Xing Yao, Bufan Liu, Liu Xinyu, Hu Li, Tianjiao Zhang, Xiaoqing Yu
Open Access
Article
Conference Proceedings
Updated Comparative Analysis between Recent 3D Object Generation Models. Adaptation to Immersive Technologies
Artificial Intelligence-based Content Generation (AIGC) and Extended Reality (XR) enable scalable creation of immersive 3D assets, but single-image-to-3D models still involve trade-offs between quality, responsiveness, and runtime performance. This study evaluates two generative models, Trellis and TripoSR, across Animals, Objects, Humanoids, and Places categories. The main comparison uses original generated meshes, while humanoid assets are additionally analysed after polygon reduction through original, intermediate, and low-polygon versions. Metrics are grouped into generation complexity, geometric accuracy, texture fidelity, perceptual realism and semantic consistency, and VR performance. Results show that Trellis achieves the best reconstruction quality but requires approximately 30 seconds for generation. TripoSR provides the fastest generation and stable XR performance, while Trellis reaches the highest CLIP-Similarity, indicating stronger semantic alignment but heavier meshes. These findings show that model selection for XR depends on the target application, balancing reconstruction quality, semantic consistency, and runtime constraints.
Geovana Amorim Abensur, Agustín A Ortiz Diaz, Sergio Cleger Tamayo, Delrick De Oliveira Freitas, Paulo Sergio Da Silva Farias, Gilberto Oliveira Neto, Gabriel Rossi Zanini
Open Access
Article
Conference Proceedings
Evaluation of Gesture Separability and Stability Based on Surface Electromyography for General Interaction Design
With the increasing demand for natural human-computer interaction, surface electromyography (sEMG)-based gesture recognition has become an important topic in intelligent interfaces and wearable technologies. Compared with conventional input methods, sEMG provides a non-invasive and relatively responsive means of capturing muscle activation associated with movement intention, which makes it suitable for virtual reality control, rehabilitation support, and assistive systems. In practical applications, however, system performance depends not only on the recognition algorithm, but also on the inherent separability of the selected gestures and their consistency across users. This study investigates the sEMG characteristics of representative hand gestures using the public NinaPro DB1 dataset (Atzori et al. 2014), with a focus on gesture separability and inter-subject stability. Root Mean Square (RMS) features are extracted from three subjects and visualized using Principal Component Analysis (PCA) to examine the distribution of six representative gestures (Jolliffe and Cadima, 2016). In addition, boxplot analysis is used to compare RMS variations across subjects. The results suggest that large-scale wrist motions, such as wrist flexion and wrist extension, show clearer clustering patterns and greater consistency both within and across subjects. In contrast, fine finger gestures, such as fist clenching and interphalangeal pinch, exhibit stronger overlap and larger inter-subject variability. Based on these findings, this study proposes a gesture selection strategy for generalized sEMG interaction systems, emphasizing gestures with higher separability and stability while suggesting supplementary features or user calibration for more ambiguous gestures.
Hao Shi, Yaqi Li
Open Access
Article
Conference Proceedings
SPARC: An Interactive Virtual Environment for Eliciting and Organizing Place-Based Memories in Historic Districts
Memories of residents connect historic districts with everyday activities, social relationships, and change. Changing uses, names, and landmarks, however, complicate the identification and organization of places and periods in oral accounts. Developed for Lingshui Old Street, China, SPARC (Spatial Platform for Assisted Recollection and Collection) combines a high-fidelity street-block environment, spatiotemporal slices, and information entry and processing. Interviews and guided trials with 12 current or former residents and feedback from four researchers were analyzed qualitatively. Discussion in the virtual environment elicited memories and recognition cues. Residents identified discrepancies for historical verification, while researchers clarified locations, floors, and uses and compared accounts. SPARC thus provided a shared spatial reference connecting recollection with information organization. Linking accounts to places, times, uses, and people, while retaining sources and unresolved information, provides a foundation for investigating heritage value.
Tianyi Liu, Zaixun Chen, Quhang Wu, Huishu Deng
Open Access
Article
Conference Proceedings
Toward a Human-Centered Framework for Digital Health Accessibility Among ASL-Dominant Users
Digital technologies are increasingly important in healthcare; however, ASL-dominant users still face accessibility and usability challenges when accessing English-dominant digital health interfaces. Existing accessibility standards, including Section 508 and the Web Content Accessibility Guidelines (WCAG), primarily emphasize sensory accessibility and technical compliance. Still, they do not adequately address the linguistic, cognitive and interactive challenges that ASL-dominant users face when interacting with English-dominant digital health interfaces. This paper presents the design of an ongoing research study involving ASL-dominant users to develop and evaluate an accessibility framework for English-dominant digital health interfaces: (1) participatory co-design activities involving ten ASL dominant participants; (2) a planned expert review of the resulting prototype by five HCI, accessibility, and ASL experts; and (3) a planned final independent review of a refined prototype by five ASL-dominant users. In addition to the study design, data and findings from the first stage were reported. Preliminary findings from these initial sessions emphasized key accessibility needs related to simplified language, clear navigation and interaction pathways, reduced cognitive load, visual and multimodal support, ASL-aligned accessibility, and step-by-step procedural guidance, going beyond linguistic adaptation. These preliminary findings provide empirically grounded, Deaf-centered design guidance for creating more inclusive digital health environments that address the linguistic, cognitive, and interaction needs of ASL-dominant users.
Olarinde Farayola, Kathy Wang
Open Access
Article
Conference Proceedings
The Impact of Gamification Elements in Art Therapy on Behavioral Intention among University Students with Trait Anxiety: A Self-Determination Theory Perspective
Trait anxiety is a relatively stable tendency to interpret situations as threatening, leading to persistent worry and unease. For university students navigating academic, interpersonal, and career pressures, accessible early emotional support is paramount. While art therapy leverages creative media (e.g., painting, music, and audio-visual interaction) to facilitate nonverbal expression and emotion regulation, sustaining engagement remains challenging. Gamification can provide structure and enhance adherence; however, its specific impact on the acceptance of gamified art therapy among trait-anxious students remains underexplored. Drawing on the Stimulus-Organism-Response (SOR) model and Self-Determination Theory (SDT), this study investigates how achievement-, immersion-, and social-related gamification elements drive behavioural intention by fulfilling the basic psychological needs of autonomy, competence, and relatedness. Using partial least squares structural equation modelling (PLS-SEM), we analysed survey data from 250 trait-anxious university students. Results supported all hypothesized paths. Notably, immersion-related elements exhibited the strongest association with autonomy, which in turn was the primary predictor of behavioural intention. These findings underscore that gamified art therapy should prioritize immersive contexts, user autonomy, supportive feedback, and low-pressure social interactions to effectively address the psychological needs of trait-anxious students.
Yingjiao Jia, Hsin-Hsien Chiu
Open Access
Article
Conference Proceedings
Customer Churn Prediction for Consumer-Facing Delivery Apps
Customer disengagement in non-contractual on-demand services must be inferred from activity rather than explicit cancellation, motivating repeated risk assessment. This study proposes a rolling-window prediction and decision-support framework using 401,164 bookings from 101,765 customers of a doorstep car-wash platform. Matched 30/30-, 15/15-, and 7/7-day observation/evaluation windows generated customer states with 54 behavioral and operational features. Logistic Regression, Random Forest, XGBoost, LSTM, GRU, and Temporal ResNet were tuned and evaluated using customer-grouped partitions, with recall prioritized. LSTM achieved the highest recall at 30, 15, and 7 days (96.0%, 96.4%, and 90.9%), while the highest ROC-AUC values were 0.944, 0.951, and 0.954. Retrospective 30-day analysis identified engagement-depth-specific behavioral patterns, including service failure (65.8%), promotion dependence (52.7%), inactivity (62.0%), and subscription loss with price burden (50.2%), which informed candidate retention actions. External Instacart replication achieved 94% recall and 0.9777 ROC-AUC with 30-day XGBoost, providing preliminary evidence of cross-dataset applicability.
Muhammad Jawad Mufti, Omar Hammad, Haitham Saleh, Muqaddas Gull
Open Access
Article
Conference Proceedings
Exploring User Perceptions of their Role in Personalization on YouTube
Recommender systems (RS) increasingly rely on user behavior and feedback to provide personalized recommendations, positioning users as participants in the personalization process. However, users may differ substantially in how they understand their role in this process and the extent to which they believe they can influence recommendation outcomes. Despite users’ growing involvement in personalization, little is known about how they perceive their role and what factors influence these perceptions. Using a content analysis of semi-structured interviews with 21 YouTube users, we explore how users perceive their role in personalization on YouTube. Our Analysis revealed four distinct perceived user roles in personalization, including Active Trainers, Contributors, Collaborators and Passive Consumers. Participants’ perceptions of these roles were shaped by differences in their understanding of personalization and control mechanisms, reliance on recommendations, and beliefs about responsibility for personalization. By identifying distinct perceived user roles and the factors that shape them, this work provides insights into how users understand and participate in personalization. Based on our findings, we discuss implications for designing recommender systems that support diverse levels of user involvement and more transparent forms of user control.
Mohammed Muheeb Faizan Ghori, Arman Dehpanah, Bamshad Mobasher
Open Access
Article
Conference Proceedings
Qualitative study updated with recent 3D object generation models for immersive technologies
As a meeting point between AI-generated content and extended reality (XR), 3D generative content has rapidly evolved from text and image synthesis to the creation of 3D objects. This article investigates user perceptions of two 3D generative models—Trellis and TripoSR—which operate from a single image within an XR environment using a head-mounted display. Employing a qualitative analysis methodology with 21 evaluators, the study provides valuable insights into overall 3D object quality, image-to-object correspondence, and user satisfaction regarding generation time, highlighting the models' potential for immersive applications. The two models were tested using a minimal image dataset categorized into four groups: humanoids, animals, places, and objects. A direct comparison of the two models yielded clear results: users noted that the Trellis model provided a significantly more satisfying experience than the TripoSR model.
Pio Ordozgoith, Gilberto Oliveira Neto, Agustín A Ortiz Diaz, Sergio Cleger Tamayo, Heider Betcel Lima De Souza, Geovana Amorim Abensur, Delrick De Oliveira Freitas, Gustavo De Freitas Martins, Osvaldo Junior
Open Access
Article
Conference Proceedings
Beyond Immersion: The Influence of Interactivity on User Experience in VR-Based Robotics Learning
This study presents an empirical evaluation of an immersive virtual reality experience designed to support learning and training in industrial robotics. The experience comprised three modules combining playful introduction, guided conceptual learning, and simulation-based problem-solving. Ninety-one participants were distributed across the modules. Usability, intrinsic motivation, simulator sickness, perceived workload, recommendation intention, XR experience quality, and open-ended feedback were assessed. Results indicated an overall positive reception, with usability scores within an acceptable-to-good range and no statistically significant differences between modules. The motivational profile, including Interest/Enjoyment, Perceived Competence, and Value/Usefulness, was also comparable across modules. Nevertheless, the simulation module showed slightly higher descriptive tendencies in Interest/Enjoyment and recommendation intention. Simulator sickness symptoms were low for most participants, although isolated cases of discomfort occurred. Workload results suggested that the simulation module required greater cognitive involvement without increasing frustration or temporal pressure. Qualitative feedback showed that participants particularly valued interactivity, hands-on exploration, and 3D visualization, whereas more passive instructional segments generated more suggestions for improvement. The findings indicate that the value of VR in industrial robotics training does not derive from immersion alone, but from the integration of meaningful interaction, narrative purpose, instructional clarity, user comfort, and active experiential learning.
Iara Margolis, Eduardo Santos, Francisco Silva, Carlos Costa, Miguel Nunes, Alexandre Ferreira, Andreia Mendes, Alexandre Carrança, Arabela Fabre, Diogo Silva, Carolina Pimenta, Ana Alves, José Rocha, Nuno Sousa
Open Access
Article
Conference Proceedings
Predicting Workplace Conflict Through Multimodal AI-Based Behavioral Analysis: An Ethical and Cognitive Framework
Workplace conflict is frequently managed as an operational issue addressed only after formal complaints or visible disputes arise. This reactive approach fails to account for the gradual cognitive and behavioral processes that underlie interpersonal tension. This paper addresses the central research question: How do managers perceive the effectiveness, acceptability, and practical challenges of using AI-driven multimodal analysis for early workplace conflict detection? To investigate this question, we propose a multimodal AI-powered framework for the early prediction of workplace conflict by analyzing behavioral patterns across multiple data modalities, including facial expression recognition, body language analysis, and text-based communication indicators such as sentiment shifts, response latency, and collaboration network changes. Drawing on theoretical foundations in affective computing, computer vision, and cognitive science, we present an integrated model that tracks temporal behavioral escalation rather than static, single-instance observations. We further address the critical ethical dimensions of such a system, including privacy preservation, algorithmic bias, and the necessity of human-centered design principles. Qualitative data were gathered through semi-structured interviews with 11 participants spanning managerial, technical, and frontline professional roles, guided by seven research hypotheses. Interview responses reveal a nuanced landscape of conditional acceptance: participants broadly support temporally aware, multimodal detection approaches while expressing strong reservations about facial-expression-only systems, raising consistent concerns around privacy, cultural bias, and the risk of managerial misuse. The paper concludes with recommendations for responsible, transparency-first deployment of AI conflict-prediction systems in organizational environments.
Damina Townes, Rushabh Gandhi, Apoorv Singh, Azhar Mohammed Abdul Haleem, Youngsoo Shin
Open Access
Article
Conference Proceedings
Measuring Mental Workload in Immersive Fighter-Flight Simulation: A Multimodal Approach
Workload indicators in fighter aviation rarely agree, and it is not known in advance which one to trust. This study examined how subjective, performance and oculometric indicators associate, dissociate or remain insensitive in immersive fighter-flight simulation. Fifty-five military pilot trainees flew a virtual-reality mission crossing three flight phases (autopilot, low-altitude terrain following, air defense) with three secondary-task levels (memorization alone, or with one or two spatial decisions every 30 s), a 3 × 3 within-subjects design. Perceived workload (NASA-TLX) rose with both factors. Near-display gaze was the most sensitive indicator of task level, separating all three levels even outside the decision windows, yet it fell across flight phases, moving opposite to perceived workload on that factor. Decision performance did not change: Bayesian analyses favored the absence of an effect for every term except task level on accuracy, where they were inconclusive, and performance was stable at a modest level, with little room to degrade. SWORD ratings and gaze both revealed effort shifting toward the decision task. Recall of a concurrent memorization task registered the demand of flying (confounded with time on task), and the added decisions only in autopilot, where capacity remained to be consumed. Pupil diameter decreased as demand increased, a reversal attributable to the light emitted by the decision task rather than to cognitive load. Each indicator thus answered a different question, and which one depended on the factor varied. No single measure suffices in a realistic cockpit; this study maps what each one answers, for which factor, and when it can be read at all.
Jean-christophe Hurault, Marianne Jarry, Gregory Froger, Anne-lise Marchand, Colin Blättler
Open Access
Article
Conference Proceedings
Zili Patrol Robot: A Human–Robot Collaborative Inspection System for Sub-Canopy Orchards
Dense canopies, slopes, gullies, and root obstacles in hilly orchards prevent traditional manual inspection, drones, and wheeled robots from reliably reaching leaf undersides, stems, and occluded near-ground areas. Drawing on embodied intelligence, this study proposes a sensory extension paradigm through a quadruped inspection robot, Zili Patrol Robot, and a companion app. Together, they form an embodiment–intelligence–environment coupled orchard inspection system that makes the robot a perceptual medium for farmers rather than their replacement. A single-group simulated system test was conducted in a laboratory orchard. The system uses a Mini Pupper 2 quadruped robot equipped with 2D LiDAR and an RGB camera. Simultaneous localisation and mapping (SLAM) supports real-time mapping and autonomous navigation, while colour-block recognition triggers anomaly detection. Anomaly coordinates, image evidence, and inspection reports are synchronised to the mobile app. Across 15 prototype trials, cumulative pass rates for five sequential stages were 100% (autonomous mapping), 93.3% (inspection-route navigation), 86.7% (obstacle avoidance), 73.3% (disease-anomaly triggering) and 73.3% (report generation).The full workflow from mapping to reporting was completed in the controlled setting, but only 73.3% of trials reached the final stage, and triggering remained sensitive to low light, occlusion, and targets at the edge of the camera's field of view. This study advocates embodied sensory extension instead of automated substitution. The robot extends the farmer's vision, mobility, and capacity for structured recording. Both diagnostic interpretation and intervention decisions remain with the farmer.
Jianan Li
Open Access
Article
Conference Proceedings
Focus Pup: A Design Study on a Companion Robot for Self-Management Among Adults with ADHD
Adult Attention-Deficit Hyperactivity Disorder (ADHD) is a condition characterised by poor attentional stability, impaired executive function and weak self-management ability. Most existing non-pharmacological interventions rely on external mandatory constraints as their core logic, and lack sustainable intervention approaches integrating physiological and psychological dimensions.This study explores attentional and self-management traits in adults with ADHD, constructing a two-dimensional programme to boost self-management executive capacity based on flow theory. First, a literature review establishes the theoretical foundations of flow theory and ADHD cognitive intervention. Second, semi-structured interviews and naturalistic behavioural observations are carried out with clinically diagnosed adults with ADHD to identify core attention pain points and self-management demands. Building on these findings, this work systematically maps flow theory to design practice, proposes Focus Pup — an intelligent emotional companion device — and develops its hardware prototype. A between-groups controlled experiment is planned to examine the solution’s efficacy, combining data from the Flow State Scale (FSS), NASA Task Load Index (NASA-TLX) and behavioural observations via quantitative statistics and qualitative analysis.This paper elaborates on the function of flow theory in cultivating self-management capacity for people with ADHD, and proposes that physical isolation, physiological soothing and instant feedback may serve as external compensations for executive function deficits, with their effects to be examined in a planned between-groups controlled experiment. Empirically, it supplements real-world evidence describing adults’ ADHD attention and self-management patterns. Theoretically, it broadens the application scope of flow theory within neurodiversity intervention research. Practically, it offers design references for tangible interaction and attention intervention tailored to ADHD groups in human-computer interaction (HCI).
Zilin Xiong, Sidi Yang
Open Access
Article
Conference Proceedings
E Contextibus Unum: When Should Your AI Share Your Information with Others?
As large language models (LLMs) become increasingly integrated into collaborative environments, questions arise about how user contexts such as prompts, conversation histories, and preferences might be shared across users to enhance collective intelligence and personalization. Although technical implementations of context sharing have been explored, it remains unclear how to design interfaces that appropriately engage users with context sharing. This study investigates how users perceive the trade-offs between usefulness, privacy, and control when their contexts are shared with other users through LLM-mediated interactions. Through a Wizard-of-Oz study with scenario-based probes, we examine both the sharer’s perspective (those whose context is being used) and the receiver’s perspective (those benefiting from others’ contexts). Our work aims to surface guidelines for designing LLM interfaces for transparent, user-centered context sharing systems that balance collective benefit with individual privacy and autonomy. This provides engineers with opportunities to design LLM assistants that leverage aggregated user data to provide useful insights. Our key findings reveal that: (1) context sharing acceptability varies by domain: professional contexts showed high acceptance while health contexts demanded expert authority over community insights, and mental health contexts produced polarized responses based on personal disclosure comfort; (2) most participants preferred statistical aggregations over individual anecdotes, demonstrating an LLM-specific trust mechanism where data-driven validation matters more than personal stories; and (3) control is essential for acceptance, but must avoid decision fatigue through domain-based preferences and keyword rules rather than requiring per-interaction consent decisions.
Enya Bullard, Xiang Chen
Open Access
Article
Conference Proceedings
A Lifelog System for Automatically Linking Physical Operations and Measurement Data in Hardware Debugging
In hardware debugging, waveforms observed on measuring instruments must be interpreted together with the physical context of the measurement, such as which pin was probed or which circuit change had just been made. Manual note-taking preserves this context but interrupts the debugging process. We developed a lifelog system that detects the worker’s contact with an oscilloscope using an externally attached piezoelectric sensor and, upon each detected event, automatically captures the waveform via SCPI and extracts a 10-second first-person-view (FPV) video clip centered on the event (5 s before and 5 s after). In a within-subjects pilot study with five undergraduate students, four of five participants completed the debugging workflow, including documentation, more quickly with the proposed system (mean 389.0 s vs. 564.8 s). Mean ratings on three subjective workload items adapted from NASA-TLX were lower with the proposed system, although no statistically reliable differences were observed. In contrast, all five participants took longer to identify the measurement context of a waveform during review with the video logs than with handwritten notes. These observations suggest that automated capture may reduce documentation burden during debugging while shifting part of that burden to later retrieval. We discuss the implication that capture automation should be accompanied by retrieval-oriented interface design, such as an event-centered interface that integrates waveforms and FPV clips.
Kairi Miyazaki, Masato Yazawa, Yutaro Takeuchi, Norihisa Segawa
Open Access
Article
Conference Proceedings
Wilderness Vanguard: Design and Implementation of a Quadruped-Robot-Based Proactive Terrain Verification System for Mitigating Visual-Deception Risks in Outdoor Hiking
With the growing popularity of outdoor exploration, falls and missteps caused by visually deceptive terrain have become a major safety concern for hikers. This study proposes a quadruped-robot-based proactive terrain verification system and investigates the role of multimodal feedback under high cognitive load. Based on a Risk Proxy design concept, the robot physically probes potentially hazardous terrain before the user proceeds. The prototype was developed through three iterative sensor configurations. V1.0 employed only 2D LiDAR; V2.0 added an OAK-D depth camera; and V3.0 integrated LiDAR-based topographic ranging, depth-based ground geometry detection, and RGB image recognition into a multi-sensor fusion architecture. Robot kinematics were also optimized by adjusting body inclination, friction, and centre-of-gravity parameters to improve locomotion over uneven terrain. V3.0 addressed the limitations of individual sensing modalities, in particular the inability of LiDAR-only sensing to detect cavities beneath a visually continuous surface. Detection performance has not yet been systematically quantified; a formal evaluation protocol is reported as part of the planned validation work. Based on this configuration, Wilderness Vanguard integrates a quadruped robot, a mobile application, and a smartwatch-based haptic interface to support pre-trip preparation, on-trail monitoring, and hazard warning. A planned controlled experiment will further evaluate whether multimodal, particularly high-frequency haptic, feedback improves reaction time, cognitive load, and system trust under simulated high-pressure conditions. The study aims to shift outdoor safety from reactive response to proactive prevention, while providing a new approach to wearable interaction and human–robot collaboration in high-risk environments.Keywords: proactive terrain verification; quadruped robot; risk proxy; multimodal haptic feedback; visual deception; human–robot collaborative wearable system
Haozhe Ma
Open Access
Article
Conference Proceedings
Design Innovation of a Remotely-Controllable Portable Door Lock for Physical Security
Portable door locks are personal security mechanisms designed to secure a door independently of its installed hardware. These devices are commonly advertised to be used in commercial applications such as hotel stays, short-term rentals, and shared residential environments where users seek to supplement existing lock systems. In addition to general travel and privacy use cases, portable locks have grown to be employed in emergency scenarios, including active threat incidents and domestic safety situations. Most existing designs require the user to be located inside the secured space to deploy the portable door lock, which limits functionality in scenarios requiring external locking or unattended room security. An infrastructure-agnostic, remotely-actuated portable locking device was developed to address this operational gap. The system interfaces directly with standard residential and commercial strike plates without modification to the door or frame, and functions independently of any pre-installed lockset. All mechanical and electronic components reside entirely on the interior side of the door, with actuation enabled via remote wireless communication. No exterior-facing elements are present, minimizing detectability and physical vulnerability. Utilizing the Design Innovation methodology, we incorporated methods such as user-centered requirements gathering, scenario analyses, and iterative prototyping. Primary operational scenarios included personal use in hotels, tenant-controlled room security in shared housing, and temporary field deployment for government or military personnel. The device architecture includes physical anti-tamper features, concealed input mechanisms, wired power option with battery back-up, tamper detection, and other design features to resist manipulation, unauthorized access, and signal spoofing. This system expands the design space of portable door locks by enabling unattended, remote-controlled physical security in non-permissive or infrastructure-limited environments.
John McIntire, Matt Srnoyachki, Jason Sopko, Adam Adducchio
Open Access
Article
Conference Proceedings
Flying While Coordinating: A Work Domain Analysis of Goggles-Only FPV Pilots as Nodes in a Scalable Strike Element
This paper presents a Work Domain Analysis of coordinated military first-person-view (FPV) small uncrewed aircraft operations, framed from the viewpoint of a single strike pilot who must simultaneously fly an immersive analog video feed and act as one node within an element of dozens to hundreds of 1:1 pilot-drone pairs. Drawing on Rasmussen's Abstraction Hierarchy and the Cognitive Work Analysis tradition formalized by Vicente and by Burns and Hajdukiewicz, we construct a five-level Abstraction-Decomposition Space of 70 nodes and 165 means-ends links, spanning functional purposes (coordinated mass strike, immersive flight fidelity, resilient shared picture, signature survivability, auditable human authority) through abstract functions (finite attention budget, latency and throughput conservation, kill-chain integrity) to the physical functions and forms of goggles, body-worn radios, common operating picture systems, and dispersed team roles. The analysis foregrounds GNSS denial, localized electronic-warfare jamming bubbles, tens-of-milliseconds latency margins, network saturation, and the competition between immersion and coordination awareness for one operator's attention, and finds that current interfaces externalize coordination almost entirely onto an adjacent navigator, leaving the in-goggle pilot a perceptually impoverished node. We derive ecological interface design requirements that push filtered, latency-bounded coordination cues into the field of view without disrupting terminal flight, and present a front-end prototype of a monitoring-class goggles overlay whose configural, emergent encodings—a route-conflict glyph, a coordination-budget polygon, freshness halos with a picture-confidence band, and a human-authority kill-chain ribbon—are each traced back to specific Abstraction Hierarchy nodes.
Michael Jenkins
Open Access
Article
Conference Proceedings
Comparison of Threshold-Based and Continuous Measurement Methods for Assessing Wrist and Finger Posture During Hand-Controller Interaction
Background. Hand-controller ergonomics are important in precision-control applications, including robot-assisted surgery, but comparisons depend on how wrist and finger posture is measured. This study examined whether threshold-based and continuous measures produced similar controller ergonomic comparisons and whether wrist-only assessment represented hand postural exposure. Methods. Twelve participants completed 1-min and 5-min trials of a game using six controllers. Wrist and finger angles were recorded using a validated wearable data glove. Outcomes included mean absolute and 95th-percentile joint angles, time beyond wrist and finger reference postures, and wrist-only, finger-only and simultaneous wrist-finger exposure. Mixed-effects models included controller, trial length and their interaction, with participant as a random intercept. Associations were examined using repeated-measures correlation. Results. Controller type significantly affected all eight kinematic outcomes, χ²(5) ≥ 40.89, p< .001. Wrist threshold exposure was strongly associated with mean absolute wrist angle (rᵣₘ = .89) and moderately associated with 95th-percentile wrist angle (rᵣₘ = .67). Associations were weaker for the fingers (rᵣₘ = .63 and .38, respectively; all p < .001). Threshold exposure and mean absolute angle produced the same controller ranking for the wrist (ρ = 1.00) and a similar ranking for the fingers (ρ = .89). Across all trials, 67.6% of threshold-defined exposure occurred in the fingers without concurrent wrist exposure. Conclusion. Threshold exposure and mean absolute angle produced similar controller comparisons, particularly for the wrist. Wrist-only assessment missed substantial finger exposure. Hand-controller assessments should include wrist and finger kinematics and select outcomes according to whether the aim is to describe exposure duration, overall posture or peak posture.
Olivia Hosie, Mats Isaksson, John McCormick, Oren Tirosh, Chrys Hensman
Open Access
Article
Conference Proceedings
Purported and Measured Waveform Characteristics of Vagus Nerve Stimulation Devices
In this work, we conduct a first-of-its-kind comparative evaluation of commercially available non-invasive vagus nerve stimulation (nVNS) devices, providing detailed waveform characteristics for the devices’ outputs and comparing the purported waveform characteristics with those from laboratory measurements. The vagus nerve travels from the brainstem through the neck and has connections to many major organs, such as the heart and stomach, via the parasympathetic nervous system. There is a rising prevalence of VNS technology aimed at improving health. Wearable nVNS devices deliver (output) either electrical current, acoustic, or magnetic signals through the skin to the vagus nerve near either the ear, neck, skull, or chest cavity. While the development and use of these devices have increased, there is very little published data on specific characteristics of the waveforms these devices produce. This paper aims to conduct a review of 7 popular commercial off-the-shelf nVNS devices by providing the product details, purported waveform specifications, and measured waveform specifications for each device. The devices we investigated are the gammaCore®, Neuvana Xen®, Pulsetto Fit®, Vagustim V2®, ZenoWell Vita®, VeRelief Mini®, and the AmofitS®. The measured waveform specifications were obtained using an oscilloscope to determine the waveform characteristics. These are compared with the purported waveform characteristics. Notable inconsistencies are addressed. The detailed waveform characteristics could be used in future studies to correlate the waveform variables with specific treatment success.
Lindsey Mcintire, John McIntire, Ben Mandani, Gideon Farhadian, Jackson Tell, Kassandra Guinto, Lukas Goodworth, Will Allison, Daniel Jensen
Open Access
Article
Conference Proceedings
Effects of Customer-Side Environmental Noise on Interactions in Avatar-Mediated Remote Customer Service
Our research group has been investigating factors contributing to comfortable working conditions for operators providing avatar-mediated remote customer service. However, the effects of customer-side environmental noise on these interactions remain unclear. This study was focused on speech comprehension and investigating whether customer-side environmental noise interferes with service interactions. To this end, we reproduced an airport crowd-noise environment and examined its effects on remote customer service interactions. Mock customer service interactions were conducted between customers and a remote operator. Ten participants (five men and five women) served as customers, while one operator performed all customer service tasks. All experimental trials were conducted in a soundproof room. The acoustic environment was created using a digital mixer and two loudspeakers and calibrated to provide a flat frequency response. Objective measures included recorded speech data and the operator’s gaze behavior during the interactions. In addition, subjective evaluations of customer service quality (0‒100 points) and self-reported stress (0‒100 points) were obtained after each trial. The results showed that both customers and the operator tended to experience higher subjective stress levels under noisy conditions. Furthermore, under noisy conditions, subjective stress was apparently influenced more strongly by listening effort and difficulties in verbal communication than by interpersonal distance itself. These findings suggested that customer-side environmental noise should be considered when designing avatar-mediated customer service systems. Based on these findings, guidelines for avatar robot placement and service-environment design in practical service settings should be developed in future studies.
Manabu Chikai, Kentaro Watanabe, Bach Quang Ho, Jooho Park, Yui Murakami, Kayo Koike, Min Ma, Masahiro Tsutsu
Open Access
Article
Conference Proceedings
A Design Methodology for Explainable Supervisory Interfaces Supporting Situation Awareness and Human-AI Collaboration in Augmented SOCs
The increasing integration of Artificial Intelligence (AI) agents into monitoring environments, such as Security Operations Centers (SOCs), is significantly transforming analysis, decision-making, and incident response activities. While these agents allow the automation of certain processes, they also raise new challenges regarding transparency, oversight, and human–AI coordination. Operators must be able to understand the status and behavior of AI agents, oversee their actions, adjust task delegation, and maintain awareness of their own ability to intervene. In this paper, we propose a methodology to design an explainable monitoring dashboard intended to support interaction between a human operator and a set of AI agents. The design principles are based on the concepts of Situation Awareness and Human-AI Teaming, as well as recommendations from research in Human-AI Interaction. The methodology is applied to the context of augmented SOC where an analyst monitors and analyzes security events in real time to identify potential threats, assess alerts, and determine appropriate responses. More specifically, the dashboard is structured around four complementary dimensions: (i) visualizing the operational status of the AI agents, (ii) maintaining a shared understanding of the collaborative activity between humans and the agents, (iii) supporting adjustable task delegation through supervision and reallocation mechanisms, and (iv) taking into account the operator’s cognitive and operational state. This approach leads to the definition of an interface that incorporates mechanisms for transparency, explainability, and human control, aimed at improving understanding of AI agents’ behavior while keeping the operator at the center of the decision-making loop
Kahina Amokrane, Reda YAICH, Emeline DURIMEL, Guillaume Giraud, Christelle FOUQUIAU
Open Access
Article
Conference Proceedings
The Invisible Footprint: Awareness, Attitudes, and Designable Solutions for Sustainable Human-AI Interaction
What people know about an AI system shapes their relationship with it just as much as what that system can do. As generative AI (GenAI) becomes embedded in student life, an open question in human-AI interaction is how new information, including a system's environmental footprint, reshapes that relationship. This study tested whether a short micro-learning intervention could shift students' attitudes, subjective norms, perceived behavioral control, and intentions toward more sustainable use of GenAI. Using a two-arm design, 125 undergraduate and graduate students across four universities were randomly assigned to view a five-minute video on GenAI's water, energy, and hardware costs either before or after completing surveys measuring these constructs. Following the intervention, students showed large gains in perceived knowledge and behavioral intention, and a moderate gain in perceived behavioral control; attitudes and subjective norms did not change significantly. A regression model explained 72% of the variance in intention, with attitudes being the strongest predictor and knowledge showing no independent effect. These findings suggest that informing students about GenAI's environmental costs can measurably shift how they relate to the technology. That attitude change drives intention to act, not knowledge alone, which is a finding with direct implications for human-AI interaction design. The results extend responsible-AI discourse to include environmental stewardship, pointing to concrete strategies such as usage feedback, eco-mode defaults, and AI-literacy curricula to embed sustainability in everyday human-AI interaction.
Jasmine Mckenzie, Mark Holden, Wesley Junkins, Alaina Smith, Juan Gilbert
Open Access
Article
Conference Proceedings
Visualizing Pay-What-You-Want Donation Bursts Using Bubble Charts and LLM-based Conversation Analysis: Evidence from Twitch Subscription Gifting
Pay-What-You-Want (PWYW) donations, such as social tipping and subscription gifting on live-streaming platforms, such as Twitch, have become important monetization mechanisms for online services. Although streamers and viewers interact continuously through voice, text chats, and emojis, previous studies have not systematically examined how these interactions induce PWYW donations or characterized their temporal dynamics. This study proposes a framework for analyzing streamer–viewer interactions and PWYW donations by combining bubble chart visualization with a Large Language Model (LLM). Bubble charts plot the timing and scale of subscription gifting along the livestream timeline to identify "donation bursts," defined as periods during which multiple viewers donate consecutively. The LLM analyzes the streamer's voice during these bursts. A case study of a top Twitch streamer examined three streams from different genres: chatting, IRL', and gaming. The bubble charts reveal that gifting patterns vary by genre, with one genre exhibiting markedly distinct dynamics. The LLM analysis further suggests that donation bursts emerge from several distinct types of streamer–viewer interactions. These findings offer new insights into the relationship between live streamer–viewer interactions and viewers' donation behaviors. The framework contributes to research on live streaming, human-computer interaction, and digital monetization.
Hisayuki Kunigita
Open Access
Article
Conference Proceedings
Sustainability-Driven Assistive Technology: A LEGO Prototype for Enhancing Safe Navigation
Ensuring safe and independent mobility remains a significant challenge for people with visual impairments, particularly in contexts where specialized assistive technologies are costly or difficult to access. This paper presents the design and evaluation of an Assistive Navigation Prototype (ANP) based on LEGO Mindstorms to support obstacle detection and non-visual feedback during mobility. The prototype was developed through a participatory, user-centered process based on the iPlus methodology and integrates proximity sensors, haptic feedback, and a modular robotic structure. The system was designed to identify obstacles and provide immediate feedback to support navigation in indoor and controlled urban environments. Evaluation in simulated and semi-real conditions examined obstacle detection, system response, and perceived usability. The results showed reliable obstacle detection, consistent system response, and positive usability perceptions, supporting the feasibility of LEGO-based prototyping for exploratory assistive navigation applications. The study contributes a low-cost and adaptable approach to the development of mobility-oriented assistive technologies, particularly for contexts where access to specialized solutions is limited.
Marco Santórum, David Morales-Martínez, Mayra Carrión, Daniel Carvajal, Juan Benavides
Open Access
Article
Conference Proceedings
Effects of Virtual Agent Style and Decision Control on Sense of Agency and Trust in AI-Enabled Telecare
AI-enabled telecare agents are shifting from information providers to decision-making partners, yet how agent presentation and the allocation of decision authority jointly shape users' psychological experience remains unclear. This study examined the effects of virtual agent style and decision control mode on sense of agency and trust. In a 2 × 3 mixed design, 60 participants interacted with an expert or partner agent (between-subjects) under AI-led, user-led, and user–AI negotiation modes (within-subjects), completing five simulated telecare tasks per mode. Participants rated positive and negative sense of agency (SoPA, SoNA) and trust and distrust. Compared with the AI-led mode, the user-led and user–AI negotiation modes yielded higher trust and lower distrust, and user–AI negotiation yielded the highest SoPA and the weakest negative sense of agency. Agent style had no main effects but interacted with decision control mode for SoPA, SoNA, and trust: the expert agent supported sense of agency under the AI-led mode, whereas the partner agent elicited higher trust under user–AI negotiation. Telecare agents should align role presentation with how decision authority is shared.
Che Liang-yuan, Chien-Hsiung Chen
Open Access
Article
Conference Proceedings
Appropriate Reliance in Wearable AI for Parkinson’s Differential Assessment: Calibration, Selective Referral, and Subgroup Reliability
Smartwatch-based machine learning for Parkinson's disease (PD) is often evaluated as forced-choice classification against healthy controls, although clinically similar disorders create the harder decision boundary. This retrospective study evaluates a human-centered selective-prediction framework on the Parkinson's Disease Smartwatch (PADS) dataset. The primary task included 276 participants with PD and 114 with differential diagnoses (DD); 79 healthy controls formed a secondary benchmark. From bilateral accelerometer and gyroscope recordings across 11 motor tasks, 3,366 participant-level time-domain, spectral, magnitude, and wrist-asymmetry features were derived, with demographic variables excluded from model inputs. Regularized logistic regression, Extra Trees, and radial-basis-function support-vector models were selected by nested five-fold outer and three-fold inner cross-validation and sigmoid calibrated. We evaluated discrimination, calibration, selective referral, subgroup uncertainty burden, class-conditional split conformal prediction, sensor/task ablation, and leave-one-DD-subtype-out shift. For PD versus DD, calibrated balanced accuracy was 0.674 and AUROC 0.714; Brier score and expected calibration error improved to 0.180 and 0.047. Referring the 20% least-confident cases reduced retained-case error from 30.5% to 25.0%. At 90% nominal conformal coverage, empirical coverage was 94.6%, but 76.7% of outputs were ambiguous. Under unseen DD subtypes, confidence referral did not consistently remove false-PD predictions, whereas conformal ambiguity increased. The results support evaluating wearable AI as a referral-and-review system rather than a forced-choice classifier.
Mohammad Mahdi Nazari, Negar Heidari Matin, Yaser Michael Banad
Open Access
Article
Conference Proceedings
Visual Quality, Perceived Safety, and Spatial Vitality in Shichahai, Beijing: A Day–Night Assessment of a Cultural Waterfront
Cultural waterfronts require evaluation after dark, yet photographs, human perceptions and observed activity represent different kinds of evidence. This study develops an image-based audit of Shichahai, Beijing, as the environmental component of a framework linking visual quality, perceived safety and spatial vitality. An archive of 1,556 photographs was screened; the newest available daytime imagery at each location was compared with author-confirmed 2026 nighttime photographs. After removing a duplicated daytime source, the primary analysis contained 71 nominally corresponding sites and 568 images. A fixed ADE20K-pretrained SegFormer-B0 model estimated water-class pixel coverage, while photographic proxies described darkness and the spatial distribution of brightness. Mean site-level dark-area coverage increased from 29.34% to 45.83%, a difference of 16.49 percentage points, and the grid brightness ratio declined from 0.259 to 0.214. Predicted water coverage averaged 1.43% by day and 1.70% at night, with no stable directional difference supported by the descriptive resampling interval. Brightness perturbations exposed sensitivity in water predictions. Historical daytime dates, unequal camera views and absent segmentation ground truth limit interpretation. No perception ratings or activity observations were collected. The contribution is a reproducible environmental audit and an explicit account of the validation needed before inferring human experience or waterfront use.
Jingjing Ma, Wu Yuwei, Huang Yan
Open Access
Article
Conference Proceedings
Prompts to Impact: User Experience Design in AI-Enhanced Online Learning Environments
The growing integration of Artificial Intelligence (AI) into higher education requires a human-centered perspective that extends beyond technical functionality to examine how learners perceive, understand, and engage with AI-generated instructional content. This study evaluates the effects of prompt-engineered educational resources on user experience in a Moodle-based Business Intelligence workshop conducted in Ecuador. The intervention involved 10 advanced students from Information Technology and Software Engineering programs who independently interacted with AI-generated videos, code examples, simulations, scripts, and dashboard-related materials, with tutor support. The workshop was structured according to the Plan–Do–Check–Act framework and assessed through a pre-test/post-test design focused on learning facilitation, content comprehension, and engagement through multimedia resources. Results showed statistically significant improvements in all three dimensions. Learning facilitation increased from 3.5 to 4.3 (p = 0.0107, Cohen’s d = 1.01), content comprehension rose from 3.5 to 4.4 (p = 0.0294, d = 0.81), and multimedia engagement increased from 3.4 to 4.6 (p = 0.0002, d = 1.89). These findings indicate that prompt-engineered instructional materials can enhance motivation, conceptual understanding, and participation in online learning. The study contributes a methodological foundation for positioning prompt engineering not only as a content-generation technique, but also as a user-centered instructional design strategy for AI-enhanced education. Future research should examine affective responses, collaborative learning, scalability, and explainability in larger and more diverse populations.
Julian Galindo-losada, Vilma Rodríguez, Viviana Parraga, Mayra Carrión, Marco Santórum
Open Access
Article
Conference Proceedings
Designing Human–AI Interaction to Foster Ambidextrous Thinking and Learner Ownership in Product Ideation
Generative AI is increasingly used in design education and product ideation, yet its role tends to be reduced to producing many ideas efficiently, whereas creative product development also requires deepening questions, testing assumptions, refining concepts, and maintaining ownership of the design process. This paper presents a human–AI interaction approach for fostering ambidextrous thinking and learner ownership in engineering design education at a Japanese university. It builds on a generative AI–based method that scaffolds ideas along two axes—exploration (qualitative shifts in design direction, extended with the notion of “darkness” from dark horse prototyping) and exploitation (quantitative refinement of feasibility and implementation quality)—in which custom GPT evaluators score sketches and return feedback. We add a Japan–Germany Global PBL in which students co-designed a companion robot, with findings from a questionnaire and interviews. Ownership was uniformly high and rooted in having physically realized the prototype; AI helped students verbalize the problem and expand their options, whereas convergence and design requirements were derived from users, the prototype, and experience. On this basis, we propose and discuss the design of human–AI interaction in design education in the age of AI.
Daigo Misaki, Joohan Bae, Komei Miura
Open Access
Article
Conference Proceedings
Designing Perimenopause Care Using Disaggregated Data for Asian American Women
The "Design for All" principle often reflects the needs of a general population rather than specific needs of those not represented in data. For Asian American (AsA) women's health, biases stem from grouping diverse population groups from over 50 countries into a single category. Moreover, to our knowledge, no health technology platform provides culturally tailored perimenopause care for this diverse population. Through Jobs-To-Be-Done research with AsA women and clinicians, we identified differences in awareness of specific symptoms, access to culturally tailored information, and preferences for care plans beyond Western hormonal therapy. These insights informed a platform featuring guided self-assessment, personalized health reports, and an AI-powered health assistant. Our work demonstrates how disaggregated data can translate from principles to practice by surfacing meaningful insights on symptom profiles and care options for AsA women at the population group level while giving them transparency and control over their data.
Jennie Lai, Qin Jing, Nancy Jiang, Stella Yi
Open Access
Article
Conference Proceedings
Exploring the Factor Structure and External Correlates of Design Attitude
Design Attitude offers a way to conceptualize how individuals engage with uncertainty and generate possibilities in creative practice. However, the relationship between theoretically differentiated characteristics of Design Attitude and its empirically observed dimensional structure remains unclear. This study examined the factor structure of 36 theoretically derived Design Attitude items and their associations with creative self-efficacy (CSE), general self-efficacy (GSE), and a study-specific indicator of PERMA-related well-being. An online survey yielded 828 analyzable responses, randomly split into an exploratory sample (n = 414) and an evaluation sample (n = 414). Exploratory factor analysis yielded an interpretable five-factor solution comprising 34 items. A subsequently refined confirmatory factor analysis model comprised 33 items (CFI = .918, TLI = .910, RMSEA = .059, SRMR = .058). Scores for the five dimensions were positively correlated with CSE, GSE, and PERMA-related well-being. In hierarchical regression, adding the five Design Attitude dimensions after demographic and design-related characteristics, sleep indicators, and GSE increased R2 for the PERMA-related indicator by .170. Overall, the findings indicate that the nine theoretically differentiated characteristics were not reproduced one-to-one in questionnaire responses but were reorganized into five broader, correlated dimensions. Because the CFA model was refined within the evaluation sample and the well-being measure was study-specific, the resulting structure should be regarded as a provisional measurement model requiring further validation rather than as a finalized Design Attitude scale.
Mika Isobe, Yumin Wang, Yoshikuni Edagawa
Open Access
Article
Conference Proceedings
Parametric Design of HEV Intake Grilles Through Traditional Pattern Translation Based on FCM and XGBoost
As automotive electrification accelerates, hybrid electric vehicles (HEVs) still need to retain intake grilles, which support heat dissipation, brand recognition, and emotional expression. Based on Kansei Engineering, fuzzy c-means clustering (FCM), XGBoost regression, and semantic translation of traditional patterns, this study proposes a parametric design method for HEV intake grilles. First, 60 HEV front-face samples and 12 Kansei descriptors were collected, and user evaluation with FCM clustering was used to extract three Kansei dimensions: Intelligent Order-Transparency, Oriental Elegance, and Dynamic Strength-Stability. Second, the grille was decomposed into five types of design features, each containing six discrete categories, and three XGBoost regression models were built separately. The 12-fold cross-validation results show that the test-set R² values of the three dimensions are 0.928, 0.967, and 0.950, while the RMSE values are 0.260, 0.239, and 0.216, indicating good predictive performance. On this basis, 7,776 design combinations were predicted and ranked, and the segmentation, curves, overlap, direction, density, and boundary relationships of ice-crack, ruyi-cloud, and scale patterns were transformed into grille generation rules. Taking the scale pattern as an example, this study demonstrates the application of the workflow in the parametric translation of traditional patterns and provides a methodological reference for mapping user Kansei needs, grille morphological features, and the compositional rules of traditional patterns.
Yichen Tian, Zimo Chen, Shuài Zhāng
Open Access
Article
Conference Proceedings
Postpartum Women’s Experiences with Digital Health Resources and Support Needs: A Qualitative Study and Design Implications
Digital health resources have become an important means for postpartum women to access health information and support, yet existing resources may not fully align with their everyday needs and patterns of use. This study explored postpartum women’s experiences with digital health resources and the challenges and support needs arising from their use. Semi-structured interviews were conducted with 12 women within 12 months postpartum and analyzed using thematic analysis. Eight interrelated themes were identified. Women drew on different support sources according to their needs but still had to verify information, assess its credibility, and adapt general advice to individual circumstances. Use was constrained by limited time, energy, caregiving demands, and the effort required for continued engagement. Participants described reassurance and worry in their use of digital resources, the involvement of digital information in family caregiving discussions, and limited space for their own interests in parenting-oriented digital environments. Based on these findings, four design implications are proposed: organizing support around different needs; structuring information around its sources, applicability, and actionable guidance; enabling low-burden, interruptible interactions; and supporting women’s control over the presentation and scope of parenting-related content.
Qi Kong, Buzhou Tang
Open Access
Article
Conference Proceedings
What May and May Not Be Digitized: Two Boundaries on Digital Transformation in Traditional Crafts and Their Implications for Service Design — A Case Study of Japanese Kaga Yuzen
Digital technology has advanced remarkably in recent years, and the trend toward digital transformation (DX) in society is expected to accelerate even further in the future. Moves to incorporate various digital technologies can also be seen in Japan’s traditional craft industries. However, there is not yet a sufficient body of research documenting the difficulties and internal conflict encountered when incorporating digital technology into the production process of traditional craft products, or addressing the institutional constraints involved. Taking institutional constraints into account, this paper examines what may and may not be digitized in the production process of traditional craft products, where value is placed on handcrafted work. In doing so, we also adopt a service design perspective and use Kaga Yuzen in Japan as a case study. In this study, we conducted research mainly using anthropological ethnography. Our fieldwork revealed two forms of line-drawing, “line-drawing by the institution” and “line-drawing by practitioners,” and showed that these two forms of line-drawing lead to questions of service design.
Yuka Ishiba, Yasunobu Ito
Open Access
Article
Conference Proceedings
Flipping Through My Day: Reimagining the Digital Calendar Through Novelistic Translation and Progressive Narrative Delivery
Contemporary digital life is governed by efficiency-driven tools that flatten qualitative human experiences into standardized metrics, promoting cognitive externalization and eroding episodic memory. To address these critical human factors within time-management infrastructures, this study proposes an alternative temporal interaction paradigm. We introduce "novelistic translation" and "emotional design" as interpretive interfaces, leveraging generative AI to transform quantified schedules into continuous literary narratives that reconstruct lived experiences. To evaluate this paradigm's potential in evoking situated awareness, we conducted a two-phase, longitudinal field deployment with efficiency-reliant users. In the first phase, a 3-day field experience introduced a progressive narrative transmission mechanism via a "10-5-2-0" cadence. By delivering condensed narrative fragments prior to events, the interface intentionally injected "temporal friction" to simulate memory recall, followed by immediate semi-structured interviews. In the second phase, one month post-deployment, all fragments were compiled into a physical "novelistic diary" to examine how this temporal span reshapes autobiographical reflections through literary text. Qualitative thematic analysis of both interview phases highlights the core capacity of novelistic interfaces to disrupt the instrumental rationality of time. The findings demonstrate that by introducing poetic narrative friction, the proposed interaction paradigm successfully reclaims users' situated awareness, deepens emotional connections with their daily schedules, and counters efficiency-oriented temporal governance.
Tzu-Wen Tsai, Rung-Huei Liang
Open Access
Article
Conference Proceedings
Creating a Path to Mastery: Impacts of Machine-Learning Guidance on Novice Software Users
Novice users of complex software often face a steep learning curve, and many disengage before reaching proficiency. This study examined whether machine-learning-guided, in-application scaffolding can lower that barrier. Using cognitive apprenticeship theory (Collins, 2012) as a framework, an online study with U.S. science teachers tested a novel creative software environment, the DaVinci Science Comic Maker (DSCM). In the experimental condition, a machine-learning algorithm adjusted the complexity of available tools and supports to each user’s demonstrated skill level; the control condition offered no scaffolding and full tool complexity from the outset, mirroring typical unsupported software adoption. Participants created three comics within DSCM and completed a survey drawn from the Unified Theory of Acceptance and Use of Technology (UTAUT; Venkatesh et al., 2003). Three data sources were analyzed: artifact complexity and efficiency scores, clickstream data, and UTAUT responses. Experimental-group participants produced increasingly complex comics across trials (F(2,18) = 17.60, p < .001), while control-group complexity did not change (F(2,20) = 0.36, p = .705). The experimental group also used fewer clicks to complete its first comic. No significant between-group differences emerged on UTAUT subscales. These findings suggest that machine-learning-enabled cognitive apprenticeship scaffolding can support measurable skill growth among novice software users without external training, with implications for reducing adoption barriers in educational contexts and for the future design of complex creative software environments.
Melissa Peterson
Open Access
Article
Conference Proceedings
Effects of Viewpoint Position and Display Device on Operators’ Driving Workload in Vehicle Teleoperation Systems
In Japan, the shortage of drivers due to the declining working-age population has become a serious problem. Vehicle teleoperation systems, in which remote operators monitor autonomous vehicles and intervene in difficult situations, offer a potential solution. Teleoperation may alleviate driver shortages and expand the operable area of autonomous vehicles by allowing operators to manage multiple vehicles and complement autonomous driving in difficult sections. In such systems, viewpoints providing wider environmental information, such as third-person and top-down views, have been investigated in addition to the conventional cockpit view. Additionally, the use of head-mounted displays (HMD) that improve situational awareness has been explored. However, the effects of the viewpoint position and display device on the operator driving workload have not been sufficiently investigated. Therefore, this study investigated these effects to identify an appropriate viewpoint presentation method for vehicle teleoperation. A driving experiment using a simulator was conducted using a monitor and HMD and three viewpoint positions (cockpit, third-person, and top-down views), resulting in six conditions. Situational awareness and vehicle operation difficulty were subjectively evaluated after each driving course. The results indicated that the characteristics and effectiveness of each viewpoint differed depending on the driving situation. In addition, the stereoscopic vision provided by the HMD potentially reduced the driving workload. These findings suggest that switching or combining viewpoints according to the driving situation may enable a more appropriate visual presentation and reduce the operators’ driving workload.
Taiga Fukuzawa, Kazunori Kaede, Yusuke Osawa, Keiichi Watanuki
Open Access
Article
Conference Proceedings
Cycling by Disabled People in Japan: Current Situation and Future Possibilities
Many previous studies have viewed cycling as a dangerous activity for disabled people. However, changes in social conditions and technological advancements are now enabling disabled people to cycle. This study aims to identify the processes through which disabled people become able to cycle. The current situation in Japan was examined through a review of previous studies, a questionnaire survey, and expert interviews. Responses to the questionnaire survey were obtained from 133 disabled people and 159 non-disabled people across Japan and quantitatively analyzed to examine differences between disabled and non-disabled cyclists in terms of reasons for cycling, motivations, learning processes, and related factors. The analysis yielded three main findings. First, becoming able to cycle was found to be associated with improved well-being among both disabled and non-disabled people, although cycling may serve different functions for these groups, such as leisure or mobility. Second, awareness, participation, and experience sharing emerged as key factors in promoting cycling, as reflected in the responses from current para-cyclists. Third, facilitating this process requires the development of appropriately designed environments and opportunities for participation, as noted by current para-cyclists based on their experiences and insights. Overall, these findings offer implications for human-centered design and the promotion of inclusive mobility.
Yoshito Dobashi
Open Access
Article
Conference Proceedings
User-Centered Transmedia Storytelling Design for Urban Landmarks: A Case Study of Shenzhen’s Bay Glory Ferris Wheel
Against the backdrop of media convergence and urban digital transformation, urban landmarks are increasingly evolving from physical attractions into cultural and experiential spaces integrating narrative experiences and public participation. However, existing practices often emphasize visual symbols or isolated media content, with limited attention to narrative continuity across media touchpoints. Grounded in transmedia storytelling theory, this study examines urban landmark design through three dimensions: storyworld, media touchpoints, and audience participation. A user-centered exploratory mixed-methods approach was adopted, combining field observation, semi-structured interviews, a user survey, and a participatory co-design workshop to investigate user experiences, narrative and interaction preferences, and media touchpoints. The findings were synthesized into design requirements and strategies for subsequent design development. The results reveal key concerns related to interaction gaps, digital engagement, urban narrative identification, emotional retention, and experiential value. The co-design process further translated these findings into experience directions and media touchpoints across different stages of the visitor journey. Based on these findings, the study proposes a user-centered pathway from user exploration and participatory co-design to design translation. It argues that urban landmark transmedia storytelling should move beyond repetitive multi-platform communication and instead construct coherent experiences through a shared storyworld and complementary media touchpoints, providing practical insights for more interactive and emotionally engaging urban landmark experiences.
Xinliang Chen, Xin Xie
Open Access
Article
Conference Proceedings
From Resources to Stories: An Experiential Narrative Design for Children’s Ecological Field Studies in OCT Wetland Park
This study constructs an experiential narrative design framework for children’s ecological field studies in urban constructed wetlands. This framework was practically validated at Shenzhen Overseas Chinese Town (OCT) Wetland Park, China. The study explores the mechanisms through which the experiential narrative design fosters a connection between people and ecological systems, evokes emotional resonances, and fosters recognition of sustainable development values. For this research, a thematic field study activity – the ‘Nature Wetland Magic Academy’ – was designed and implemented. The children’s observations and feedback were recorded for qualitative analysis. The results indicate that experiential narrative designs can effectively transform ecological knowledge into emotionally engaging exploration narratives, significantly enhancing children’s emotional connection to wetland ecosystems and the complexity of their cognition of such systems. The interweaving of narrative threads and the integration of role-based tasks are key to transforming children from passive observers into active story participants. This study provides a theoretical framework and a practical paradigm for deeply integrating knowledge transmission, emotional connection, and ecological responsibility cultivation into wetland ecological field studies. Future research should further explore balancing scientifically accurate ecological knowledge with narrative appeal in field study designs and establishing evaluation mechanisms for long-term educational outcomes.
Xiaobao Yu, Miao Fu
Open Access
Article
Conference Proceedings
An Analysis of Handoff Processes in Global Teams
In the current software development sector, effective design handoff processes are essential. Teams must ensure seamless collaboration and knowledge transfer among members in diverse geographical locations. This study explores challenges and opportunities in design handoff for global teams, aiming to identify best practices and strategies for improving delivery methods. The analysis includes factors like cross-cultural communication, time zone differences, communication technologies, and semiotics. We administered a questionnaire to identify gaps in design handoff processes between multicultural teams. Based on the results, we discussed the findings to further explore issues surrounding these workflows. By examining this critical aspect of global team dynamics, the research offers significant insights for businesses and organizations seeking to enhance design delivery processes in international collaboration environments.
Daniel Cunha, Katleen Picanço, Barbara Silva, Lanier Santos, Erick Bezerra
Open Access
Article
Conference Proceedings
Towards Participatory Requirements Gathering of Constantly Evolving User Needs for Smart Cane+
“Smart Cane+” systems combine smart canes, which already expand conventional white canes with technical features, with connected technologies such as smartphones or other wearable devices. However, the adoption rate among blind and visually impaired people (BVIP) is very low, primarily because these assistive technologies do not meet their needs. This paper compares requirements identified through an exploratory literature review with those obtained from semi-structured interviews involving BVIP. Literature-derived requirements were synthesized into six categories and systematically compared with interview findings to identify confirmed, refined, and newly emerging requirements. The results show that many established requirements remain relevant. At the same time, participants refined existing requirements and introduced additional ones. Conflicting participants’ preferences underline the importance of significantly more adaptable and modular concepts. Overall, the findings indicate that requirements for Smart Cane+ systems are constantly evolving rather than relatively stable. We therefore argue that requirements gathering for assistive technologies should be understood as continuous, participatory process that combines existing knowledge with ongoing user involvement to address changing technologies, user expectations, and mobility needs.
Monika Wiechers, Celina Mae Kong, Birgit Bomsdorf
Open Access
Article
Conference Proceedings
Rethinking Inclusivity in Dissonant Heritage
Nowadays, approaches to cultural heritage increasingly promote objectives of inclusivity, participation and accessibility. Yet can heritage interpretation and enhancement processes truly be considered inclusive if certain memories and narratives remain marginalised or excluded?The paper follows a progressive line of inquiry, moving from questioning inclusivity as applied to heritage, to analysing how this concept may be redefined when dealing with dissonant heritage and finally examining selected case studies illustrating how such a reframed understanding may become operational.Cultural dissonance originates from the selective nature of heritagisation, through which certain cultural expressions are assigned the status of heritage. It arises when heritagisation encounters dramatic memories, contested values and conflicting narratives, which challenge traditional practice with heritage.Thus a paradox emerges: while the “Design For All” is generally geared towards conflict resolution, in dissonant heritage contexts inclusivity may instead be achieved by making conflict visible. Interpretative tensions should therefore not be removed but revealed to foster constructive dialogue.Yet making conflict visible also means acknowledging the existence of silenced memories and marginalised narratives, thus challenging the assumption that heritage interpretation and enhancement are inherently inclusive processes. On the contrary, they reflect social and political power structures, therefore making inclusivity both a technical and political issue.In conclusion, the paper argues that inclusivity in dissonant heritage should not be understood as the resolution of interpretative conflict, but rather as its recognition. It should depend less on achieving consensus than on critically engaging with disagreement: revealing rather than hiding, provoking rather than appeasing.
Valeria Minucciani, Gabriella Dalla Serra
Open Access
Article
Conference Proceedings
Trend Analysis of Cognitive Ability in Three Long‑Duration Working Groups Based on Piecewise Linear Regression
Long-duration operations in confined and shift-based environments can significantly affect operator cognitive ability. This study investigates the trends of cognitive ability changes in three long-duration working groups using piecewise linear regression. Fifteen participants were divided into three shift groups (A1, A2, and A3) and completed a 288-hour simulated monitoring task. Cognitive ability was assessed before and after each duty period through five indicators: perceptual stability, attention-and-vigilance stability, working-memory stability, reasoning-and-decision stability, and executive-function stability. Group-level cognitive ability time series were constructed by averaging the five normalized indicators, and two-segment linear models were fitted to capture the overall trend of cognitive changes over time. The results show that all three groups exhibited an overall declining trend with local fluctuations and partial recovery, while the magnitude and pattern of changes varied according to shift timing. Group A2 maintained the highest overall cognitive ability level, whereas Group A3 showed the lowest among the three groups. The piecewise linear regression models achieved low root-mean-square errors for all groups, indicating good fitting accuracy and stability. These findings suggest that shift timing influences the trend of cognitive ability during long-duration operations, and the proposed piecewise linear regression approach can effectively describe the temporal dynamics of cognitive performance.
Si Li, Xiaoyi Zhou, Yang Yu, Liping Pang, Liang Zhang, Ziang Chen
Open Access
Article
Conference Proceedings
Crew Cognition Management (CCM): A Comprehensive Design Methodology for Civil Aircraft
The increasing level of automation in civil aircraft has profoundly altered the nature of flight crew duties. However, higher automation has not diminished the requirements for human-machine coordination; on the contrary, it imposes even greater demands on the flight crew's cognitive monitoring and takeover capabilities in complex failure scenarios. Analysis indicates that the majority of human-factors-related operational errors in the flight deck stem from the flight crew's insufficient understanding and lack of predictive cognition regarding the behavior of automated systems, which in turn gives rise to cognitive biases upon sudden state changes or faults, ultimately resulting in decision-making or operational errors. To address this issue, this paper proposes the theoretical framework of Crew Cognition Management (CCM), which takes crew cognitive resources as its core and focuses on the flight crew's comprehension, prediction, and co-construction of shared cognition regarding aircraft state. Drawing on representative aircraft design cases, the paper analyses the differences in crew cognition under distinct design philosophies, and elaborates on how, with crew cognitive resources as the driving factor, an integrated four-dimensional design—spanning aircraft function design, flight manual compilation, training system development, and operational safety management—can be realized. This theoretical framework is intended to steer highly automated civil aircraft away from isolated "automation function design" toward a comprehensive "human-machine cognitive co-design." It is proposed as a novel theoretical underpinning for human-factors-oriented forward design in the civil aircraft domain, with the design objective of mitigating the categories of human error potentially induced by automation and, ultimately, supporting operational safety. The framework is primarily theoretical at this stage; its practical effectiveness still needs to be evaluated with flight crews or simulator scenarios.
Lei Wang, Jiangping Chu, Lei WU
Open Access
Article
Conference Proceedings
From Awareness to Action: Designing a Progressive Cross-Device Intervention Ecosystem for Sedentary Behavior Among Young Adults
Prolonged sedentary behavior is increasingly embedded in the daily routines of young adults. Although many young adults recognize its potential health consequences, awareness does not always translate into timely physical activity, particularly during cognitively demanding tasks. Existing reminders are often either too subtle to attract attention or too intrusive to preserve task continuity and perceived autonomy. This study investigates how a progressive cross-device intervention can balance intervention salience, task continuity, and user control. A formative mixed-methods study was conducted with young adults participants. An online questionnaire collected 208 valid responses on sedentary habits, behavioral barriers, and preferences for intervention timing, modality, and intensity. Semi-structured interviews with five participants further explored the awareness–action gap and acceptable forms of sedentary behavior intervention. Findings were interpreted through the Health Belief Model. Results showed that over 70% of respondents reported sedentary bouts exceeding one hour, while only 12.23% regularly interrupted sitting every 30 minutes. More than 90% considered external reminders potentially useful, yet preferences varied substantially across contexts and intervention intensities. Based on these findings, we designed a progressive cross-device intervention system integrating a modular wearable, ambient desktop cues, and mobile and desktop interfaces. The system visualizes accumulated sitting as metaphorical “health debt” and progressively escalates from peripheral ambient cues to explicit movement guidance, while retaining postponement and override controls. Guided micro-breaks provide restorative visual feedback. This study contributes empirical evidence of the sedentary awareness–action gap, a progressive intervention framework balancing salience, continuity, and autonomy, and a cross-device design concept combining quantified self-monitoring with affective feedback. Longitudinal evaluation is needed to assess sustained acceptance and behavioral effectiveness.
Tianmei Zhang, Jie Tan, Luyu Song, Jianbin Wu
Open Access
Article
Conference Proceedings
Governing Agentic AI Workflows: Balancing Automation and Human Oversight in Enterprise Practice
Agentic AI is extending organizational automation beyond predefined tasks, requiring organizations to reconsider what should be delegated to AI and what forms of human involvement should remain. This study compares three enterprise AI development cases. One sought to reduce cost by substituting human work. Another made AI easier to use by letting practitioners assemble assistants through natural-language instructions. A third treated continued interaction as a way for an AI agent to become aligned with actual work. Greater AI capability did not lead to a stronger expectation of removing people from the workflow. In the learning-oriented case, evaluation, correction, and contextual input were not merely safeguards against AI failure. They became part of the mechanism through which the agent and the workflow were adapted. Human-centered design for agentic AI should therefore keep delegation adjustable and treat human correction as a productive component of AI-enabled work.
Natsumi Takahashi, Indranil Bhattacharjee
Open Access
Article
Conference Proceedings
Human-Centred Adaptive Reuse of Building Heritage: A Comparative Ex-Post and Ex-Ante Evaluation of abandoned areas in Maribor, Slovenia
Adaptive reuse has become an important strategy for preserving building heritage while addressing contemporary environmental, social, and cultural challenges. Yet the social dimension of adaptive reuse and users' experiences remain insufficiently explored in evaluation practice. Furthermore, existing evaluation methods are primarily applied to completed projects and rarely support the assessment of projects still under development.This paper develops a Human-Centred Adaptive Reuse Evaluation Framework (HCAREF) for building heritage, investigating whether human-centred design principles can be applied to both retrospective (ex-post) and prospective (ex-ante) evaluations. A literature review on adaptive reuse and human-centred design was conducted to identify the key evaluation dimensions, forming an integrated framework that was then applied to three case studies from Maribor, Slovenia. Two completed projects, the Minorite Monastery Complex and the Judgement Tower, were evaluated ex post through field observations and document analysis, while the Centre Rotovž project was assessed ex ante using project documentation, architectural drawings, and visualisations.The findings indicate that the proposed framework can support the evaluation of adaptive reuse projects at different stages of development, providing a practical decision-support tool for architects, planners, and heritage professionals while advancing the integration of human-centred principles into adaptive reuse practice. The framework provides a structured basis for comparing realised qualities in completed projects with intended qualities in an ongoing project.
Vesna Zegarac Leskovar, Vanja Skalicky Klemenčič
Open Access
Article
Conference Proceedings
DEM Cropper: A Web-Based Geospatial Processing System for Physical Terrain Representation
Digital Elevation Models (DEMs) are foundational geospatial datasets used in Geographic Information Systems (GIS) for terrain analysis, environmental assessment, and spatial decision-making. However, transforming elevation data into physical terrain representations suitable for collaborative visualization and tangible geospatial interfaces often requires multiple specialized tools and significant technical expertise. This paper presents DEM Cropper, a browser-based geospatial processing system that integrates DEM ingestion, interactive spatial selection, terrain mesh generation, and fabrication preparation within a unified workflow. Users can upload GeoTIFF datasets, define arbitrary geographic regions through map-based interactions, and generate fabrication-ready terrain models directly in a web environment. The system also incorporates a terrain-aware partitioning method that subdivides large models into contiguous printable components while satisfying fabrication constraints. By lowering barriers to terrain model generation, DEM Cropper enables physical-digital geospatial visualization workflows in which terrain serves as a tangible interface for communicating and exploring spatial information.
Loelle Lam, Nurit Kirshenbaum
Open Access
Article
Conference Proceedings
Selection of Underwear Materials for Comfortable Sleeping in Summer Evacuation Shelters
Global warming has increased the frequency of disasters such as earthquakes, typhoons, and floods in Japan, forcing many people—particularly vulnerable groups such as the elderly—into evacuation shelters where air conditioning is often unavailable because of power outages. This study investigated how the choice of underwear material affects sweat evaporation and thermoregulation during sleep under such hot and humid conditions. A single 68-year-old woman wore five underwear of different material compositions (A: Cotton; B: Cotton/Polyester; C: Cotton/Nylon; D: Polyester; E: Nylon) for 6.5 h/night on from July to September, in a bedroom at approximately 29 °C and 53%RH. Body weight, oral/skin temperature, the sweat weight of clothing/towels/bedding, and urine were measured, and evaporation efficiency was calculated using a heat-/mass-balance model. Fabric thermal and water-vapor resistance were also measured with a guarded hotplate. Evaporation efficiency was highest for A and lowest for B, which retained sweat within the fabric and produced a sticky sensation on the skin. The D retained little sweat but left moisture on the skin because of its low water-holding capacity. These results suggest that selecting cotton underwear may help elderly evacuees sustain effective thermoregulation through sweating when air conditioning cannot be used. Further work with larger and more diverse samples is needed to establish practical guidelines for underwear selection in evacuation shelters.
Tamaki Mitsuno, Kaoru Wakatsuki
Open Access
Article
Conference Proceedings
Human-AI Teaming in Life Coaching: Evaluating LLM Role Fit Through Coaching Competency and Mentor Coach Perspective
Large Language Models (LLMs) are increasingly being used for personal reflection, decision support and coaching conversations. However, their role in life coaching remains unclear, especially when conversations involve sensitive topics such as values, identity, relationships, family expectations, confidence, and major life transitions. This exploratory study examined Human-AI Teaming in life coaching by comparing three LLMs – Claude, Gemini and ChatGPT, to identify their potential Human-AI Teaming roles. Rather than exploring if AI replaces human coaches, the study evaluated if LLMs can support roles like reflection partner, goal-setting assistant, coach-training simulator or mentor coach support aid. The study used six standardized life coaching scenarios and 18 simulated transcripts, evaluated through two lenses – International Coaching Federation (ICF) Core Competencies and a Mentor Coach-informed review. ICF competencies assessed the models’ coaching behaviour across 8 competency areas while the Mentor Coach perspective identified their strengths, growth edges, error patterns and risks like giving advice, asking leading questions, over-validation, premature action planning or therapy drift while coaching. Claude achieved the highest observed mean (4.56), followed by Gemini (4.31) and ChatGPT (3.98). Claude demonstrated concise reflection, Gemini structured goal support, and ChatGPT detailed reflection but frequent over-talking. All three were classified as unsuitable for the therapy-adjacent scenario. The analysis identified preliminary role-fit patterns and reinforced the need for human oversight in boundary-sensitive contexts. The study proposes a practical evaluation framework for identifying appropriate LLM role-fit in coaching related Human-AI Teaming. The study follows a human-centered approach where AI is not considered an autonomous life coach but a supportive tool, whose safe and effective use depends on role clarity, professional standards and human oversight.
Ravneet Kaur, Satarupa Mukherjee
Open Access
Article
Conference Proceedings
Modeling Individual Differences in Motivational Responses for Personalized Support of Sustained Behavior Change toward Well-Being
Sustained behavior change requires more than matching interventions to average group characteristics: it requires understanding personal characteristics, attitudinal responses, and actual behavior over time. Building on our earlier category-based interventions for exercise and sleep, this study examined the information needed to design personalized support. First-year students in the Faculty of Science, Engineering and Society at Chuo University participated in a common one-day English communication experience at Tokyo Global Gateway (TGG) on April 19, 2026. General characteristics and English-related attitudes were explored in 278 baseline respondents; longitudinal attitudes were analyzed in 137 respondents matched across three surveys, and changes in English contact outside university classes were assessed in 164 respondents matched between the first and third surveys. Baseline recognition of the importance of English was high, whereas enjoyment/interest and evaluation concern followed different temporal trajectories. Of 38 participants whose enjoyment/interest moved from below the neutral threshold to at least that threshold shortly after the experience, only 12 reported a higher category of out-of-class English contact at follow-up. These findings illustrate that favorable attitudinal responses do not correspond one-to-one with later behavioral increases. We propose a human-centered design perspective that distinguishes relatively stable personal characteristics, changing activity-specific attitudes, and current behavior, and that adaptively selects value-oriented messages, concrete action opportunities, and reflective feedback. This is an exploratory basis for future personalized intervention studies, not a causal evaluation of TGG or of the proposed support cycle.
Toshikazu Kato, Rikuto Fukada, Takashi Sakamoto, Toru Nakata
Open Access
Article
Conference Proceedings
Human-Centered Design of AI Decision Support: Algorithmic Vigilance and the Role of AI Literacy
Artificial intelligence is increasingly embedded in workplace decisions, yet practitioners oscillate between algorithm aversion and automation bias. Algorithmic vigilance, i.e. calibrated, critically reflective reliance on AI, is the normative design target, but it remains unclear what it looks like in everyday professional work and which conditions support it. Existing design principles largely stem from laboratory studies and treat AI literacy as the main individual moderator.This exploratory study combines interviews with 13 practitioners from construction, production planning and school administration with the Meta AI Literacy Scale (MAILS). Interview statements were quantitized into ordinal indicators and related to MAILS scores (Spearman's ρ). Practitioners unanimously retained decision authority and verified AI outputs. AI literacy tended to go with more specific transparency requirements (ρ = .44), whereas task complexity and decision criticality shaped explanation demand (ρ = .76 and .81) and automation affinity (ρ = −.55). AI literacy was unrelated to emotional security (ρ = .19) and emotion regulation (ρ = −.03); self-efficacy was associated with more systematic verification (ρ = .59).We derive five hypotheses and an extended model of algorithmic vigilance that adds task complexity as a second dimension next to AI literacy and locates emotional security in system design. The design implications are role- and task-sensitive transparency, active competence support and error-type-differentiated confidence communication, implemented through confirmation gates, calibrated confidence signals and configurable autonomy. A confirmatory study will test the hypotheses.
Julia Czerniak-Wilmes, Lukas Messmer, Leonie Däullary, Teuta Salii, Philipp Rüegg, Frieder Loch
Open Access
Article
Conference Proceedings
When Helpful AI Feels Too Trustworthy: Engagement and Reliance in Design Ideation
Artificial intelligence (AI) assistants are increasingly used in engineering ideation, but their effectiveness depends on both the AI model’s capabilities and how users interpret and act on AI feedback. This paper synthesizes human-factors findings from the development and evaluation of the AI Design Assistant (AIDA), a modular testbed for iterative engineering ideation, focusing on trust calibration, feedback presentation, user agency, and design exploration. Across studies, adaptive and socially responsive feedback generally increased trust and perceived usefulness, yet trust did not reliably correspond to measured design improvement. More feedback yielded mixed results: longer or highly structured responses were not consistently associated with stronger outcomes, while concise feedback and longer user revisions were associated with larger score gains in turn-level analyses. Users also applied AI guidance selectively; only 38.6% of actionable suggestions were implemented, although most revisions used at least one AI-generated suggestion. These findings suggest that the next generation of AI design assistants needs to support calibrated trust, reduce unnecessary information-processing demands, preserve explicit user control over suggestion use, and distinguish user perceptions from measured design outcomes. The results provide practical design guidance for human-centered AI systems that augment, rather than replace, human judgment during creative engineering work.
Ting Liao, Jiacheng Sun
Open Access
Article
Conference Proceedings
Understanding the needs and constraints of speech-language pathologists when using technological devices in rehabilitating people with aphasia.
Background: Aphasia is a language disorder caused by brain injuries or neurodegeneration. It has been recognized as having one of the most substantial negative impacts on patients’ quality of life, with its impact exceeding that associated with several major health conditions, including cancer and Alzheimer’s disease (Lam & Wodchis, 2010). Given its significant impact on communication and daily life, aphasia rehabilitation requires intensive interventions aimed at improving functional communication. This care is promoted by Speech Language Pathologists (SLP). They use multiple techniques and approaches, including structured exercises, ecological simulations or devices (Oussama et al., 2021). Numerous applications are available to support SLP in their evaluation, in the functional treatment and in augmentative and alternative communication (AAC) to encourage patients to use any type of communication (van de Sandt-Koenderman, 2011). Despite the growing availability of these technological solutions, they do not always fully meet the needs of SLPs or take into account the specific constraints experienced by People with Aphasia (PWA). To get a better understanding of SLP’s needs and constraints regarding these technologies, a user-centered approach has been put in place. This approach will provide a set of requirements that should allow the development of more usable devices (Yardley et al., 2015). Method: Eight French-speaking SLP women volunteers participated in the study. They provide care in public and private hospitals and as independent. Each participant took part in a one-hour semi-structured remote interview. The questions focus on their clinical management and the personalization of the rehabilitation according to PWA’s individual needs. Results: SLPs highlighted the necessity to adapt both rehabilitation and technological tools to each PWA’ individual characteristics. They also reported that some devices were poorly aligned with clinical practice, with limited SLP involvement in their development. Finally, varying exercises and aligning goals with PWA’s functional communication abilities were seen as important to maintaining motivation. Discussion: The interviews provided insights into the determinants of the personalization of rehabilitation provided by SLP. Technological devices were mainly used to monitor the progress of PWA, providing varied therapeutic support and promoting rehabilitation at home, potentially contributing to improving the quality of life (Güngör Ketenci et al., 2026).
Emilie Grandgirard, Laura Wallard, Christian BASTIEN, Christophe Kolski, Eugénie Avril
Open Access
Article
Conference Proceedings
Constrained but Capable: Motivation, Persistence, and Early-Stage Attrition Among Nascent Women Entrepreneurs
Women nascent entrepreneurs experience higher early-stage quit rates than men, yet longitudinal analysis suggests motivation does not explain this pattern. Using data from the Panel Study of Entrepreneurial Dynamics II (PSED II, N=1,214), this study compares motivational orientation across gender and examines whether motivational variables predict early-stage quit behavior among women. Results confirm statistically significant gender differences in motivational orientation: women rate flexibility as a more important entrepreneurial driver than men and rate wealth-building as less important. However, motivational variables do not significantly predict whether women quit during the earliest startup wave. Women who remain in the startup process ultimately achieve business launch rates exceeding men by the final follow-up wave. These findings suggest that early-stage attrition among women reflects the constraint environment of the startup phase rather than motivational differences between genders, with implications for the design of human-centered entrepreneurial support resources.
Lake Crowell-Mitchell, Quintin Williams
Open Access
Article
Conference Proceedings
SENSEHOME: Supporting Emotion-Driven Interior Design Decisions through Color Psychology and Augmented Reality
Residential interior design is shaped not only by functional and visual considerations, but also by the emotional atmosphere users seek to create in their living spaces. However, many digital interior design tools remain primarily product- and visualization-oriented, offering limited support for translating desired feelings into concrete choices of color, material, furniture, and spatial arrangement. Drawing on research in color psychology, this study explores how color–emotion associations can be integrated with augmented reality (AR) to support emotion-driven decision-making in residential interior design. A literature-informed design approach was adopted to translate color–affect research into design cues for organizing alternative color directions within the prototype. Based on this foundation, SENSEHOME was developed as an AR-supported interior design prototype that connects users’ affective intentions with color palettes, material choices, furnishing alternatives, and situated visualization. The workflow connects open-ended affective input with multiple color-informed design directions, iterative refinement, and situated AR comparison. Users can begin by expressing a desired spatial feeling in their own words or by exploring visual starting points, then develop and compare alternative design interpretations through the workflow. The resulting prototype is presented as a human-centered design proposition for affective decision support, with affective intention positioned as the starting point for design exploration rather than as a consideration added after product selection. This work reframes AR-assisted interior design from product visualization toward affective design support and provides a basis for future user studies on preference articulation, atmosphere–design alignment, and decision confidence.
Wang Zhou, Meiyu Zhou
Open Access
Article
Conference Proceedings
Stakeholder requirements for quality-in-use
When considering the measurement of quality in use, it is necessary to confirm the meaning of the sub-sub-characteristics "impact of use by stakeholder'' (ISO/IEC25019) defined for each quality characteristics/ sub-characteristic, and to clarify for each quality characteristic what each stakeholder expects from the target when the operator uses the target. In this paper, we proposed that it is appropriate to classify quality sub-sub characteristics and to define measurement quantities for what the target, result, organization, and society require. As a result, it was extracted that "what each stakeholder expects from the target product" are “What is requirements for the object to be used”, “What is requirements for the results used”, “What is requirements by organizations”, “Things to appeal to”, “Impact on public/society” and “Social acceptance”
Shin Fukuzumi
Open Access
Article
Conference Proceedings
EdgeConv-Based Spatial Difference Extraction of Plantar Pressure Improves Mediolateral Ground Reaction Force Estimation During Walking
Falls and fractures are major concerns in super-aged societies, highlighting the need for gait assessment in daily living environments. Wearable footwear devices equipped with plantar pressure sensors and inertial measurement units (IMUs) enable remote gait monitoring without laboratory-based equipment; however, they cannot directly measure three-dimensional ground reaction forces (3D GRFs), and estimating the mediolateral component (Fx), which is related to balance control, remains challenging. This paper proposes a deep learning model that integrates Edge Convolution (EdgeConv), a graph neural network layer, to extract spatial features from plantar pressure data and improve Fx estimation. Gait experiments were conducted with 12 healthy participants walking on an instrumented treadmill, during which 3D GRFs and wearable sensor data from eight plantar pressure sensors and a six-axis IMU were synchronously recorded. The proposed model combines EdgeConv for spatial feature extraction with a Transformer for temporal feature learning and was evaluated using subject-independent six-fold group cross-validation. For Fx estimation, the proposed model achieved an NRMSE of 8.79 ± 1.26% and an R² of 0.76 ± 0.09, compared with 9.30 ± 1.52% and 0.73 ± 0.09 for the Transformer and 9.47 ± 1.77% and 0.71 ± 0.13 for the GCN. These results indicate that learning local spatial features with EdgeConv enhances mediolateral GRF estimation from wearable footwear sensor data.
Hideyuki Nagashio, Yusuke Osawa, Kazunori Kaede, Keiichi Watanuki
Open Access
Article
Conference Proceedings
Incorporating User Research Findings into a Model-Based Enterprise Architecture to Protect the Ocean
Human-centered design (HCD) practices provide many benefits to the architectural design process, however, there is a discontinuity when attempting to use HCD artifacts as input for model-based techniques such as model-based systems engineering (MBSE). This results in a loss fidelity in the transition between HCD and MBSE, where valuable insights from human stakeholders have no corresponding representation in the architectural model. To ad-dress this shortcoming, pathways need to be established between the disciplines of HCD and MBSE. In this paper, we use the U.S. Coast Guard's role as protector of the marine environment as a case study to explore how these two methodologies can be synchronized. First, we look at how HCD can be per-formed, and the process of collecting, analyzing, and sorting user research findings. The resulting data is then mapped into the Unified Architecture Frame-work (UAF), which establishes traceability between HCD and MBSE. We then discuss how this mapping provides a clear and manageable link between the disciplines, amplifying the impact of both.
Sarah Rudder, Audrey Haque, Amna Greaves, Todd Jackson, Will Harper, Amy Florentino
Open Access
Article
Conference Proceedings
A Human-Centered, Theory-Based Approach to Designing Conversational AI for Food Insecurity Support at HBCUs
Students at Historically Black Colleges and Universities (HBCUs) experience disproportionately high rates of food insecurity, yet despite available resources, persistent barriers prevent help-seeking. Conversational AI offers potential to address these barriers and shape help-seeking behaviors. This work examines the current landscape of food insecurity support to inform the design of a conversational AI intervention. To inform intervention design, a needs assessment is conducted through semi-structured interviews with four campus administrators and nonprofit leaders involved in food insecurity programming, representing six private HBCUs in the Southeast, examining resources, communication strategies, and access barriers. Analysis revealed that underutilization stems from psychological barriers: stigma prevents help-seeking even when students know resources exist; traditional communication channels fail to reach students; and fear of being seen deters usage regardless of need. Stakeholders identified technology as uniquely positioned to address barriers, offering discreet access that bypasses stigmatizing face-to-face interactions while providing 24/7 information through channels students use. Recognizing the barriers as constructs of the Theory of Planned Behavior (TPB), attitudes, social norms, and perceived control, findings informed a conversational AI prototype grounded in the TPB to shape help-seeking behaviors. The system addresses TPB constructs by normalizing language and social-proof messaging to improve attitudes and social norms, providing step-by-step guidance to enhance perceived control, and integrating diverse resources that reflect interconnected needs. This work demonstrates a stakeholder-informed approach to designing AI for sensitive contexts where psychological barriers are as significant as practical ones. Future work will evaluate system usability and its impact on help-seeking behavior.
Alaina Smith, Jasmine Mckenzie, Mark Holden, Wesley Junkins, Juan Gilbert
Open Access
Article
Conference Proceedings
A User-Centred Design Approach for Inclusive and Trustworthy Human-Machine Interfaces in AI-based CCAM Systems
Human-Machine Interfaces (HMIs) for AI-based Connected, Cooperative and Automated Mobility (CCAM) systems are too often designed around an implicit “average user”, typically a young, experienced, able-bodied driver, despite growing evidence that user needs vary sharply across age, driving experience, cognitive profile and sensory ability. This paper presents the User-Centred Design (UCD) methodology developed in the AIGGREGATE project to derive inclusive HMI requirements for three automated-driving use cases: collective perception in shared urban spaces (UC-Aa), collective decision-making at SAE Level 4 (UC-Ab), and cooperative highway Take-Over Request (TOR) management (UC-B). Grounded in ISO 9241-210:2019 and the Task–Tool–User triangle , the methodology combined a state-of-the-art review with twenty semi-structured interviews following a mock-up exposure session, involving participants with Mild Cognitive Impairment, ADHD, visual and hearing impairments, older adults, and novice and experienced drivers across six countries. Thematic analysis of the interviews, cross-referenced with pre- and post-exposure questionnaire data, produced 43 interview-derived HMI requirements distributed across user profiles and use cases. Results show that acceptance was conditional rather than binary: participants did not reject the automated-driving concepts, but identified specific, implementable design deficiencies tied to their individual profiles. The paper discusses these profile-specific findings, the resulting interview-derived recommendations, and the sample-size limitations that will guide the next, implementation-focused phase of the project.
Turkan Hentati, Katerina Touilou
Open Access
Article
Conference Proceedings
From Generic HMI to Driver-Adaptive Automation: Multimodal Feedback, Mode Awareness and Customized Takeover Timing
Current European regulations and guidelines for Human-Machine Interfaces (HMIs) in automated vehicles provide an important framework for safety evaluation, although some aspects can remain open by stakeholders from the automotive industry when translated into practical evaluation criteria. For instance, concepts such as “timely warning” do not always specify precise temporal thresholds, while assumptions based on an “average driver” may overlook substantial differences in cognitive processing speed, attentional capacity, expertise, and mode awareness. These considerations are particularly critical during transitions between manual and automated driving, where proportionate awareness of the current automation mode can contribute to inappropriate driver responses and delayed takeovers. This paper presents the experimental protocol of Use Case 1 (UC1-HMI) of the CERTAIN project, which applies an inclusive, driver-centred approach to investigate how adding haptic feedback to conventional audio-visual warnings influences mode awareness and Take-Over Request (TOR) performance, and how warning techniques should be quantitatively adapted to different driver profiles. In this experiment, fifty participants stratified by driving expertise (novice/experienced), age (including adults 60+) and neurocognitive diversity (including ADHD), with gender balance, complete a 2-hour simulator session covering manual driving, SAE Level 2 hands-on and Level 3 hands-off automation across a seven-step highway-to-urban transition scenario. Two warning conditions are compared between groups: audio-visual feedback alone versus audio-visual feedback combined with haptic stimulation. Data triangulate subjective measures (NASA-TLX, Technology Acceptance Model, comfort and trust ratings, mode-confusion survey), the E3D Incabin physiological monitoring system, and a video-cued walkthrough interview. The paper presents the protocol, the research questions, the expected outcomes in light of prior literature, and the next steps, a second design iteration informed by these findings, toward adaptive, evidence-based warning strategies that replace universal HMI thresholds with driver-adaptive ones.
Turkan Hentati, Rosen Dimov, Oana Moldovan
Open Access
Article
Conference Proceedings
A Human-Centered Framework for AI-Wearable Integration in Automotive Assembly Lines
Industry 5.0 shifts manufacturing from automation-centred production toward human-centred collaboration between operators and intelligent systems. Although artificial intelligence (AI), augmented-reality (AR) head-mounted displays, sensor-equipped wearables, and digital twins can support assembly quality, safety, training, and process optimisation, these technologies are often developed as isolated applications. This study proposes a process-oriented framework for integrating AI-assisted wearable systems into automotive assembly operations. The framework is developed through a structured synthesis of Industry 5.0, human–AI complementarity, wearable technologies, digital twins, Value Stream Mapping, Design for Assembly, concurrent engineering, and ergonomic system design. Process mapping and task decomposition are used to identify candidate operations characterised by bottlenecks, quality risks, repetitive activity, cognitive complexity, or ergonomic demands, while stakeholder analysis captures requirements from operators, quality, safety, maintenance, supervision, and management. The resulting framework combines a three-layer system architecture, a human–AI service blueprint, and a four-phase implementation roadmap that links wearable sensing and guidance with AI processing and backend systems such as digital twins, Manufacturing Execution Systems, and Quality Management Systems while retaining human decision authority. A structured evaluation plan is further proposed to assess operational performance, AI and sensing accuracy, system responsiveness, human-factors outcomes, and technical feasibility through controlled pilots, comparative testing, robustness analysis, and expert review. The framework provides a practical pathway for future evidence-based deployment of AI-assisted wearables in automotive and other complex manual-assembly environments.
Xing Yao, Yehong Zhou, Qiheng Peh, Yanhao Tan, Xiaoqing Yu, Seung Ki Moon, Chun-Hsien Chen, Lam Haw Woei Sean, James Song
Open Access
Article
Conference Proceedings
Evaluating the Usability and Decision Support Effectiveness of Public Health Dashboards Using Eye Tracking
Public health professionals increasingly rely on complex data related to social determinants of health (SDOH), food insecurity, and healthcare access to support evidence-based decision-making. Interactive dashboards can facilitate the interpretation of these data. However, their effectiveness depends on how clearly and efficiently information is visually communicated to users. This study evaluates the usability and decision-support effectiveness of three interactive public health dashboards using task-based performance, user feedback, and eye tracking analysis. Six graduate-level participants completed a within-subjects experiment involving three randomized dashboard conditions, with two decision-making tasks performed for each dashboard. Eye-tracking data was collected using a Tobii Pro Spectrum and Tobii Pro Lab to examine visual attention and interaction patterns across dashboard elements. Areas of Interest (AOIs) were used to capture measures including fixation behavior, gaze patterns, and mouse interactions. Participants achieved an overall task accuracy of 88.89%, although performance varied across dashboard tasks, suggesting potential differences in the clarity and usability of specific visualizations. Findings from this study can inform the development of more intuitive public health dashboards by identifying visual design features that effectively attract attention, communicate information, and support accurate decision making.
Michael Stubblefield, Mikaya Hamilton, Steven Jiang
Open Access
Article
Conference Proceedings
AI-Assisted Head-Shape Design for High-Speed Trains: Operationalizing Emotional Intent in a Designer-in-the-Loop Workflow
We present an AI-assisted, designer-in-the-loop workflow for high-speed train head-shape design that makes emotional intent actionable. Prior work optimizes aerodynamics or judges aesthetics post-hoc, offering little guidance while forms are being created. We elicit perception dimensions via interviews with 10 train enthusiasts and LDA, generate 50 parameterized designs with an L50 orthogonal array, and collect ratings from 27 participants. Random-forest regressors map 12 form and colour parameters to perceived responses; the resulting models are refined into an interpretable evaluation tool. Two professional designers then use the tool’s scores and cues to steer divergent concepting. Our contributions are: (1) a replicable, model-informed pipeline that links user language to editable parameters; (2) an interpretable evaluation model that supports early-stage exploration and trade-off reasoning; and (3) a practical, AI-assisted workflow prototype for complex engineered forms. This advances emotional intent from rhetoric to controllable design signals and informs decision-making during early design.
Zhengyang Wang, Ruochen Hu
Open Access
Article
Conference Proceedings
Exploring Usability and User Perception in an Immersive Robotics Learning Environment
This paper presents an early-stage formative evaluation of the usability and user experience of an immersive Virtual Reality (VR) robotics learning environment developed for the Meta Quest 3 headset. A convergent mixed-methods approach combined concurrent think-aloud, task-based observation, semi-structured post-task interviews, and four standardized questionnaires: the System Usability Scale (SUS), the Virtual Reality System Usability Questionnaire (VRSUQ), the Virtual Embodiment Questionnaire (VEQ), and the Simulator Sickness Questionnaire (SSQ). Fourteen participants with expertise in robotics and/or XR completed individual sessions in a controlled environment. SUS (M = 79.46, SD = 11.69) and VRSUQ (M = 79.17, SD = 11.20) scores indicated good perceived usability. However, VRSUQ Effectiveness scores were significantly lower than Efficiency and Satisfaction, revealing weaknesses in feedback clarity and error prevention. The qualitative findings identified recurring problems involving interaction feedback, language and speech consistency, spatial orientation, control mechanisms and physical discomfort. The findings informed seven priority redesign recommendations. Given the small, purposively selected expert sample, the results should be interpreted as context-specific pilot evidence supporting iterative development rather than as generalizable evidence of effectiveness for industrial learners.
Diogo Silva, Eduardo Santos, Miguel Nunes, Carlos Costa, Francisco Silva, Arabela Fabre, Carolina Pimenta, Nuno Sousa, Iara Margolis
Open Access
Article
Conference Proceedings
Developing an Approach to Measure Trust in Flight Testing of Autonomous Aircraft Systems: An LLM-Assisted Robotic Journaling Proof of Concept
Integrating autonomous systems into high-stakes military aviation requires robust, repeatable methods for evaluating operator trust. While developmental flight testing generates exhaustive technical log data during envelope expansion, these quantitative records capture system states but fail to document the nuanced cognitive evolution of operator trust. This paper presents a case study for applying the Robotic Journaling methodology originally developed and validated for heterogeneous human-machine teams in R3 (Real users, Real systems, Real consequences) environments to autonomous flight tests. As a precursor to Phase 2 flight testing, this Phase 1 study utilizes simulated data from a software control environment to establish and validate a Trust Calibration Indicator Codebook for flight test maneuvers. The approach uses a Large Language Model (LLM) to aggregate multi-modal data on system behavior, alerts, operator interventions, and communications into a coherent narrative for qualitative analysis. Flight logs contain extensive data about what events took place during a test point; LLMs aggregate that information into a narrative that non-engineers can understand, which can then be correlated to an operator’s retrospective account of their experience on that flight. These narratives are then used as the basis for a cross-referencing analysis, a key tool for scoping the boundaries of trust calibration. This study establishes a replicable framework for analyzing how an operator's trust evolves in response to an autonomous system's performance during flight test maneuvers. The central contribution is the Trust Calibration Indicator Codebook: a structured set of observable codes derived from over-trust and under-trust theory that operationalizes trust calibration for flight tests. Phase 2 is positioned to empirically calibrate and validate these indicator codes against actual flight test data.
Nathalie Ochoa, Martin Ha, Justin Stofik, Rachel Jager, Brenda Haslup, Boyoung Kim, Nhut Ho
Open Access
Article
Conference Proceedings
BrailleBridge: A Cost-Conscious Mechatronic Prototype for Digital Text-to-Tactile Braille Conversion
Refreshable Braille displays enable tactile access to digital text, but their cost and mechanical complexity can limit accessibility. This paper presents BrailleBridge, a modular electromechanical prototype developed to investigate the feasibility of cost-conscious refreshable Braille technology. Design requirements were informed by consultation with specialists in visual disability and Braille education. Two complementary prototypes were developed: an eight-cell electronic module with 48 independently controlled light-emitting diodes (LEDs) for text-to-Braille translation and navigation, and a three-cell tactile prototype with 18 solenoid-driven pins arranged in a staggered configuration. Following targeted corrections, the software passed all 74 selected host-based cases covering Spanish Braille translation, input handling, and eight-cell pagination and navigation. Electronic functionality was also evaluated on the assembled hardware. Mechanical testing demonstrated independent pin actuation and tactile pattern generation, while revealing intermittent sticking and return failures during repeated operation. In a retrospective follow-up interview, one blind, Braille-literate participant reported recognizing a short word through touch and navigating the display using physical controls. The findings support the initial technical feasibility of the modular architecture and identify mechanical repeatability, thermal management, and further user evaluation as priorities for continued development.
Andrea Sofía Benítez Moreno, Sergio Alberto Navarro Tuch
Open Access
Article
Conference Proceedings
From Heuristics to Evidence: A Comparative Usability Evaluation and Design Implications for Space Mission Planning Tools
Space mission planning tools have evolved to support increasingly complex analytical tasks, yet their interfaces often impose steep learning curves and usability barriers for students, engineers with varying levels of expertise, and users with diverse cognitive and sensory needs. Previous research addressed these challenges by developing a human-centered framework consisting of aerospace-specific accessibility heuristics, an architecture ruleset, and interface design guidelines grounded in Universal Design for Learning (UDL) combined with Nielsen's usability heuristics. Building on this foundation, this paper presents the implementation and formative comparative usability evaluation of TARA (Toolkit for Astrodynamics Research and Application), a web-based mission planning platform designed to incorporate these principles into practical engineering workflows. A within-subjects evaluation was conducted by performing representative mission planning tasks using both TARA and NASA's General Mission Analysis Tool (GMAT). Task completion outcomes, execution times, usability ratings, and qualitative feedback were collected to assess workflow efficiency, perceived ease of use, and user confidence. This study demonstrates a systematic approach to translating usability heuristics into aerospace software design and evaluating their effectiveness through practical user interaction. Results revealed TARA's task-oriented architecture facilitated faster mission planning, while visualization controls and icon discoverability introduced challenges in navigation and interpretation. Participants generally perceived TARA as easier to learn and less complex.
Tia Bajaj, Madhusudan Vijayakumar, Rajeev Ramprasad
Open Access
Article
Conference Proceedings
Digital Avatars and Intelligent Recommendation: Design Approach and Study of Personalized User Experience in the Textile and Clothing Sector
Evaluating fit and drape in online apparel retail remains challenging, with limited human-factors data on consumer expectations for avatar-based virtual try-on. Within the TEXP@CT project, which develops a business-to-consumer (B2C) virtual try-on platform, design thinking was applied as a documented research procedure to examine how consumers perform with a try-on prototype, which avatar parameters they select as most relevant, and which design actions follow from their feedback. A task-based evaluation of a high-fidelity prototype was conducted with 31 consumers, each performing ten tasks. On average, participants achieved a 95.2% task completion rate with a mean time of 3:48 minutes and rated difficulty of use at 4.5 out of 5 (1 = very difficult, 5 = very easy). Users prioritized hips, chest/bust (74.2% each), waist (67.7%), and height (64.5%) for avatar parameters, and favored 360° rotation (87.1%) and walking (77.4%) movements. Qualitative feedback highlighted unnoticed controls (29.0%) and unclear icons, while confirming garment size as the primary driver for past returns (67.7%). Fifteen findings (six issues, nine opportunities) were translated into design knowledge and design actions, including credible fit cues, a common core of body parameters, direct body-shape adjustment on the avatar and explicit affordances for secondary controls, illustrating how user evidence can be traced to design decisions for a try-on tool.
DIogo Lopes, Isabel Varajão, Nuno Sousa, Yusbel castilla, Yanelys llerena, Pedro Pereira, Pedro Magalhães
Open Access
Article
Conference Proceedings
From Human Narrative to Harmonic Structure: A Human-Centered Investigation of Algorithmic Music Generation through the Chord Wheel Diagram
Contemporary algorithmic and AI-based music generation can produce compositions that satisfy formal requirements of tonality, harmonic progression, and musical coherence. Nevertheless, structural plausibility raises a fundamental question: whether musical expression can be adequately described by formal and mathematical properties alone. Human composers create music within personal, emotional, and cultural contexts that may influence decisions concerning harmony, repetition, tension, resolution, and deliberate departures from established patterns. This study investigates six narrative-driven popular songs by Bob Dylan, Johnny Cash, and Ritchie Valens with documented compositional, biographical, or cultural contexts. Original human harmonies are compared with outputs of an explainable computational harmonizer operating on the same melodies without access to the original chord progressions. Harmonic vocabulary, functional persistence, repetition, non-diatonic events, and tension–resolution patterns are examined. The Chord Wheel Diagram and BPMN-based representation provide complementary interpretable views: tonal-functional space and temporal harmonic development. Results show that high melody–chord compatibility does not necessarily imply preservation of the original human harmonic decision pattern. Some generated harmonizations retain the economical functional structure of the reference, whereas others increase harmonic diversity, reduce characteristic persistence, or suppress distinctive harmonic events while remaining highly compatible with the melody. Rather than quantifying artistic quality or assuming causal relationships between narrative and individual chords, the study introduces narrative-conditioned harmonic structure as a complementary perspective for computational music analysis. The findings suggest that generative systems may benefit from modeling not only harmonic correctness, but also structural identity, context, and human compositional intention.
Josef Pavlicek, Petra Pavlickova, Irena Štrausová
Open Access
Article
Conference Proceedings
Constructing a best-fit measurement scale for evaluating IEQ, Productivity, and Wellbeing for Individual Attributes of Employee workspaces
Employees spend extended time confined within their workplaces where workspace conditions can influence their wellbeing and productivity. Indoor Environmental Quality (IEQ) factors are therefore important consideration in workspace design. However, existing workspace assessment approaches evaluate environmental quality, wellbeing and productivity separately limiting their ability to account for differences in individual employee preferences. This study addresses this gap through the development and preliminary validation of an integrated measurement scale informed by IEQ, Flourishing Theory and Person-Environment Fit Theory and includes individual user attributes. Measurement items were adapted from previously validated instruments and organized into constructs assessing IEQ factors, personalisation and overall comfort. A quantitative approach was adopted, with the adapted measurement scales pilot tested among 25 office-based employees in South Africa. Descriptive statistics were used to examine participants’ responses, while Cronbach’s alpha was used to assess the internal consistency of the constructs. Five of the six constructs demonstrated acceptable to excellent internal consistency although these findings are interpreted cautiously given the small pilot sample. The findings provide preliminary support for the measurement properties of the adapted scales while identifying areas requiring further development. The proposed integrated approach contributes to workspace assessment by considering individual user perceptions alongside IEQ, wellbeing and productivity. Further validation with a larger sample is required to examine the construct structure and strengthen the psychometric properties of the proposed scale. The study provides a foundation for developing a more comprehensive and user-centred tool to inform the design of work environments responsive to diverse employee needs.
Mosa Ramohlokoane, Bankole Awuzie, Thabelo Ramantswana
Open Access
Article
Conference Proceedings
From Individual Songs to Music Corpora: Evaluating Explainable Harmonic Decision-Making on HookTheory
Automatic melody harmonization is often evaluated through individual compositions or data-driven corpus models. This study examines whether an explainable harmonic decision model developed for individual songs can be applied to independently collected human annotations without changing its rules. We evaluate a frozen Explainable Melody Harmonizer (EMH) on selected HookTheory excerpts containing aligned melody and chord annotations. EMH generates chord sequences from melody alone using explicit criteria for melodic compatibility, harmonic function, persistence, cadence, meter-aware segmentation, and Viterbi-like optimization. Human chords are withheld during generation and used only for post-hoc comparison. Nineteen excerpts yielded 439 evaluable harmonic segments. Exact chord agreement reached 20.50%, chord-root agreement 31.66%, and harmonic-function agreement 47.62%, with functions assessed relative to the key inferred by EMH. Decision-level analysis identified 203 segments (46.24%) where the human chord lay outside the model's candidate vocabulary and 146 (33.26%) where it was available but not selected. These findings distinguish representational limitations from differences in chord selection without treating the human reference as the only musically valid solution. The study demonstrates a reproducible, interpretable approach to corpus-based evaluation of harmonic decision-making.
Josef Pavlicek, Martin Pelikán
Open Access
Article
Conference Proceedings
AHP-Based Cultural Gene Extraction and AIGC-Assisted Immersive Museum Experience Design for Chenghai Lantern Riddles
With the growing use of AIGC in cultural creative design, issues such as weakened regional context, cultural misinterpretation, and superficial symbol reproduction have become increasingly evident in traditional cultural content generation. Chenghai lantern riddles, as a local intangible cultural heritage, embody distinctive dialect features, ritual rules, and spiritual meanings, requiring a more systematic culture-constrained approach for digital translation. This study takes Chenghai lantern riddle culture as its object, applies cultural gene extraction and the AHP to identify core cultural factors, and translates them into AIGC generation constraints and immersive experience design elements. Accordingly, it proposes an innovative framework of “cultural gene extraction—AIGC translation—immersive experience design.” In practice, the study develops museum-oriented immersive narrative content and integrates riddle learning, creation, sharing, and solving into an interactive process, enabling a transition from cultural cognition to behavioral participation. User evaluation results show strong performance in decorative symbol presentation, riddle ritual expression, narrative communication, and emotional identification, suggesting the potential of cultural-gene-guided AIGC generation. This study provides a quantifiable and traceable design method for the digital design of intangible cultural heritage and offers practical reference for cultural communication and living heritage transmission in museum-based immersive exhibitions.
Yi Han Dong, Xin Xie
Open Access
Article
Conference Proceedings
Cabin Layout Optimization of Underwater Transport Vehicles Based on Personnel Physiological and Cognitive State Maintenance
Operators aboard underwater transport vehicles must maintain stable physical and cognitive states to ensure mission success. Prolonged sitting in narrow, enclosed cabins often leads to fatigue, discomfort, and cognitive decline, reducing operational efficiency. To enhance working efficiency and riding comfort, this study employs a combined simulation-experimental approach to identify cabin layout schemes that mitigate the adverse effects of long-duration transportation on personnel.A full-scale cabin simulation model was first established to define reasonable ranges for inner diameter and basic seating layouts. Considering spatial dimensions, comfort, and accessibility, three seating schemes were proposed and evaluated under varying diameters. Systematic simulation narrowed the candidates to two layout modes (face-to-face and same-direction seating) and three diameters (1.6 m, 1.8 m, and 2.0 m) for further testing.Subsequently, six-hour prolonged sitting experiments were conducted on a simulated cabin platform with professional operators, comparing cognitive performance and overall comfort across the independent variables (diameter and layout). Dependent variables included cognitive level and long-term riding comfort. Results show that the face-to-face layout with a diameter of at least 1.8 m is the optimal design. This research provides a solid human factors engineering foundation and technical guidance for underwater vehicle cabin design.
Si Li, Zhanshuo Zhang, Xiaoyi Zhou, Ziang Chen, Di Yao, Hongyu Wang, Kaili Yin, Tiancheng Huang
Open Access
Article
Conference Proceedings
Living Geometry as a Human-Centered Design Framework for Context-Aware Multi-Stage Injury Prevention
Human-centered safety design requires more than detecting people and hazards; it must account for how risk emerges from changing relationships among human capabilities, behavior, objects, and everyday environments. This paper presents a prototype injury-prevention service based on Living Geometry, a framework that transforms physical observations into context-dependent relational structures such as reachability, attention, caregiver availability, and hazardous coupling. The framework connects RGB-D-based estimation of a person's position and gaze direction with route-aware reachability analysis and a multi-stage intervention strategy. Potentially hazardous appliances, including an electric fan and an electric pot, are evaluated in both floor-level and tabletop placements. For elevated hazards, the system considers not only direct distance but also whether an accessible route exists through surrounding furniture or structures. When an active hazardous relationship is detected, the service first notifies a caregiver, preserving human judgment as the primary response. If no response is obtained, a bear-shaped robot redirects the person's attention and trajectory. If the hazardous relationship persists, the system escalates to device-level control, such as turning off the appliance. A proof-of-concept implementation in the AIST Living Lab demonstrates that heterogeneous sensing and intervention mechanisms can be coordinated through a common relational representation. The contribution is a human-centered design framework that selects among interventions with different degrees of human involvement and technological control according to the evolving person-environment-hazard relationship.
Togo Hane, Yuya Kawabe, Philip Blenwall, Koji Kitamura, Mikiko Oono, Tatsuhiro Yamanaka, Yoshifumi Nishida
Open Access
Article
Conference Proceedings
Analysis of Driver Risk Perception at Intersections Across Various Road Contexts: A Driving Simulator Study
Road safety is directly tied to how drivers perceive risk. Recognizing and reacting to potential hazards allows drivers to take necessary preventive measures to avoid accidents. To support this, contemporary safety methodologies rely on the 'Self-Explaining Road' principle, designed to help drivers automatically adapt their behavior by clearly understanding the road layout. This study aims to measure driver reactions close intersections, a singular point on the road, through indicators recorded by a virtual reality driving simulator. Specifically, the study compared risk perception across three different configurations of at-grade intersections: a rural road and two suburban roads. The two suburban configurations differed by the presence of a pedestrian crossing in the second one. The risk factors considered in the study include approach speed, deceleration, brake reaction time, and time to collision. Deceleration was also measured by comparing the friction demand during braking with the friction available from the pavement, considering the latter as the threshold value beyond which loss of control phenomena can occur, perceived by the driver as a safety alarm. The validity of the indicators and the statistical significance of the sample were verified using dedicated statistical models. The data derived from the processing offer the necessary tools to verify the effectiveness of road design and management. The analysis highlights the need to adopt a systemic approach that integrates the surrounding environment; indeed, it emerges that the criticality of a scenario is closely correlated with the surprise factor, rather than an absolute intrinsic risk.
Maria Rosaria De Blasiis
Open Access
Article
Conference Proceedings
From “Disgust” to Acceptance: A Participatory Design Study of Insect-Based Foods Based on Cognitive Friction
In the context of sustainable dietary transformation, insect protein is regarded as an important food resource with high efficiency and low carbon characteristics, but its promotion is limited by cultural taboos and psychological aversion. This paper takes insect food as the object, based on the theory of cognitive friction, combined with interviews, material experiments and participatory workshops, analyzes the key factors that affect consumers ' acceptance of insect food. On this basis, the strategy of 'cognitive friction design ' is put forward, which is verified by the practice of ' insect macaron ' and blind box packaging design. The results show that the moderate introduction of sensory and symbolic conflicts can stimulate users ' reflection, promote the change of attitude from ' disgust ' to ' acceptance ', and provide new ideas and methods for new food design.
Yuzhu Liu, Shuai Sun
Open Access
Article
Conference Proceedings
A Human-centered Multi-node Multimodal Design Framework for Interactive Installations in Cultural Heritage Sites: A Case Study of Qingcheng Mountain
Interactive installations are increasingly used to support cultural heritage interpretation and smart tourism. However, existing studies often focus on individual technologies or stand-alone installations, with limited attention to the systematic coordination of visitor needs, spatial nodes, and interaction modalities. This study proposes a Human-centered Multi-node Multimodal Design Framework (HMMD Framework) for interactive installations in cultural heritage sites, using Mount Qingcheng as a case study.The framework integrates experience economy, human-centered design, sense of place, and multisensory experience and operationalizes these perspectives through five visitor-need constructs. Survey evidence from 162 respondents informed five design requirements, which were consolidated into three core dimensions: Human-centered, Multi-node, and Multimodal. The framework was then operationalized through functioning prototypes for entrance navigation, cultural interpretation, and landscape experience. Cross-case analysis examined the correspondence between the survey-derived design requirements, the three framework dimensions, and the interaction strategies implemented in each prototype. The cases provide application-oriented evidence of the framework’s design applicability by linking visitor needs, site-specific nodes, and appropriate interaction modalities within a coherent visitor journey.
Weizhen Xiao
Open Access
Article
Conference Proceedings
Exploring Multisensory Interaction in Fibre Art: Implications for Design Strategies
Fibre art has the potential to support multisensory interaction through material flexibility and the combined use of textile texture, colour, and spatial organisation. This study examines material properties, spatial environments, and audience participation from a multisensory perspective to identify interrelated and complementary design strategies for multisensory interaction design. This study adopts an exploratory multiple-case study approach. Five projects were selected through purposive sampling. Within-case analysis and cross-case comparison examined four dimensions: material properties, sensory modalities, modes of participation, and interaction feedback mechanisms. The findings show how multisensory interaction is organised across these dimensions and identify three design strategies: multisensory integration, embodied interaction, and interaction feedback. The study extends artwork analysis beyond visual form to encompass material perception, audience participation, and spatial environmental experience. The findings offer guidance for material selection, sensory coordination, and the design of audience participation.
Heyao Li, Zuojun Yu
Open Access
Article
Conference Proceedings
Designing for Behaviour Change: A Systems Design Approach to Sustainability in Cosmetic Consumption
This research investigates overconsumption in the cosmetics industry, which contributes to environmental harm, particularly plastic packaging waste. The cosmetics industry generates substantial plastic packaging waste annually, much of which is difficult to recycle. This waste stems from a culture of overconsumption, in which consumers spend heavily on cosmetic products yet use only a fraction of what they buy. Secondary research revealed that psychological, informational, and social factors shape overconsumption and waste, including impulse purchasing, social standards, an influencer-driven market, and limited consumer literacy about cosmetic product specifications. Social media is a major accelerator, contributing to unrealistic beauty standards and never-ending “trends” that normalize overbuying. The study conducted interviews and analysed consumer discourse to identify key consumer behaviours. The study developed a framework comprising four stages of cosmetic consumption: Exploratory, uninformed and with limited product knowledge; Influenced, peak spending driven by trends and recommendations; Aware, recognizing overconsumption behavior and initiatives to reduce it; and Minimalist, maintaining routine with a replenishment-only mindset. Most spending patterns occurred during the Exploratory and Influenced stages. A key finding across the interviews and consumer discourse analysis was that seeing many of the products they had accumulated increased participants' awareness of their own consumption and led to an overall change in how they view future purchasing. This suggests that increasing the visibility of one’s own consumption patterns, rather than physically or intentionally restraining purchases, may serve as an effective catalyst for sustainable behavioral change. Based on these findings, the study proposes a systems design approach that makes consumers’ existing cosmetic collections and consumption patterns more visible, supporting greater self-awareness of overconsumption and more informed purchasing decisions. By helping consumers recognize their product inventory, including unused or unfinished items, the system aims to shorten the Exploratory and Influenced stages, reduce redundant purchases, and decrease product and packaging waste associated with cosmetic overconsumption.
Eubin Tak, Min Kang
Open Access
Article
Conference Proceedings
The Impact of Visual UI Elements on Information Comprehension in Websites
Visual user interface (UI) elements can enhance engagement but may also affect how users understand information. Our previous study compared six existing websites and identified two broad user experience (UX) dimensions: cognitive-functional clarity and emotional-aesthetic engagement. Static-oriented websites tended to support clarity and operability, whereas dynamic-oriented websites tended to enhance visual appeal. However, differences in content, brand familiarity, and information architecture prevented the effects of visual elements from being isolated. The present study addressed this limitation using an original fictional tourism website, "Soraka Island." A 2 x 2 between-subjects design manipulated static visual elements and dynamic visual elements at low and high levels, producing four conditions. 44 participants, 11 per condition, browsed the website and created a travel plan. Questionnaire responses were collected using a five-point Likert scale. Medians and interquartile ranges were calculated, and the four conditions were compared using the Kruskal-Wallis test followed by Dunn's test with Holm correction. Four of 31 questionnaire items showed significant overall differences. The high-static/high-dynamic condition was rated higher than the low-static/low-dynamic condition in visual attractiveness. The high-static/low-dynamic condition was rated higher than two other conditions in perceived liveliness. For the perceived helpfulness of pop-up and hover displays for information comprehension, the low-static/low-dynamic condition was rated higher than the low-static/high-dynamic condition. Perceived movement also differed overall, although no pairwise difference remained significant after correction. These findings suggest that visual richness can improve attractiveness, while increasing dynamic presentation does not necessarily increase the perceived helpfulness of dynamic displays for information comprehension. Effective website design therefore requires balancing visual engagement with cognitive clarity.
Koki Takashima, Lei Luo, Shuhei Wada, Masato Takanokura
Open Access
Article
Conference Proceedings
Differential Sensitivity of EEG, ECG, Respiration, And GSR To Mental States Classification: A Comparative Machine-Learning Analysis
Understanding a pilot's cognitive state is critical for improving flight safety, optimizing human performance, and enabling intelligent adaptive cockpit systems. Although electroencephalography (EEG) has long been considered the benchmark for measuring cognitive workload, there has been a growing interest in the ability of other physiological signals, such as electrocardiography (ECG), respiration, and galvanic skin response (GSR) to reliably detect changes in mental state either individually or combined. This study uses four supervised machine learning algorithms: Logistic Regression, Support Vector Machine (SVM), Random Forest, and Multilayer Perceptron (MLP) to compare the sensitivity of these physiological modalities for classifying pilot mental states and examines whether multimodal physiological fusion provides superior performance. Model performance was assessed using Macro-F1 score, confusion matrices, and state-specific recall. The study showed that fusing multiple physiological signals consistently produced the most accurate mental-state classification. Multimodal fusion achieved the highest overall performance with Logistic Regression (Macro-F1 = 0.973); outperforming every individual single modality. Among the individual physiological modalities, MLP achieved the highest EEG performance (Macro-F1 = 0.930). More importantly, the fusion of ECG, Respiration, and GSR substantially improved classification accuracy with a maximum Macro-F1 score of 0.902 using the SVM classifier. This represents an approximately 18-37% point increment over the individual peripheral modalities. Overall, the study highlights the value of multimodal physiological sensing for mental state assessment. This work supports the development of robust, real-time pilot monitoring systems capable of enhancing adaptive cockpit technologies, decision-support systems, and human-centered aviation operations.
Francis Logah, Younho Seong
Open Access
Article
Conference Proceedings
Evidence-Informed Modular Design for Restorative Campus Open Spaces: A Case Study at Harbin Institute of Technology
University students need accessible places for short recovery during intensive study periods, yet campus open spaces often provide limited support for switching among rest, communication, reading, and light dining. This study examines how user evidence can inform a modular open-space prototype at Harbin Institute of Technology. The research combined a site survey of four campus spaces, behavioral observation in two study-intensive settings, a questionnaire with 95 valid responses from 101 collected, and 12 semi-structured interviews. Among the questionnaire respondents, 77.9% reported high end-of-term pressure. A 67-response evaluation subset rated existing rest spaces positively overall, but functional suitability had the lowest mean and greatest dispersion. Across the full questionnaire, short rest was the most frequently selected function (76.1%), followed by communication (55.2%), reading (47.8%), and light dining (47.8%); semi-open or semi-enclosed space was selected by 62.1%. In the 28-response on-site questionnaire, environmental preferences favored wood, low-saturation colors, and warm tones; greenery was supported qualitatively by observation and interviews. These findings were synthesized into three design requirements: functional diversity, spatial adaptability, and restorative environmental quality. The resulting modular proposal provides different activity settings and degrees of enclosure. As an unbuilt conceptual prototype, it demonstrates a traceable evidence-to-design pathway rather than verified post-occupancy effectiveness.
Tian Wang, Yuning Jiang, Jingyi Mu, Yuanhong Ma
Open Access
Article
Conference Proceedings
When More Becomes Too Much: Cognitive Support and Interpretive Constraint in Generative-AI-Assisted Reading of Chinese Classical Poetry
Generative AI can enrich classical Chinese poetry reading through paraphrases, images and video, but these representations also shape how poems are presented, interpreted and understood by readers. We compared four reading packages in a within-participant study with 64 Chinese readers: poem only (P), poem plus AI paraphrase (PT), PT plus a four-panel image (PTI), and PT plus a silent video (PTV). Paraphrases improved perceived comprehension, vividness and engagement and reduced mental effort, while increasing perceived interpretive constraint by 0.75 points on a seven-point scale. Static imagery further increased vividness by 1.29 points and constraint by 1.22, alongside greater engagement and lower effort. Video increased vividness by 0.43 points and distraction by 1.40; the distraction pattern persisted in sensitivity and paired-response analyses. Objective comprehension accuracy was near ceiling, with no detectable differences in the focal comparisons. These findings reveal two costs of richer AI support: stronger perceived direction of interpretation and, for video, greater distraction. With static imagery, poems became more vivid while their meanings felt more predetermined. We therefore argue that literary AI interfaces should balance experiential benefits with interpretive freedom and attention, prioritizing optional support, explicit presentation of interpretive choices and reader-controlled playback.
Rui Peng, Yuankai Lin
Open Access
Article
Conference Proceedings
Interpreting User Narratives in CAD: A Graph-Based Approach
Users’ narratives offer architects insights into real everyday experiences that can guide design processes to address diverse preferences and needs. They are, nevertheless, seldom applied within the computer-aided design (CAD) environments commonly used in architectural practice and research. This absence of user narratives within CAD may be explained by the applications being designed for formal geometric information. However, we argue that this absence indicates a potential for introducing user narratives within CAD to align computational design decisions with the preferences and needs of actual users. The pilot study presented here explores this potential by developing a plugin for the CAD application Rhinoceros that enables linking user narratives with CAD geometry. Based on protocol analysis involving 14 architects employed at the Royal Danish Academy, the study critically examines both the plugin and the underlying approach. The architects were asked to interpret a user’s preferences and needs from six user narratives linked with a Rhinoceros model of the user’s house, and to report whether and how the linked narrative-geometry might be used in other design contexts. Results show that participating architects interpreted preferences and needs relatively consistently, while their reported imagined use of the linked narrative-geometry in other design contexts varied considerably. These findings indicate practical feasibility for linking and interpreting user narratives within CAD, pointing towards directions for future studies - both to determine whether the plugin can meaningfully support design processes aiming to accommodate actual and diverse users, and to examine more specifically how CAD frameworks may constrain the use of narrative to inform computational design decisions.
Emil Ballegaard, Masashi Kajita, Isak Worre Foged, Kirsten Marie Raahauge, Jens Jørgensen
Open Access
Article
Conference Proceedings
UX challenges for conceptualizing interfaces on smart glasses with displays
Smart Glasses with displays are making significant strides in the wearable device market distinct from extended reality (XR) devices (Hillmann, 2021) by integrating into daily activities and projecting experiences like smartphone usage via interfaces displayed directly within the user's field of view. In this context, the primary challenges for UX designers creating interfaces for Smart Glasses involve visual constraints associated with compact screens, as well as usability and energy efficiency issues. These factors are compounded by the innovative nature of conversational interfaces linked to AI interactions and the way visual interfaces affect the user experience and physical interactions. The development of these interfaces expands the discourse surrounding interaction and usability limitations in Smart Glasses. This article presents research into interfaces for Smart Glasses with displays, mapping the challenges that shape UX designers' approaches. We structured an exploratory study (Gil, 2021) to gather information on the device’s UX specifications. The collected material includes manufacturer documentation and enthusiast evaluations sourced from forums and social media. The identified challenges include distinct mental models and gesture patterns that increase interaction complexity; compact displays that affect content legibility and input comprehension; colors and visual hierarchy, which impact user experience and influence interface decisions. Therefore, this article aims to broaden the discussion on Smart Glasses, considering the market's growth and the UX design challenges involved in conceiving interfaces for this new technological platform.
Alef Santos, Najara Paixao, Marcos Silbermann
Open Access
Article
Conference Proceedings
A better understanding of vibrations : First insights on the influence of vibratory stimuli on postural adjustments in the sitting posture context
Prolonged static sitting causes discomfort, neuromuscular fatigue and poor blood flow that can lead to health problems. Nevertheless, even slight changes in posture can improve comfort and spinal health. However, individuals who sit for long periods often fail to perform these adjustments due to task engagement or lack of awareness. While vision dominates task-oriented contexts, the tactile modality, particularly vibrations, remains underexplored as an intervention medium to induce postural changes. This study proposes to integrate a vibratory stimulation system in a seat to trigger postural adjustments and alleviate neuromuscular fatigue. To this aim, we implemented carefully configured dynamic vibrations, following results of our previous study that showed the positive influence of vibrations on postural muscle tone (Zenasni et al., 2025). The influence of vibratory stimuli on seated posture and perceived discomfort has so far been little examined. We therefore investigated this relationship by measuring objective parameters such as centre of pressure displacement as well as neck and torso joint angles using sensor mats and cameras. For several conditions, neck and trunk inclination angles varied significantly over time. Vibratory stimulations significantly influenced postural activity, with effects varying across multiple seat regions, particularly under the same experimental conditions. These results reveal that vibratory stimuli, depending on their dynamic parameters, can induce specific postural micro-adjustments during seated posture. Going further, we aim at investigating the influence of additional modalities, such as auditory stimulations on postural adjustments to assess whether combining tactile and auditory cues enhances postural adjustment efficiency across a wide range of contexts (transportation means, workplaces, gaming …).
Salim Zenasni, Tanguy Marqueste, Richard Kronland-Martinet, Solvi Ystad, Mitsuko Aramaki, Erick Dousset
Open Access
Article
Conference Proceedings
Unraveling the Process of Consummatory Engagement to Enable Fulfilling Personal Experiences
Recent discourse on human wellbeing has shifted from objective indicators such as economic prosperity and social status toward subjective dimensions, particularly eudaimonic wellbeing (EWB), a state concerned with self-realization, meaning, and purposeful living. Among the conditions that foster EWB, consummatory activity, engagement whose value resides intrinsically within the process itself rather than in external rewards, has been identified as a potentially significant contributor. Yet cultivating EWB requires individuals to identify and engage in activities consummatorily, which requires a deeper understanding of the motivational process through which consummatory engagement emerges, a process that remains insufficiently understood. This study aimed to clarify the motivational structure and developmental process underlying the emergence of consummatory value in personal activity. Nine adults who demonstrated consummatory engagement across activities such as cycling, swimming, cooking, driving, and map exploration were recruited through snowball sampling and interviewed individually using a semi-structured format. Data were analyzed using the Modified Grounded Theory Approach (M-GTA), a qualitative methodology in which concept generation and theory building proceed simultaneously. Analysis revealed two experiential tendencies underlying consummatory engagement: a self-realization tendency involving mastery pursuit, goal setting, and life integration, and a hedonic tendency involving novelty seeking and stress-free continuation. These tendencies were not mutually exclusive; individuals moved fluidly between or combined elements of both. These findings offer a process-level understanding of how consummatory engagement emerges and sustains itself, providing a foundation for design principles that help individuals cultivate fulfilling, meaningful experiences.
Sayaka Noda, Toshihisa Doi
Open Access
Article
Conference Proceedings
UX Research for Designing Plastic Household Products -Perspectives of Time Efficiency, Mindful Living, and Environmental Awareness-
This study explored design opportunities for plastic household products by focusing on two lifestyle perspectives: time performance and mindful living, and environmental awareness and circularity. A screening survey was conducted to identify participants who represented these lifestyle values, followed by semi-structured interviews with nine participants. The interviews explored concrete everyday behaviors and product use related to activities such as food preparation, storage, organization, cleaning, and waste disposal. The interview data were hierarchically analyzed by organizing relationships among concrete behaviors (Have), underlying goals and needs (Do), and higher-level values (Be). The results showed that participants did not simply seek to maximize efficiency or environmental performance, but balanced these values with personal preferences, effort, and priorities in everyday life. Three design opportunities were derived from the identified relationships: reducing unnecessary effort in everyday life、preserving personally meaningful experiences while supporting efficiency and integrating circularity into everyday choices and product use. These findings provide a basis for translating lifestyle values into product concepts and user experiences for plastic household products.
Luo Weiyang, Toshihisa Doi, Sayaka Noda, Kuroda Toshiki, Sayaka Yamane, Masahiro Kuwa, Shun Ando
Open Access
Article
Conference Proceedings
An Exploratory Study of Weekly Satisfaction Tracking for Long-Term UX Evaluation
This study explored what kinds of data can be obtained when satisfaction is tracked over a long period using weekly reports. Ten participants used a Fitbit Inspire 3 for eight weeks, and weekly overall satisfaction and retrospectively reported daily satisfaction ratings were collected. The data were analyzed using a linear mixed-effects model. No clear uniform temporal trend in satisfaction was observed across the eight weeks. Approximately 30% of the variability in weekly satisfaction was attributable to differences between users. Among the summary measures derived from retrospectively reported daily satisfaction, Mean and Min were significantly associated with weekly overall satisfaction, whereas Max and End were not. These findings suggest that weekly reporting may provide a relatively simple approach for tracking individual satisfaction trajectories during long-term product use. However, because daily satisfaction was retrospectively reported at the end of each week, the relationships between daily and weekly satisfaction should be interpreted cautiously. Although this was an exploratory study with a small sample and a single product, the findings provide a basis for further examining weekly reporting as an approach to long-term UX monitoring.
Toshihisa Doi
Open Access
Article
Conference Proceedings
The Role of Mental Models in Learning How to Use Smartglasses with Display Among First Time Users
Smart glasses with display introduce interaction methods that remain unfamiliar to most users, requiring them to interpret the experience through piror technological knowledge. This study investigates how first time users construct, adapt, and consolidate their understanding while learning to use this emerging category of wearable interface. Ten participants took part in two sequential sessions involving free exploration, guided tutorials, and practical tasks using the device’s touchpad and wristband. Findings show that participants relied on established mental models from smartphones, smartwatches, voice assistants, mice, joysticks, and remote controls to interpret the device and decide how to interact. However, understanding an interaction conceptually did not immediately translate into consistent practical execution. Over time, participants also shifted from initially favoring the more familiar touch interface to preferring the wristband for more complex tasks. The findings suggest that learnability in smart glasses depends on how interaction models relate to prior knowledge and on how effectively the system supports users in building new interaction repertoires.
Alan Turatti, Marcos Silbermann, Randerson Queiroz, Rodrigo Cavalcanti, Hanah Correa
Open Access
Article
Conference Proceedings
From Uncertainty to Confidence: Emotional UX and Cross-Cultural Co-Design in Campus Wayfinding
University wayfinding systems frequently develop through incremental additions, creating inconsistent signs, unclear entrances, limited awareness of one's current location, missing information at decision points, and weak connections between exterior and interior navigation. These conditions affect more than efficiency: uncertainty can produce anxiety, hesitation, dependence, and a diminished sense of welcome. This practice-based case study examines how emotional user experience principles can be incorporated into campus wayfinding through Campus Connections: Reimagining the Wayfinding Experience, a collaborative workshop involving graphic and communication design students from the University of South Dakota and Chiba University in Japan. Cross-cultural teams investigated campus facilities through observation, route analysis, emotional journey mapping, collaborative ideation, and prototyping while comparing culturally situated assumptions and negotiating shared design decisions. Preliminary analysis of workshop artifacts and critiques identified recurring challenges at entrances, transitions, decision points, and destinations. The resulting concepts emphasized information hierarchy, repeated confirmation, recognizable landmarks, accessible route guidance, visual continuity, and supportive language. Japanese public wayfinding provided a comparative reference through its frequent use of advance notice, layered cues, and considerate behavioral guidance. The study proposes six connected stages: arrival and place recognition, orientation and awareness of current location, route decision, progress confirmation, recovery from error, and destination recognition. The study presents reassurance as an essential measure of wayfinding quality, emphasizing that users should feel confident, independent, and included throughout their journey.
YoungAe Kim, Ikjoon Chang
Open Access
Article
Conference Proceedings
Designing for the Brain: The EPA Framework for Translating Cognitive Evidence into Design Decisions
Human-centered design depends on understanding how people perceive, remember, and respond to information, yet translating cognitive research into practical interface decisions often risks oversimplifying scientific evidence. This work-in-progress proposes the Evidence–Principle–Application (EPA) Framework to express and qualify evidence-to-design inferences without treating psychological findings as deterministic rules. Drawing on a selective synthesis of 31 sources across attention, perception, memory, emotion, and agency, EPA structures design reasoning across five connected stages: evidence, interpretation, candidate principle, application, and boundary condition. A constructed subscription-cancellation walkthrough demonstrates the rationale for separating review from commitment and keeping decision-critical consequences visible at the moment of commitment. A competing immediate-action design with reliable undo illustrates how changing a system assumption alters the resulting interface recommendation. EPA is positioned alongside Questions, Options, and Criteria (QOC) and Claims Analysis as a design-rationale representation, with Motivation, Engagement, and Thriving in User Experience (METUX) offering a complementary need-based perspective. Future empirical evaluation should isolate the commitment sequence from information placement and assess user comprehension, error rates, effort, and framework utility for practitioners.
Diana Lodi
Open Access
Article
Conference Proceedings
From Supervision to Self-Correction: TuneTrace, a Persistent Visual Intonation Companion for Practice Autonomy in Autistic Young Musicians
Instrumental practice requires learners to notice pitch deviations, decide what to change, and evaluate another attempt. Adult guidance can support this process, but feedback that depends entirely on adult interpretation may leave limited opportunities for learner-initiated decisions. This paper presents TuneTrace, a browser-based intonation companion designed to support practice autonomy in autistic young musicians through persistent, learner-interpretable visual feedback. The design links pitch deviations to notation, retains annotations for review, and lets learners adjust when and how much feedback they receive. Its rationale combines presuming competence, self-regulated musical learning, and research on augmented feedback. The public demonstration exposes instrument and tuning settings, measure looping, and three feedback schedules: Live Support, Measure Review, and Standard Notation with end-of-session review. An exploratory eight-week evaluation with six autistic musicians aged 11–15 describes fewer adult interventions, more reported unassisted practice time, and more unprompted re-attempts during use. Re-attempts are distinguished from successful pitch correction, and these descriptive observations do not establish causal effects or learning transfer. The contribution is a design account of persistent feedback as a resource for learner-controlled practice, together with feasibility observations and a focused agenda for technical, accessibility, and educational validation.
Lujing Yang
Open Access
Article
Conference Proceedings
Human-Centered Design of a Multi-Modal Safety Interface for RTK-GPS-Based Coastal Harvesting Mobility
Coastal harvesting activities expose workers to various safety risks, including tidal changes, unstable terrain, and physical fatigue. These risks are particularly critical for elderly harvesters, who often experience reduced physical and cognitive abilities during prolonged outdoor work. Although recent studies have developed RTK-GPS-based coastal harvesting mobility systems to improve transportation efficiency and location tracking, limited attention has been given to the design of human-centered safety interfaces that effectively communicate risk information to users. This study proposes a human-centered multi-modal safety interface for an RTK-GPS-based coastal harvesting mobility system. User requirements were derived based on coastal work characteristics, hazard scenarios, and elderly user characteristics. The proposed interface integrates visual, auditory, and haptic feedback to provide intuitive safety warnings according to different levels of risk. In addition, tide information, real-time location tracking, and emergency notification functions were incorporated to support proactive safety management during harvesting activities. The proposed interface was preliminarily evaluated through a scenario-based review involving eight experts in ergonomics, service design, and coastal safety management. The evaluation indicated positive assessments of the proposed design in terms of usability, safety, and operational applicability, providing preliminary support for its conceptual feasibility. The proposed interface extends previous mobility-centered studies by focusing on human-centered interaction between users and intelligent mobility systems. The findings provide practical HMI design guidelines for intelligent coastal harvesting mobility and contribute to the development of safer and more user-centered coastal work support systems.
Kwangtae Jung, Yejin Lee
Open Access
Article
Conference Proceedings
Exploring Experiential Value Categories for Product Use Through a Sentence Completion Test: Toward an Experiential Value Scale
Existing UX evaluation scales are useful for quantitatively assessing predefined dimensions of product experience, but they do not necessarily capture the qualitative meanings and values that users derive from their experiences. To understand product-use experiences themselves more comprehensively, it is therefore necessary to clarify what kinds of experiential values can emerge from such experiences. This study aimed to inductively identify and categorize these experiential values as a basis for developing a scale specifically designed to assess experiential value in product use. A web-based survey was conducted with participants who had used at least one product continuously for six months or longer, and 438 valid responses were analyzed. Participants recalled a positive and memorable product-use experience and completed three types of sentence completion test (SCT) items. Using a qualitative analysis procedure based on the KA method, 1,316 experiential values were extracted and categorized through affinity diagramming. As a result, 30 intermediate-level categories were inductively identified and further organized into six higher-level categories. Many of the identified values corresponded to concepts addressed in existing UX research, whereas others, including relaxation, nostalgia, and integration into everyday life, were less explicitly represented in major UX evaluation scales. These categories provide a basis for developing an experiential value scale that captures the diverse values users derive from product-use experiences.
Toshihisa Doi
Open Access
Article
Conference Proceedings
Impact of Body Mass Index, Tenure Length, and Age on Physical Exertion of Transportation Maintenance Workers during Simulated Asphalt Shoveling
Shoveling is the most frequent pushing/pulling task at the Indiana Department of Transportation and is associated with work-related musculoskeletal disorders (WMSDs), particularly of the lower back and shoulders. Physical overexertion is a well-established risk factor for WMSDs, and individual factors (e.g., body mass index (BMI), tenure, and age) may influence exertion. However, few studies have quantified physical exertion during shoveling or examined how individual factors influence it. This study examined the effects of BMI, job tenure, and age on physical exertion among 20 maintenance workers during simulated asphalt shoveling. Perceived physical exertion was assessed with the Borg 6–20 Rating of Perceived Exertion scale, and muscle activation was assessed with surface electromyography normalized to maximum voluntary contraction (%MVC) of the medial deltoid, biceps brachii, brachioradialis, and lumbar erector spinae. ANOVA with post-hoc tests was used to compare physical exertion across BMI, tenure, and age groups. Workers with obesity reported higher perceived exertion than overweight workers in the biceps brachii (p=0.045) and brachioradialis (p<0.001), and higher lumbar erector spinae activation than normal-weight (p=0.002) and overweight workers (p<0.001). Workers with 0–2 years of tenure showed higher biceps brachii perceived exertion (p=0.024) and lumbar erector spinae activation (p=0.006), whereas those with over 2 years showed higher brachioradialis activation (p=0.005). Workers aged 30–39 reported higher perceived exertion than those aged 50–59 in all four muscles, while age had no significant effect on %MVC. This study contributes to knowledge on worker groups at higher overexertion risk during shoveling, supporting demographic-specific fatigue management.
Xinran Hu, Xingzhou Guo, Yunfeng Chen, Jiansong Zhang
Open Access
Article
Conference Proceedings
Beyond True or False: Measuring Information Discernment in a Simulated Joint Operations Environment
Information disorder and adversarial exploitation of human cognition pose growing challenges for military decision making in saturated information environments. Although most research focused on veracity discernment, operational settings require broader information discernment: evaluating, contextualizing, and determining whether information should influence decisions and action. This study addressed limitation of laboratory paradigms and classified operation settings through an iterative experiment in the U.S. Air Force Academy’s Multi-Domain Laboratory (MDL), an unclassified simulation of a joint all-domain operations center. Across two semesters, measures were refined for assessing information interaction in a naturalistic, high-information density (HID) environment. Initial instruments emphasized decision and response selection but could not distinguish participants who failed to notice information from those who identified it as inaccurate or irrelevant. Fall 2025 methods incorporated component-specific, time-stamped information supplements and retrospective debrief measures of information belief and decision influence. Findings and results from 771 observations are presented to explore the debrief instrument’s utility for assessing metacognitive processes underlying information discernment without disrupting wargame execution.
Melissa Mclain, Sarah Bostrom, Richard Niemeyer, Leanne Hirshfield
Open Access
Article
Conference Proceedings
A Multidimensional Framework for Assessing Human-AI Collaboration in Military Operations: The CogARMY Teamwork Index (CTI)
The increasing integration of autonomous systems and artificial intelligence (AI) into military operations is transforming teamwork, requiring new approaches to assess performance in hybrid human–AI teams operating in dynamic, high-stakes environments. Existing teamwork assessment frameworks offer robust theoretical foundations for evaluating leadership, communication, coordination, and adaptability, but were developed primarily for human-only teams and are not fully suited to Human-Autonomy Teaming (HAT), defined by O'Neill et al. (2022) as interdependent collaboration between one or more humans and one or more autonomous agents, each fulfilling distinct roles while pursuing shared mission objectives. This paper proposes a CogArmy Team Index (CTI), a multidimensional framework for assessing teamwork effectiveness across human-only and hybrid human–AI military teams. Developed through the Cognitive Training and Teamwork Assessment of Army Personnel (CogARMY) initiative and NATO Science and Technology Organization Human Factors and Medicine research task group HFM-374, the CTI extends established teamwork theory with autonomy-specific dimensions, including trust calibration, explainability, human–AI coordination, adaptive task allocation, and human–AI synchronisation. The framework adopts a multilevel Input–Process–Output architecture linking contextual inputs to teamwork processes, coordinating mechanisms, autonomy management, regulation, and mission outcomes. The CTI leverages a multimodal, data-driven methodology integrating behavioural, physiological, and AI-derived metrics, overcoming limitations of self-report measures while enabling assessment of both human and autonomous team members. Composite scores combine theory-driven and empirical weighting to support adaptability across operational contexts. The proposed framework offers a scalable, scientifically grounded approach for evaluating and optimising teamwork in next-generation military environments, supporting operational readiness and future human–AI collaboration.
Tanya Paul, Daniel Lafond, Dale Richards, Gianluca Borghini, Johan DeHeer
Open Access
Article
Conference Proceedings
Bridging the Gap Between Human Factors and Autonomous Vehicle Decision-Making and Control: A Conceptual Framework for Behavioural Acceptance
Autonomous vehicles (AVs) have emerged as a promising technology with the potential to improve road safety, efficiency, transport accessibility, and traffic flow. These anticipated advantages have motivated substantial investment in the research, development, and deployment of AVs. Considerable progress has been made in decision-making and motion control strategies, that govern the AV behaviour, with hierarchical architectures coordinating the two. Existing assessment and feedback mechanisms have largely been limited to safety and performance related considerations, while research in human factors has established constructs determining acceptance without considering the layers that play a significant role in generation of AV behaviour being judged. To bridge this gap, this paper proposes a conceptual framework for acceptance within the context of AV behaviour. The framework comprises three main elements within a unified structure: the high-level decision-making, low level motion control, and the human factor acceptance layer. The decision-making layer establishes planning and initiation of candidate actions, while motion control layer executes the desired decision, with their combined operation producing the observable AV behaviour. This resulting behaviour forms the input to the acceptance layer, which appraises it and returns an assessment to the layers that generated it. By establishing this linkage between the three layers, the proposed architecture provides a conceptual interface for incorporating human-centred dimensions into the model-based design, modelling, simulation, and evaluation workflows of the AV systems.
Joshua Dsouza, Jisun Kim
Open Access
Article
Conference Proceedings
Bridging the Gap between Saviours and Survivors: Exploring Uncertainties in Search, Rescue, Emergency, and Evacuation Operations in Earthquake and Flood
This exploratory study reports the behaviors in situations in which transportation services are required in earthquakes and floods for search, rescue, emergency, and evacuation exit operations. The study focuses on a comparison of how two disasters, predictable i.e., floods and unpredictable i.e., earthquake happening in emerging and emerged economies and how institutions are responding with their own set of available resources and facilities. The case studies are three in number, which include a high-population density city facing extreme seasonal floods from a river, under-resourced, and belonging to a developing country. second case study is a high-population-density city, richly resourced, and an advanced city in a developing country considered for earthquakes. The third case study for floods is an advanced city in a developed country that faces floods almost every year. The data were collected through a questionnaire, and a systematic sampling technique was adopted by considering the most disaster-exposed households. The results gave a broader perspective regarding the use of information and communication technologies (ICT) used by disaster-affected residents to gain and exchange transport-disaster-related information, the frequency of ICT-usage, reliable information sources, and preferred institutions to contact among rescue authorities, police, and government authorities. The study also explores the behavioral differences in pre-, during-, and post-disaster situations in which disaster-affected people engaged ICTs in choosing evacuation vehicles, helping dependents to evacuate safely, informing other disaster-affected community members, and relying on authorities for support and services. The study is helpful in managing transport-disaster scenarios to save lives, maintain information flow among people and authorities, improve the efficiency of transport operations, build resilience, and design technologically advanced, well-connected future smart cities.
Izza Anwer, Irfan Yousuf
Open Access
Article
Conference Proceedings
Does AI Assistance Leave a Temporal Fingerprint? Detecting Overreliance in AI-Assisted Writing and Programming
The rapid adoption of generative AI has made final artifacts unreliable evidence of student learning, and AI detectors that examine only the finished product are inaccurate and ethically contentious. Process data offers an alternative, but prior work covers only English essay writing. We ask whether AI assistance carries a temporal signature, whether it generalizes from writing to programming, and whether it distinguishes ordinary collaboration from wholesale delegation. We analyze three public corpora: CoAuthor (1,447 keystroke-level co-writing sessions), RealHumanEval (editor telemetry from 243 programmer records), and a pre-LLM CS1 corpus (5.1 million keystrokes) as a human-only baseline, comparing minimal-AI work, collaborative AI use, and simulated wholesale delegation. Three findings emerge. First, the signature generalizes: AI contributions arrive in bursts far outside the author's own baseline in both mediums (paired d_z = 1.13 and 3.54). Second, engagement diverges by medium: 93% of AI-inserted characters survived to writers' final documents, while only 14% of accepted code suggestions survived intact. Third, classifiers using only observable temporal features separate simulated delegation from authentic work nearly perfectly (F1 ≥ 0.997; at most 0.5% of real work misclassified), while ordinary collaboration remains hard to distinguish from unassisted work. Temporal evidence flags wholesale delegation rather than assistance, positioning process visibility as a candidate evidentiary basis for academic integrity, pending validation in authentic coursework.
Eduardo Davalos, Yike Zhang
Open Access
Article
Conference Proceedings
Calibrating Managers' Trust in ICT-Enabled Visualized Information: A Case Study of Heat Illness Prevention at Infrastructure Maintenance Sites
In infrastructure maintenance, where worksites are geographically dispersed, ICT systems that visualize workers’ heat illness risk are increasingly introduced, yet how managers trust and utilize the visualized information remains insufficiently understood. Regarding the system as an automation technology, this study examined the process by which managers calibrate their trust, based on the human factors concepts of “trust in automation” and “appropriate reliance.” For a three-month trial introduction at an expressway maintenance and repair company, we combined a questionnaire survey (29 managers and 100 workers) with interviews (5 managers) in a mixed methods design. In the questionnaire, managers’ change in safety management awareness was significantly above the scale midpoint, and the M-GTA analysis of the interviews derived a process in which managers calibrated their trust through comparison with experiential knowledge and recognition of measurement limitations, and reconstructed their hazard recognition while adjusting their reliance under that trust. The calibrated trust appeared to connect to changes in management behavior through the justification of interventions, and as numerical information became a shared criterion for judgment, the reconstructed hazard recognition was beginning to be shared within the organization. These findings suggest that proactive and continuous trust calibration by managers can contribute to realizing the value of system introduction.
Atsunori Someya, Kotomichi matsuno, Manabu Sawaguchi
Open Access
Article
Conference Proceedings
Redesigning Work in Industrial Control Rooms: Intelligent System Interfaces That Preserve Human Decision Authority While Maintaining Efficiency
While significant research has focused on the technical performance of intelligent systems and the design of human-machine interfaces, comparatively little attention has been given to the redesign of work itself. This introduces emerging sociotechnical risks, including authority ambiguity, skill degradation, automation complacency, reduced situational engagement, and diminished adaptive capacity during abnormal operating conditions. This paper introduces Human Authority-Centred Work Design (HACWD), a methodology that treats human authority as a property of the whole work system rather than as an individual trait or a final approval step. HACWD extends the author's previously developed Human Authority Index (HAI), which conceptualises human authority as a measurable sociotechnical property comprising four domains: Decision Clarity, Cognitive Sustainability, Psychosocial Safety and Work Design. Where the HAI assesses the extent to which intelligent systems preserve meaningful human authority, HACWD translates these principles into a work design model built around four interdependent design domains: Interface Design, Work Design, Capability Design and Authority Design. Using autonomous mining operations as an illustrative context, the paper examines how intelligent systems may reshape conventional control-room roles and proposes the human authority engineer or decision architect as an example of a future operational role. This is research in progress: the HAI weightings, items and interpretation bands are preliminary and are the subject of ongoing field, psychometric and cross-sector validation. The paper positions human authority as a critical organisational capability and argues that future operational success will depend not only on technological sophistication, but on the ability of organisations to redesign work, workforce capability and decision structures around effective human-machine collaboration.The research investigates: 'How should work, roles, and decision authority be redesigned in intelligent industrial operations to preserve and enhance human capability while realising the benefits of automation and AI?'
Peta L. Chirgwin
Open Access
Article
Conference Proceedings
Is situation awareness related to pupillary changes during a flight navigation simulation?
A lack of situation awareness (SA) is often cited as contributing to Naval aviation mishaps. Yet, SA is not measured in real-time as doing so is too intrusive. Previous work suggests eye tracking metrics may be able to serve as a non-intrusive measure of SA. If so, it is critical to study how these metrics trend over time and their relation to SA. The present work begins to investigate this with nonlinear, longitudinal models of pupil size of Naval aviation trainees who completed a flight simulation in virtual reality. Specifically, pupil size during the four seconds before and after approaching the assigned waypoint was modeled via a generalized additive mixed model (GAMM). It found that pupil size differed per the waypoint and the person’s overall SA. This preliminary investigation suggests pupil size changes over time may inform a real-time measure of SA in Naval aviation environments and the like.
Shannon McGarry, Allison Bayro, Michael Kukenberger, Heejin Jeong, Joseph Coyne
Open Access
Article
Conference Proceedings
Intra-Crew Dynamics of Multi-Unit Operations in a Nuclear Power Plant Simulator
Although multi-unit operation has long been standard practice in non-nuclear facilities, its application in the nuclear industry remains relatively immature, with NuScale bringing the concept to prominence in 2016. To prepare for a larger study on multi-unit crew dynamics, a pilot study was conducted using a nuclear power plant simulator. The initial goal of the pilot was to assess the research infrastructure, identify simulator bugs, and refine the study plan. The pilot revealed a different critical and unexpected finding: a significant gap in intra-crew communication capabilities introduced by the multi-unit environment. Research (e.g., Park et al., 2023) demonstrated that data from student participants can be a useful proxy for professional operators in single-unit single-operator operations. To test multi-unit operations, our pilot involved simulator sessions utilizing two groups of two student Reactor Operators (ROs), with a single confederate researcher acting as the Senior Reactor Operator (SRO) across all sessions to maintain consistency. The sessions revealed that the multi-unit environment exacerbates the differences between students and trained professionals. Professional operators are highly trained in specific, formalized communication protocols, whereas the student ROs frequently failed to communicate their actions or announce specific milestones in the expected manner during reactor startups and shutdowns, leaving the SRO under-informed. This paper documents this communication gap, highlighting the possibility that the relationship between data ascertained from student operators and qualified professionals may fundamentally change in a multi-unit, multi-operator context.
Dylan Jurski, Thomas Ulrich, Ronald Boring, Justin Watson, Emma Hone
Open Access
Article
Conference Proceedings
From Recommendation Correctness to Clinical Action: Transparency, Trust, and Verification in AI Decision Support
Artificial intelligence assisted clinical decision support systems are commonly evaluated using aggregate measures such as accuracy, sensitivity, and specificity. These measures do not establish whether a recommendation is dependable for an individual case. Clinical safety also depends on how interfaces communicate reliability and how clinicians interpret, verify, and use the output. This paper examines the pathway connecting AI recommendation correctness to clinical action through transparency, trust calibration, and verification. A qualitative, theory guided integrative framework approach was used to synthesize peer reviewed empirical and theoretical literature identified through Scopus, Web of Science Core Collection, and PubMed. The full multilevel safety framework provides context for a focused conceptual pathway linking recommendation correctness, interface communication, clinician reliance, verification, and action. The reviewed evidence suggests that correct recommendations can improve diagnostic performance, while incorrect recommendations can impair judgment when accepted without sufficient evaluation. Explanations, confidence displays, and uncertainty cues may help clinicians assess recommendations, but can also create false confidence or additional cognitive demands. Their value depends on whether they support interpretation and verification during clinical work. Clinician responses may include questioning, modifying, independently checking, or overriding an output. The proposed pathway identifies opportunities for interface design to support informed reliance and preserve clinician agency. It emphasizes communicating uncertainty and helping clinicians compare recommendations with clinical evidence before acting. These relationships remain conceptual and require empirical evaluation. Future studies should examine how reliability and uncertainty displays affect verification, reliance, diagnostic performance, and workload in realistic clinical workflows.
Simranpaul Dhami, Dennis Vincenz, Kat Moran
Open Access
Article
Conference Proceedings
Autonomous Hospitality Yachts: Designing Trustworthy Human-AI Experiences in Future Maritime Environments
The development of Maritime Autonomous Surface Ships (MASS), artificial intelligence and cyber-physical systems is reshaping future marine environments. While maritime autonomy has mainly addressed cargo transport, surveillance and industrial operations, the leisure marine sector remains underexplored. This paper proposes a human-centred conceptual framework for Autonomous Hospitality Yachts (AHY), understood as supervised intelligent vessels integrating marine autonomy, AI-driven hospitality services, remote oversight and adaptive human-AI interaction. The framework combines a layered system architecture with four human-factors requirements - perceived safety, trust in AI behaviour, emotional comfort and perceived operational reliability - grounded in neuroscientific and cognitive models of guest experience. Rather than advocating fully unsupervised yachts, the paper argues for progressive, domain-bounded and explainable autonomy aligned with MASS principles and Operational Design Domains. The framework is intended as a research and design model requiring future empirical validation, not as a validated deployment architecture.
Laura Pagani, Paolo Gemelli, Alessandro Bertirotti, Mario Ivan Zignego
Open Access
Article
Conference Proceedings
Assessing Embodied-AI Humanoid Robots for Construction: A Task-Capability-Human-Compatibility (TCHC) Framework
Embodied-artificial-intelligence (AI) humanoid robots offer a potential pathway for automating human-scale construction work, yet advances in isolated robot skills do not establish readiness for complex construction operations or effective participation in human crews. This paper proposes a Task-Capability-Human-Compatibility (TCHC) framework to assess humanoid suitability at the level of construction action sequences and collaborative work. The framework comprises three connected stages. First, building construction work is decomposed from trade operations into executable actions, with contextual, physical, quality, safety, and interaction requirements. Second, actions are mapped to humanoid capabilities in locomotion, perception, manipulation, force regulation, planning, and adaptation. Third, six Human Embodiment Compatibility (HC) dimensions conceptualize morphological, environmental, tool-handling, sensorimotor, interaction, and cognitive-social compatibility. The integrated TCHC framework is illustrated through qualitative construction task scenarios involving carpentry tool handling, roofing, and piping placement. The cases reveal distinct bottlenecks: precise and safe tool use, stable contact-rich work on sloped surfaces, and coordinated handling within constrained shared workspaces. These analyses do not constitute empirical robot validation; rather, they identify testable capability and human-humanoid teaming requirements. TCHC contributes a construction-focused bridge between physical humanoid capabilities and human-autonomy teaming, supporting more discriminating use-case selection and future experimental evaluation.
Di Liu, Mengyu Liu
Open Access
Article
Conference Proceedings
Beyond the Employer: Solar UV Risk Profiles of Self-Employed Farmers and Artisanal Fishers
Occupational exposure to solar ultraviolet (UV) radiation causes nearly one in three deaths from non-melanoma skin cancer worldwide, according to WHO/ILO joint estimates. Most sun-safety frameworks, however, rely on employers to provide shade, organize work schedules, and supply protective equipment, leaving self-employed outdoor workers outside this model. This study compared the sun-safety risk of two self-employed occupations in central Chile, small-scale farmers (n = 138) and artisanal fishers from four coastal communities (n = 148). Data were collected in 2018–2019, before the COVID-19 pandemic. The Solar UV Risk Assessment for Outdoor Workers (Sun Safety at Work Canada) was administered in person. It scores three dimensions, environment, operational factors, and personal protection, whose sum gives an overall risk score (high risk > 98 points). Fishers were assessed in two work positions, navigation at sea and shore-based tasks. Both occupations scored high on environment (farmers 44, fishers 46; high > 38). Farmers had a mean overall score of 109.8 ± 7.9, with 91.3% at high risk. Their risk was exposure-driven: the operational score was high (43.4 ± 6.5; high > 37) and the personal protection score was medium (22.4 ± 4.2). All farmers worked in the sun throughout the day even though nearly all reported being able to reschedule tasks, 62% lacked shade during work, 47% reported regular sunburn, and 97% reported regular exposure to photosensitizing agrochemicals. Fishers showed the opposite profile, protection-driven risk: operational scores were medium (20.4 at sea, 25.0 on shore) and personal protection scores were high (25.8 and 31.3; high > 25). None used UV-protective clothing or UV-filtering eyewear, and 75% never or rarely used sunscreen. Shore-based work raised the overall score from 92.2 ± 10.2 to 102.3 ± 7.9 (p < 0.001), the proportion at high risk from 30% to 66%, and peak-hour exposure from 18% to 41%. These contrasting profiles show that sun-safety needs differ by occupation and task, and that autonomy over work schedules does not translate into protective behavior. Protecting self-employed outdoor workers requires occupation-specific, human-centered strategies delivered through cooperatives, unions, fishing communities, and primary health care.
Rodrigo Dominguez, Maria Paz Caceres, Fanny Mondaca, Sebastian Amaro
Open Access
Article
Conference Proceedings
A Human–AI Collaborative Mechanism for Commercial Packaging Design with Generative AI
This study The rapid adoption of generative artificial intelligence (GenAI) has significantly expanded its role in commercial packaging design, enabling designers to produce visual concepts with unprecedented speed. However, existing studies have primarily emphasized generation efficiency and visual quality, while paying limited attention to the stability of generated results, their commercial applicability, and the evolving relationship between designers and AI throughout the design process. This study investigates the aesthetic generation characteristics and practical limitations of GenAI in commercial packaging design and proposes a designer-centered human–AI collaborative creation framework. A mixed-method approach combining literature review, controlled experiments, and case analysis was employed. Under identical generation conditions—including the same model, prompts, and generation parameters—commercial packaging images were repeatedly generated and comparatively analyzed in terms of visual style, packaging structure, brand consistency, cultural expression, and commercial feasibility. The results indicate that GenAI can efficiently produce visually coherent design concepts while maintaining a consistent overall stylistic direction. Nevertheless, substantial variations remain in packaging structure, typographic legibility, brand identity, material organization, and cultural interpretation due to probabilistic sampling, training data distribution, and model-specific aesthetic preferences. These findings suggest that current GenAI systems are effective for early-stage ideation but remain insufficient for directly producing commercially deployable packaging solutions. Based on these findings, this study proposes a designer-led human–AI collaborative workflow in which designers retain responsibility for brand strategy, cultural interpretation, aesthetic judgment, and final decision-making, while GenAI supports concept exploration, design expansion, and iterative refinement. The proposed framework provides both theoretical insights and practical guidance for integrating generative AI into commercial packaging design while preserving design quality, brand coherence, and commercial viability.
Zibin Liu
Open Access
Article
Conference Proceedings
Assessing Organisational Capability for Meaningful Human Oversight of AI-Supported Work: A Human Factors Framework
Human oversight is adopted as a control measure for workplace artificial intelligence; assigning responsibility does not establish whether workers can identify errors, question recommendations, or intervene effectively. This paper proposes a human factors framework for assessing conditions that enable meaningful oversight within a defined AI-supported task. The framework was established using a purposive conceptual synthesis of human factors research, organisational studies, EU AI Act provisions, and AI assessment guidance. It builds on questions raised by the authors’ published study (Lawal, Koppel, and Putans, 2026), which provided background motivation. Its findings are not presented as new evidence or validation of this proposal. The assessment covers four areas: understanding, intervention authority, working conditions, and organisational support. Each specifies assessment questions, evidence of effective implementation, and organisational responses to gaps. The procedure defines the task, identifies personnel responsible for performance and supervision, and specifies plausible erroneous AI recommendations. It assesses whether overseers have sufficient information and time to evaluate outputs and consequences and respond appropriately. Evidence may include organisational records, direct observation of human-AI work, and realistic failure exercises assessing recognition, challenge, and response. Findings distinguish absent evidence from demonstrated deficiency within each domain. A hypothetical maintenance allocation example illustrates organisational barriers to intervention. The framework translates oversight expectations into workplace assessment criteria and organisational responses. It remains empirically unvalidated. Operational evaluation should compare it with existing approaches and examine content coverage, usability, inter-assessor reliability, and its relationship to personnel’s intervention capacity. The proposal does not establish legal compliance or demonstrate improved safety.
Ayoola Rilwan Lawal, Romans Putans
Open Access
Article
Conference Proceedings
Emerging Technologies and AI in Aviation: Reshaping the Roles and Responsibilities of Human Factors
Artificial intelligence (AI) is generating significant interest across the aviation industry, with applications such as aircraft design, operations, aviation data analysis, aviation maintenance and aircraft retirement. The Federal Aviation Administration (FAA), industry stakeholders, and researchers are actively exploring how these technologies can improve aviation safety, efficiency, operational capability, and potentially reduce human error. Historically, human error has accounted for over 80% of aviation accidents, yet the aviation industry is an ecosystem that involves humans at every level. However, before AI-enabled systems can be widely deployed in industry, they must demonstrate compliance with aviation’s rigorous safety requirements. This paper examines how AI transforms human factors in the aviation ecosystem. It considers the implications of AI across different phases of aviation, from design, system safety, training, workload, decision-making, trust, explainability, regulatory oversight, and risks associated with it.
Yuwei He, Kris Borgen
Open Access
Article
Conference Proceedings
Assessment of Curved Body Armor Plate Designs for Improved Protection and Mobility in Soldiers
This study aimed to increase warfighter survivability and operational capacity. Reported discomfort and mission performance from poor anthropometric fit (Mitchell et al., 2023) suggests that body armor design for female warfighters warrants further refinement. Alternative body armor designs were evaluated to improve fit, performance, and comfort by better accommodating female anthropometry. Three systems were tested (A) the Army standard issue ESAPI, (B) a novel sizing system, and (C) the enhanced female curvature with B’s sizing. Participants also completed the tasks in a baseline with no plate. Thirty-four female participants were classified into two breast volume groups (Small and Large) based on the difference between Chest and Chest at under Breast Circumferences. Participants underwent anthropometric measurements for plate sizing, plate fitting, performance assessments (including range of motion (ROM) and treadmill running), and subjective ratings of comfort, pain and fit. However, objective performance benefits were observed only for standing trunk flexion. Seated trunk flexion and cross body reach did not differ significantly among plate configurations. No significant effects of breast volume or interactions between breast volume and plate configuration were observed, although descriptive trends suggested greater improvements in standing trunk flexion among participants with larger breast volumes. Pain ratings did not differ significantly among plate configurations, potentially due to the short wear duration. Future research should evaluate longer wear periods and a broader range of Soldier relevant tasks to better understand the operational benefits of plate designs with enhanced curvatures.
Meghan DeClue, Scott Marriner, Lavina Burman, Iris Matthews, Blake Mitchell, Ericka Rovira, Hyegjoo Choi-rokas
Open Access
Article
Conference Proceedings
Human Testing of Spacesuit Size Preference Using 3D Printed Mockups
This study presents a novel spacesuit fit assessment method that parameterized suit-to-body interaction patterns as a predictor of fit quality. A statistical model was constructed by leveraging historical NASA fit check data and was validated via Human-in-the-Loop (HITL) testing. Fully customized Hard Upper Torso (HUTs) were developed for five subjects. Motion capture, laser scanning, and subjective surveys were used to collect data on each HUT’s impact on fit, indexing, mobility, and reach. The developed model opens the door for customized hardware and extended population analysis.
Cassandra Vining, Chris Miller, K. Han Kim, Nate Newby, Yaritza Hernandez
Open Access
Article
Conference Proceedings
Radiological Incident Awareness and Preparedness Among the Public and First Responders: A Systematic Review
This study examines awareness and preparedness for radiological incidents among the public and first responders within the context of occupational safety, risk management, and emergency preparedness systems, with particular attention to human factors and system-level performance in real-world environments. The objective of this study is to identify gaps in the literature concerning hazard knowledge, safety training practices, and operational preparedness in radiological incident scenarios. A systematic literature search was conducted across PubMed, IEEE Xplore, and Web of Science to identify English-language studies published between January 2014 and January 2024. A total of 15,047 records were identified, and 12 studies met the inclusion criteria for final review following PRISMA screening procedures. The reviewed studies primarily focused on radiation exposure and health outcomes, including cancer, cataracts, cardiovascular disease, and thyroid disorders, while limited research directly examined preparedness, hazard awareness, responder decision-making, or human performance during radiological emergencies. Most included studies were observational and focused primarily on health outcomes rather than operational preparedness. The findings indicate that the literature provides limited direct evidence on how radiation hazard knowledge translates into preparedness-related practices and limited reporting of human factors, training approaches, hazard recognition, and preparedness protocols. These findings highlight an evidence gap regarding the relationship between hazard awareness, training, and operational preparedness during radiological incidents and support consideration of an occupational radiological preparedness framework connecting these factors.
Anna Kucharski, Avishek Choudhury
Open Access
Article
Conference Proceedings
Tangible Card Topologies or Text Prompts? Comparing AI-Human Co-Creative Story writing Methods
Most AI-driven storytelling tools rely on text-prompt interaction that conceals narrative structure within the model. We developed a tangible card-topology method: users arrange story-element cards into spatial configurations, photograph them, and the AI generates text from the recognized topology. In a within-subjects experiment (N = 48) across four story contexts (multi-thread plot, emotional arc, goal-directed narrative, learning module), the tangible method yielded higher creative quality [F(1, 47) = 38.72, p < 0.001, ηp² = 0.45], 19–27% lower cognitive load, and stronger perceived creativity support (d = 1.1–1.6).
Yaojiong Yu
Open Access
Article
Conference Proceedings
Decoding the Language of Team Performance: A Machine Learning Framework for Sentiment-Based Performance Modeling and Team Situational Awareness
This research introduces a multi-faceted AI-enabled big data sentiment analytics framework for modeling and predicting team performance in high-stakes simulation environments. Communication data from 40 teams participating in a firefighting simulation were collected. Big data and sentiment analysis techniques were used to examine communication dynamics and identify relationships among individual factors, team communication dynamics, and team situational awareness (TSA) and their collective predictors of mission success. By leveraging these insights, the proposed big data analysis framework provides a data-driven approach to enhance TSA, improve team interaction, and optimize human–machine coordination. It also supports the development of adaptive feedback systems, AI/ML model training, and improving performance prediction capabilities across diverse operational contexts. The findings highlight the potential of sentiment-based analytics to uncover subtle behavioral and cognitive indicators of team performance that traditional evaluation methods often overlook. This integration of human communication data with advanced analytics provides predictive indicators for team success and training optimization. This research advances the field of industrial and systems engineering by bridging behavioral analytics and performance modeling. It provides a systematic method for transforming unstructured team communication data into actionable insights that enhance decision-making, training, and human-AI collaboration in complex environments.
Gracie Gray, Vineetha Menon, Kristin Weger, Bryan Mesmer, Nicholas Loyd
Open Access
Article
Conference Proceedings
The AXTENI Framework: Redefining Leadership Authority in the Age of Agentic AI
Commercial aviation is rapidly integrating artificial intelligence across maintenance, dispatch, and customer service, but adoption is slower on the flight deck and in air traffic control. As AI expands across airline operations over the next decade, both direct supervisors and organizational leaders must develop a new understanding of AI governance and accountability. The Fifth Industrial Revolution has moved the industry toward agentic systems that decide and act autonomously, yet prevailing leadership frameworks remain anchored in Fourth Industrial Revolution assumptions. This misalignment creates a serious accountability deficit in high-consequence environments, particularly within commercial aviation. This paper proposes a shift from executive oversight to active governance of AI systems. The Aviation Explainable AI Trust-Ethics Nexus Integration (AXTENI) framework provides the structural foundation, replacing static preparation with strategic readiness, and establishing rational skepticism as a core leadership duty. Applying the Human Factors Analysis and Classification System (HFACS) to predictive ramp maintenance shows how agentic AI migrates accountability upward, concentrating governance responsibility at the supervision and organizational levels where existing frameworks offer the least guidance. AXTENI is infused into this analysis to surface targeted leadership priorities across management, supervisory, and technician tiers, positioning the human leader as the final, informed authority able to interrogate and override autonomous logic.
Leila Halawi, Mark Miller, Sam Holley
Open Access
Article
Conference Proceedings
A Rank-Based Analysis of AI Consistency in Shortlisting Top Software Systems by Usability: Learning Management Systems as an Example
This study serves as a furtherance of a prior work, which demonstrated a moderate-to-strong correlational consistency between usability factor scores generated by Large Language Model (LLM)-based Artificial Intelligence (AI) systems and a human benchmark score from Capterra. While the prior work focused on parametric and non-parametric correlations, this paper investigates the consistency of the rankings produced by three AI systems (Claude-Opus-5, GPT-5.6-Sol, and Gemini-3.5-Flash) when shortlisting top-performing Learning Management Systems (LMSs) by usability as gauged by seven usability attributes based on ISO/IEC 25010, 9241 Standards and usability models proposed in the literature so far: effectiveness, efficiency, error protection, accessibility, operability, learnability, and satisfaction. Rank-Biased Overlap (RBO) was employed to compare the AI rankings against each other and against Capterra’s human benchmark score. The results show that all AI systems exhibit moderate-to-high ranking agreement with each other (RBO up to 0.878). However, the agreement between the AI rankings and the human benchmark score is substantially lower, diminishing from a moderate level at a shallow depth for top 10 LMSs (RBO ~ 0.7) to a rather weak level for the full list of 97 LMSs (RBO < 0.2). These findings indicate that while AI systems are highly consistent among themselves in ranking LMSs by their usability, their rankings somewhat diverge from the human benchmark as more LMSs are considered. This suggests that AI may be a viable tool for rapidly identifying a shortlist of potential top-usability LMSs but is not yet a reliable replacement for comprehensive human evaluation. Such findings on rankings offer more practicality than the prior work on correlations.
Victor K Y Chan
Open Access
Article
Conference Proceedings
Designing with and Beyond Humans: An LLM-Based Human–AI Co-Creation Framework for Remote Curatorial Collaboration
As cross-border curatorial practices become increasingly common, remote collaboration has become an inevitable mode of work for curatorial teams—including curators, artists, and exhibition production teams—yet it also brings recurring challenges such as fragmented communication context, difficulties in consensus building, and reduced decision-making efficiency. Recent advances in large language models (LLMs) and multi-agent systems enable AI to shift from a task-oriented assistant to a reflective, collaborative co-designer that can reason, dialogue, and integrate perspectives across roles, helping teams form consensus more effectively in complex situations. Building on this premise, this study proposes an LLM-based self-collaborative design mechanism and constructs a Human–AI co-constructed curatorial knowledge model. In-depth interviews with curators and artists in real-world curatorial contexts were conducted and analyzed to identify key pain points and latent needs in remote curatorial collaboration, which were then translated into concrete design issues. A structured multi-agent discussion and reasoning process was conducted to generate actionable, traceable, and more feasible insights. The outcomes contribute: (1) a replicable LLM-based multi-agent research framework that provides an end-to-end methodological pathway from field data to multi-role reasoning, and from consensus extraction to guideline generation; and (2) a set of design guidelines for remote curatorial collaboration, offering strategic solutions and an eight-layer functional architecture. Therefore, this study not only addresses the practical needs of remote curatorial collaboration but also provides a knowledge-production mechanism that can be directly integrated into workflow and tool design, enabling teams to sustain collaborative quality and decision-making efficiency under resource constraints and high uncertainty.
Yi Pei Wu, Chien-Hsu Chen
Open Access
Article
Conference Proceedings
"Listen, Who is Anxious?": A Multi-Agent Coordination and Self-Dialogue Mechanism Based on Internal Family Systems Theory
As AI increasingly becomes a "digital tree hole" for emotional expression, existing tools remain limited by their lack of theoretical grounding, susceptibility to sycophancy, and treatment of the self as a singular entity. This study integrates Internal Family Systems (IFS) theory with multi-agent AI technology, externalizing internal parts — Managers, Firefighters, and Exiles — as distinct agents in a fully automated conversational system for mild everyday anxiety. Ten participants completed an IFS pre-assessment, a multi-agent dialogue session, and a post-session interview. Qualitative findings suggest that participants experienced the agents' contrasting stances as challenging their habitual thinking, and several described shifting toward a more observational "Self" perspective. Most participants described the automatically generated emotional summaries as cognitive anchors that organized scattered emotional content into structured self-understanding. These exploratory findings suggest IFS-grounded multi-agent design as a promising direction for emotional well-being in HCI, warranting controlled evaluation.
Ni Jin Li, Rung-Huei Liang
Open Access
Article
Conference Proceedings
Qualifying NEV User Experience: LLM-Based Dimensional Encoding and Hierarchical Clustering of Online Comments
As competition in the new-energy-vehicle (NEV) market increasingly shifts from technical specifications to holistic user experience, understanding user-generated feedback has become essential for product optimisation. This report investigates NEV user experience across 12 dimensions, including powertrain performance, handling, space and comfort, intelligent driving, smart cabin functions, charging convenience, and after-sales service, with the aim of identifying distinct user segments from online reviews. Based on a sample of 500 user reviews, the study applies a text-mining pipeline that combines rule-based term standardisation, LLM-based summarisation and classification, semantic encoding with Sentence-BERT, and hierarchical clustering. This workflow converts large-scale, unstructured comments into structured experience features. It also enables the comparison of users according to the similarity of their expressed concerns and evaluations. The analysis identifies five user clusters with clearly differentiated priorities: smart-cabin enthusiasts, comfort-driven users, pragmatic technology users, range-anxious drivers, a space-oriented majority. Research indicates significant diversity among NEV consumers regarding their priorities – in more details, that is intelligent capabilities, seat comfort, driving range, charging accessibility, and powertrain dependability. These results provide data-driven support for targeted product development, user-segment-specific marketing, and strategic expectation management, while also demonstrating the value of text mining for extracting actionable insights from NEV user reviews.
Siyun Zhang
Open Access
Article
Conference Proceedings
Consistency of AI-Driven Usability Evaluations for Software: Learning Management Systems as an Example
The evaluation of usability is a critical, yet resource-intensive, component of Human Systems Integration (HSI) for complex software systems like Learning Management Systems (LMSs). This study investigates the potential of using Large Language Model (LLM)-based Artificial Intelligence (AI) systems to conduct automated usability evaluations by analyzing textual user feedback. The author compared the consistency of usability assessments for 100 LMSs generated by three leading AI systems (Claude-Opus-5, GPT-5.6-Sol, and Gemini-3.5-Flash) against each other and against a baseline of human user aggregate scores. The AI systems scored seven usability attributes based on ISO/IEC 25010, 9241 Standards and usability models proposed in the literature so far: effectiveness, efficiency, error protection, accessibility, operability, learnability, and satisfaction. Principal Component Analysis (PCA) was used to derive a single usability factor score for each AI system’s evaluations. Subsequent correlation analyses (Pearson, Spearman, Kendall’s tau-b) revealed strong, statistically significant positive correlations between the AI-generated factor scores and the human benchmark scores, with correlation coefficients ranging from 0.459 to 0.626. Notably, the inter-AI correlations were exceptionally high, with values as strong as 0.850 between Claude-Opus-5 and Gemini-3.5-Flash. The findings suggest that LLMs possess a significant capability to perform human-like usability evaluations (or at least humans’ aggregate product rating) from text, supporting their potential as a powerful tool for rapid, scalable, and cost-effective usability analysis in HSI and systems engineering. However, the moderate strength of the correlations with human benchmark scores also indicates that current AI systems should be seen as complementary to, rather than a complete replacement for, traditional human-centered evaluation methods.
Victor K Y Chan
Open Access
Article
Conference Proceedings
Human–AI Collaboration in Supply Chains: A Framework for Resilient Decision Making Under Disruption
Supply chain inventory management under disruption is increasingly complex, requiring planners to integrate fluctuating demand, delivery delays, and inventory shortages across multiple periods. Standard evaluations of AI decision support systems evaluate accuracy within single-trial settings, overlooking how algorithmic errors propagate into persistent, multi-period operational disruptions. To solve these issues, this article introduces a humancentered AI decision-support framework and inventory simulation system designed to aid planners during dynamic disruptions. Built on an OR-Gym single-echelon formulation and calibrated using three real-world datasets—Walmart M5 for demand, Olist for lead times, and DataCo for disruption frequencies—the system supports realistic stress-testing and state tracking. The system supports decision workflows by pairing automated order recommendations with real-time stockout risk predictions from a calibrated LightGBM classifier across three visual bands (Low, Moderate, High). These functions are combined into a two-stage ordering interface where planners lock in an initial independent order before viewing AI recommendations, preserving human decision authority and minimizing cognitive bias. A usability and human performance evaluation was conducted through a 30-week pilot demonstration to assess advice adoption, operational costs, and cognitive workload (NASATLX). With high classification performance and results showing that propagating AI errors cause 3.45 times more extra stockout weeks than absorbed errors, the platform demonstrated excellent usability and behavioral engagement. The findings demonstrate that the proposed decision-support framework effectively facilitates cognitive ergonomics, appropriate reliance, and human–AI collaboration in supply chain inventory management, underscoring the promise of human-centered AI systems to improve managerial performance and operational resilience.
Md Asfaqur Rahman, Shakila Azad, Numair Bin Sharif, MD Hasibur Rahman Murad, Razia Sultana, Mohammed Tanjim Uddin Chowdhury, Md Shahadat Hossain
Open Access
Article
Conference Proceedings
Development of an AI based PHM system for instrumentation and control (I&C) systems of Korea’s APR1400 NPPs
This paper presents an artificial intelligence (AI)-based prognostics and health management (PHM) framework for estimating the remaining useful life (RUL) of control cards and electronic components in the instrumentation and control (I&C) systems of Korea's APR1400 nuclear power plants (NPPs). The framework integrates a reliability and failure database, AI-based RUL prediction models, and a hierarchical maintenance interface. As a representative application, 40 photocouplers were aged at 130 °C for approximately 90 days. The accelerated-aging data were converted to plant-relevant temperature conditions using the Arrhenius relationship, and a long short-term memory (LSTM) network incorporating Monte Carlo (MC) dropout and physics-informed regularization was developed to estimate RUL and predictive uncertainty. Using operating time, temperature, cumulative temperature exposure, and minimum and maximum temperatures as inputs, the proposed model achieved a mean absolute error of 4.29 days and a reported prediction accuracy of 97.77% on the test dataset. The prototype provides component-, control-card-, and cabinet-level condition and RUL information together with abnormal-state alarms and trend graphs. The proposed framework provides a basis for condition-based and predictive maintenance of nuclear I&C equipment by combining data-driven prognosis with system-level visualization.
Chang Hwoi Kim, Sangjun Park
Open Access
Article
Conference Proceedings
Acceptance of Virtual Reality Among Older Adults: An Exploratory Study of a Culturally Embedded VR Application
The use of virtual reality (VR) applications among older adults has received limited attention in prior research and has been investigated mainly within clinical contexts. The aim of the present study was therefore to investigate the acceptance of a VR application and potential influencing factors within a sample of older participants.The sample comprised N = 119 participants aged between 55 and 89 years. In addition to demographic variables (age, gender), perceived presence, physical well-being, technology affinity, and personal relevance were assessed. The VR application immersively recreated a historically significant hotel from the region, selected for its potential to establish regional and biographical relevance.Descriptive results indicated a consistently high level of acceptance and positive user experience. Qualitative feedback from participants emphasized the realistic representation, visual clarity, interactive elements, and exploratory nature of the application. Inferential statistical analyses (ANOVAs, Mann-Whitney-U-Test and correlations) revealed no significant differences between age groups or genders and no significant correlations with the psychological variables assessed. No significant differences or correlations were detected; given the exploratory design and uneven subgroup sizes, these findings should be interpreted cautiously. Drawing on theories of cultural memory and embodied experience, the results suggest that VR enables users to engage with historical environments as shared experiential spaces. Overall, the study contributes to the emerging body of research on VR and older adults by highlighting the potential of immersive technologies to facilitate access to cultural memory and enable meaningful engagement with historical contexts that have not been personally experienced.
Caroline Schon, Vanessa Fehlig, Rong Huang, Tim Wachsmuth
Open Access
Article
Conference Proceedings
Evaluating Nasal Fit Interfaces in XR Headsets Using Anthropometric Clustering Techniques
Achieving secure, comfortable, and properly registered fit remains a persistent challenge in head-mounted displays (HMDs), particularly at the nasal interface where variation in nose shape drives discomfort, slippage, and positional variance. This study presents a reproducible, data-driven workflow for deriving typified nose pad geometries from large-scale (~1700) anthropometric scans. An analysis of predicted nasal landings compared to anatomical landmarks indicated differences across ethnicity supporting the need for multiple interface geometries. To generate representative shapes, we aligned head scans in a novel Nasal Ridge coordinate system, clustered points along the z-axis, and applied Locally Estimated Scatterplot Smoothing (LOESS) to create 5th, 50th, and 95th percentile surfaces. Comparative tests with alternative meshing and curve-fitting methods showed LOESS produced the most stable and design-ready geometries. This framework demonstrates how anthropometry-based modeling can directly inform inclusive and efficient HMD nose pad design.
Thomas Schnieders, Karen Bredenkamp, Paul Foster, Brian Heckman
Open Access
Article
Conference Proceedings
Digital Personas in Immersive Learning: How Avatar-Mediated Interaction Shapes Competency Development in Higher Education: A Systematic Review
How do digital personas and avatar embodiment influence learning and behavior in metaverse environments? We systematically reviewed 184 studies on immersive technologies in higher education (2018 - 2025), focusing on the socio-behavioral mechanisms through which avatar-mediated interaction shapes competency development. Our findings are sobering: while the Proteus effect and social presence enhance short-term engagement, fewer than 25 studies employed experimental designs and no longitudinal evidence exists. We propose a causal mechanism chain from immersion to real-world skill transfer and identify specific break-points where the chain fails. Digital persona design emerges as a key moderating variable that has been conceptually recognized but barely investigated empirically. The integrative framework we develop positions ethical considerations as cross-cutting rather than peripheral and generates testable propositions for future research. We discuss implications for metaverse platform design and institutional adoption strategy.
Safa Omri, Florian Urmetzer, Jens Neuhüttler
Open Access
Article
Conference Proceedings
How AR guidance shapes behavioral change in walking: A stage-based analysis of gaze and trajectories
This study models behavioral change during real-world walking as a process rather than only an outcome. We propose three stages—Active Observation, Pre-action Behavior, and Active Behavior—quantified from gaze and walking trajectories. In a university corridor, 34 participants completed two sessions under either an augmented-reality (AR) guidance condition or a control condition; in the AR condition, a virtual object demonstrated the target action in real space. Successful execution was associated with a characteristic movement tendency toward the target. AR did not directly induce the target action, but was associated with earlier target recognition and reduced individual variability in exploration depth and trajectory straightness. Repeated exposure was also associated with changes in these characteristics, suggesting that AR information may facilitate progression from pre-action to active behavior.
Shugo Matsui, Tatsuya Shibata
Open Access
Article
Conference Proceedings
Development, Comprehension Testing, and Preliminary Evaluation of Child-Adapted Measures of Visually Induced Motion Sickness for Virtual Reality
Visually induced motion sickness (VIMS) presents a salient challenge to the use of virtual reality (VR) technologies in children, and its assessment poses unique challenges for researchers. While previous work has created and employed assessment tools designed to target specific VIMS symptomology in adults, (1) the language contained in these tools may not be appropriate for comprehension across all ages, and (2) previous attempts to adapt these assessments for use in children have also fundamentally modified the original tools. In the present work, we propose updated versions of two established tools for VIMS assessment that preserve the original assessments’ questions and rating scales: the MISC-C, adapted from Bos et al.’s (2005) Misery Scale, and the SSQ-C, adapted from Kennedy et al.’s (1993) Simulator Sickness Questionnaire. In Phase 1, we modified the original language of these assessments to be appropriate for children aged 7–12 through consensus within a panel of experts in developmental and cognitive psychology. In Phase 2, we explored children’s comprehension of the initial MISC-C and SSQ-C versions within this age range and applied corresponding updates informed by these insights. In Phase 3, we conducted a preliminary evaluation of the resulting MISC-C and SSQ-C in a VR gaming use case to explore their initial application in this context, rather than to establish their broader validity or performance. We discuss initial observations regarding the use of the MISC-C and SSQ-C in children aged 7–12 and establish a foundation for future validation of developmentally appropriate VIMS assessment tools for child VR research and practice.
Nichole Breeland, Stephen Walenchok, Rachel Kelly
Open Access
Article
Conference Proceedings
Detection-Aware Generation of Fatigue-Conditioned Ocular and Head Behaviors for Virtual Drivers
Driver fatigue induces systematic changes in observable behaviors, which should be represented in virtual drivers beyond predefined animations. This study presents a data-driven framework for generating stochastic ocular and head behaviors conditioned on subjective sleepiness. Empirical behaviors were obtained from a controlled driving experiment and characterized across Alert, Intermediate, and Drowsy states based on the Karolinska Sleepiness Scale (KSS). Significant fatigue-dependent changes were observed in eyelid-closure characteristics, PERCLOS, and head-pitch behavior, and were incorporated into probabilistic behavioral models. A detection-aware generation method was developed to synthesize eyelid trajectories while accounting for discrepancies between intended behavior and behavior re-measured after detection. Monte Carlo validation showed that generated PERCLOS distributions reproduced empirical distributions across all KSS states, with normalized Wasserstein distances of 0.14–0.24 and no significant differences detected by Kolmogorov–Smirnov tests. These results provide quantitative evidence that the proposed generator reproduces fatigue-dependent characteristics and population-level variability observed in real drivers.
Woojin Kim, Seung-Jun Lee, Yonghyun Kim, Myeongkyu Lee, Brandon J. Pitts
Open Access
Article
Conference Proceedings
Design and Preliminary Evaluation of a Generative AI Role-Play System for Practicing Cognitive Empathy
When another person's psychological state is not fully verbalized, understanding that person requires active questioning and interpretation. We designed a generative AI role-play system in which the AI portrays a junior colleague who does not readily disclose their feelings. The system combines a predefined persona, rules governing psychological information disclosure, and three dialogue tasks: exploration, interpreting, and paraphrasing. In a preliminary evaluation with 16 participants, illustrative dialogues from two conditions showed both task-aligned communication and instances in which advice or directive statements replaced the intended dialogue behaviors. The mean change in cognitive empathy (CE) was +0.438 in the cognitive-empathy condition and +0.094 in the factual–situational condition. However, baseline scores differed between conditions, and the baseline-adjusted analysis did not show a difference in the same direction. This study presents a concrete design for practicing perspective-taking through dialogue and identifies response-contingent support for users' dialogue behaviors as a direction for future system development.
Tsubasa Kishiume, Taiyo Kojima, Toshikazu Kato, Toru Nakata, Takashi Sakamoto
Open Access
Article
Conference Proceedings
Improving accessible mobility for visually impaired travelers: An IPA analysis of the sightseeing verbal map in Shanghai
This study conducts an Importance-Performance Analysis (IPA) of the Sightseeing Verbal Map, an auditory assistive system, to guide accessible-mobility improvements for visually impaired travellers in Shanghai. Building on prior qualitative work that adapted the system from Japan to Chinese users, we move from descriptive validation to quantitative prioritization. A structured questionnaire was administered to 306 visually impaired respondents, who rated nine attributes across three functions (route guidance, the map of words, and environmental sounds) on both importance and satisfaction. Plotting these means against grand-mean crosshairs yields a four-quadrant priority matrix. Real-time location awareness emerges as the clearest improvement target, with the narrow-sense map of words as a borderline case in the same quadrant; start- and end-of-route guidance and walking instructions are strengths to maintain, and the soundscape is already adequately provided. The findings translate diffuse feedback into a data-driven improvement map and advance cross-cultural localization of assistive mobility technologies.
Tianyu Li, Takashi Uchida
Open Access
Article
Conference Proceedings
Do Close-Range Drone Operations Affect Workers’ Emotional States and Stress Levels? A Mixed-Reality Study
Unmanned aerial vehicles (UAVs, or drones) are increasingly deployed on construction sites for tasks such as material delivery, progress monitoring, and safety surveillance. In many applications, drones may operate near workers in shared workspaces, raising safety concerns among researchers and industry practitioners. Specifically, drones operating near workers can be perceived as a potential threat or a psychological stressor, which may elicit changes in workers’ emotional states and physiological stress, potentially impairing health and well-being. However, there is a lack of empirical evidence on how drone presence influences construction workers’ emotional states and stress levels, and whether these effects vary across proximity levels. To address this gap, this study employed a mixed-reality technique to simulate a roofing task, in which participants installed shingles on a roof while a drone approached and hovered at two distances (near: 0.45 m; far: 1.2 m). Emotional states were assessed using the Positive and Negative Affect Schedule (PANAS), and stress levels were quantified using electrodermal response (EDR). Thirty-one participants completed the experiment. Results indicate that positive affect remained relatively stable across conditions, whereas negative affect increased significantly when the drone presence in proximity. EDR-based stress levels were highest in the near-drone condition, while the far-drone condition was comparable to baseline. These findings suggest that close drone operation can impose measurable emotional and physiological costs and highlight that proximity management can be a practical strategy for safer and more effective human–drone collaboration on jobsites.
Yibo Qiang, Sébastien Hélie, Jeffrey M Haddad, Sogand Hasanzadeh, Behzad Esmaeili
Open Access
Article
Conference Proceedings
Design of Graphical User Interfaces for Validation of Steering-Wheel HMI Solutions
The increasing integration of digital functionalities, Advanced Driver Assistance Systems (ADAS), and autonomous driving technologies is transforming the steering wheel into a more complex and demanding Human–Machine Interface (HMI), requiring new interaction technologies capable of balancing usability, safety, and efficiency. Within the GreenAuto research project, this work contributes to the development of an innovative steering-wheel HMI aimed at enabling intuitive, safe, and efficient driver–vehicle interaction in conventional and autonomous driving scenarios.This paper presents the graphical user interfaces (GUIs) developed for an automotive driving simulator used to evaluate and compare different steering-wheel HMI solutions. The simulator integrates conventional vehicle interfaces, including the instrument cluster, head-up display, and central stack, while introducing an innovative steering wheel equipped with two integrated mini-displays and new interaction controls. The proposed GUIs define the context and visual interaction between the steering-wheel controls and integrated displays, including mini-displays and the head-up display, providing support for infotainment, communication, navigation, and ADAS functionalities.The design process combined ergonomic constraints, information architecture, and user-centered interface design principles to define interaction flows, navigation structures, and visual feedback mechanisms. A reachability assessment was conducted to establish the dimensions and positioning of the display areas, ensuring accessibility without compromising driving ergonomics. The resulting interfaces provide a consistent interaction framework for implementing and evaluating different steering-wheel HMI solutions within an automotive driving simulator equipped to collect behavioral, subjective, and physiological data during experimental studies.The proposed GUIs establish a reusable research platform for the experimental evaluation of steering-wheel HMIs, supporting future studies on usability, driver interaction, cognitive workload, and interface design.
Micael Gonçalves, Lídia Lemos Ribeiro, Hugo Rodrigues, Marc Schledorn, Nelson Costa
Open Access
Article
Conference Proceedings
Intelligent Design Architectures for Human-AI Teaming in Safety-Critical Operations
Intelligent systems increasingly influence operational decisions in transportation, yet their integration often occurs without considering the cognitive characteristics of the human operator. This paper proposes an intelligent design architecture that combines AI-enabled analytics, digital twins, competency modelling and systems engineering within a human-centred framework. Rather than optimising algorithmic performance alone, the proposed architecture dynamically adapts information presentation, decision support and workload distribution according to human performance requirements. Aviation case studies illustrate how computational intelligence can enhance resilience, explainability and operational adaptability while maintaining meaningful human authority.
Debra Henneberry, Dimitrios Ziakkas, Thanasis Binis
Open Access
Article
Conference Proceedings
Sharing the Edge: Multiple Concurrent Neural Workloads on a Single Microcontroller in a Smart HVAC Vent
Edge devices increasingly run multiple inference workloads at once, but a microcontroller offers no free lunch for multitasking: every model competes for the same memory, cycles, and physical environment, including the space the user speaks into. This paper examines that competition through the second prototype of ZenSys, a smart-vent system sharing one ESP32-S3 across a wake-word detector, a keyword spotter, and continuous thermal-occupancy inference. On 2,000 labeled thermal frames, an adaptive background-subtraction blob detector outperformed a compact CNN (F1 0.92 vs. 0.80) at a fraction of the memory and latency, a result we attribute to a single-room corpus rather than to any inherent weakness of learned inference. Small early voice trials point to a similar inversion: shorter commands held up better under noise, wake-word detection appeared weaker for female speakers, and actuator vibration shrank usable voice range to 3–10 cm whenever the fan ran. For the occupant, the binding constraint was mechanical rather than computational.
Yousef Abdou, Hazem Elshorbagy, Alaa Abdou
Open Access
Article
Conference Proceedings
Metacognitive Scaffolding for Novice Designers in Generative AI-Assisted Poster Design: A Formative Study of Premature Convergence
Generative artificial intelligence makes poster creation accessible to novice designers, but polished initial outputs may encourage premature convergence. This study develops an early poster-design prototype that embeds planning, monitoring, and evaluation scaffolds within a structured workflow and examines its use formatively. Thirteen participants without systematic design training completed a fixed A-to-B within-participant comparison: freely combining AI tools in condition A, followed by using the prototype in condition B. Human coding of recordings informed fixation-related process measures; intent-alignment ratings, an experience questionnaire, and interview summaries provided complementary evidence. Mean fixation-duration proportion was 34.9% in A and 18.3% in B, a difference of -16.6 percentage points (bootstrap 95% CI [-24.9, -8.1]; exact p = .005). Conceptual-frame counts increased by 1.154 on average and retained a statistical signal after Holm adjustment. Fixation-event rate did not differ significantly, and intent alignment did not retain a statistical signal after adjustment. All four experience dimensions were rated more positively in B. These observations identify broader exploration and a lower proportion of effective creation time coded as fixation-related within this sample. The small sample, fixed order, unequal creation times, bundled components, and measurement limitations restrict generalization and causal interpretation. The contribution is a concrete metacognitive scaffolding design and formative evidence that identifies priorities for controlled evaluation.
Shuai Sun, Yu Yangwenqing
Open Access
Article
Conference Proceedings
Human-Object Interaction Spatial Constraint-Driven Intelligent Generative Design for Products
The convergence of ergonomics, Kansei engineering and generative artificial intelligence is reshaping the paradigm of intelligent product design. Conventional generative design has largely been driven by geometric optimisation, yet it struggles to accommodate the emotional experience and spatial usability that arise when humans interact with products. This review examines how Kansei engineering, which quantifies users' affective responses, has been integrated with AIGC-enabled generative models to produce designs that satisfy ergonomic spatial constraints while resonating emotionally. Three dimensions are analysed: the quantification of Kansei needs, the technical pathways of AIGC-driven generation, and the embedding strategies of spatial constraints. The review indicates that the field is shifting from geometry-driven generation towards human-centred constraint-driven generation. Automotive interiors and seating systems emerge as the most extensively studied domains, where methods such as FAHP-FQFD, digital human modelling and diffusion-based generation have been most rigorously developed. Nevertheless, the alignment between user Kansei and producer design intent remains difficult to quantify, dynamic interaction is largely disconnected from static spatial constraints, and standardised evaluation benchmarks are absent. The paper concludes by outlining future directions, including mixed-initiative interaction, physiologically informed generative loops, and physically intelligent models capable of reasoning about posture, motion and contact pressure.
Yaqi Li, Hao Shi
Open Access
Article
Conference Proceedings
An Intelligent Scheduling Capability for ISR Collection Assignment
Intelligence, Surveillance, and Reconnaissance (ISR) operations require rapidly assigning prioritized collection requests to available assets while adapting to changing operational conditions. The increasing volume of collection requests and the need for asset coordination make automated scheduling essential for effective ISR operations. By automating the assignment and dynamic reallocation of collection requests, such systems can reduce planner workload while also enabling autonomous mission management capabilities. This paper focuses on the specific problem of assigning prioritized ISR collection requests to planned air missions, where the objective is to maximize the number of assigned collections while favoring higher-priority requests. This ISR assignment problem differs from standard scheduling problems due to domain-specific collection feasibility constraints and the need for rapid rescheduling in response to evolving threats, route adjustments, and weather changes. The primary contribution of this work is the design and evaluation of a domain-specific scheduling capability for ISR collection assignment that addresses collection feasibility constraints and the dynamic replanning requirements. The paper describes the system architecture and scheduling methods and presents results comparing standard (baseline) and bottleneck scheduling approaches across both static and dynamic scenarios. The resulting capability can both support human planners through decision support and operate as an autonomous scheduling component.
Jeremy Ludwig, Annaka Kalton, Forest Finnigan, Katelyn Clements, Jerred Holt
Open Access
Article
Conference Proceedings
Behavior-Aware Arbitration for Scalable Robot Swarms in Cooperative Search-and-Retrieval Tasks
Adding agents to a team can improve search speed and task coverage, but larger teams can also increase coordination load, resource contention, and task interference. This paper studies robot-only swarm behavior as a step toward arbitration support for human-autonomy teaming. We developed a maze-based simulation in which agent teams collect keys, open doors, respond to coordination requests, and attempt to reach a shared objective. Agent behaviors include normal task execution as well as error-prone conditions that produce confused key handling, request overcommitment, task abandonment, or adversarial interference. Structured event logs were parsed into run-level, agent-level, request-level, and decision-level summaries. Experiments across multiple map sizes and team sizes show nonlinear performance scaling. Small maps were nearly always solved across team sizes, while larger maps were more often solved by mid-sized teams and showed degraded performance for larger swarm sizes. Request expirations and timeout-related events increased in higher-load conditions, suggesting that coordination overhead can reduce the value of additional agents. We also trained a contextual candidate-action arbitration model that converts event-stream features into supervisory recommendations, including door-progress prioritization, timeout-risk reassignment, continued exploration, key escort, and reduced reliance on agents associated with disruptive behaviors. The model is intended to centralize swarm activity and highlight when task progress, reassignment, or additional supervision may be needed. These findings support behavior-aware arbitration as a decision-support layer for identifying when larger autonomous teams are improving task progress and when they are increasing coordination burden.
Bradley Killen, Daniel Carruth, Cindy Bethel, Kenna Henkel, Victor Paul
Open Access
Article
Conference Proceedings
A High-Availability Multi-Agent RAG Architecture for Automated Expert Assistance and Multimodal Technical Onboarding
This paper presents the design, deployment and evaluation of a high-availability, multi-agent Retrieval-Augmented Generation (RAG) architecture acting as an expert assistant for technical training. In sectors such as cyber-physical systems and automated instrumentation, onboarding depends on synchronous supervision by senior engineers, which creates bottlenecks and increases technical downtime. Retrieval is restricted to a closed, curated corpus of manuals, datasheets and approved guidelines, keeping answers traceable and reducing large language model hallucinations. The interaction layer runs on WhatsApp through the Evolution API and accepts text, documents, circuit images and voice notes; the backend combines n8n orchestration, a dedicated virtual private server and a Supabase PostgreSQL instance with the pgvector extension, with inference served by paid-tier models through Google Vertex AI with provider fallback. Retrieval scope is bound to the caller’s resolved identity rather than chosen by the model, enforcing need-to-know separation, and the interaction policy is configurable by user profile: a Socratic mode that withholds finished solutions, and a direct mode for experienced practitioners. An adoption survey administered to final-semester engineering students at the end of an intensive training cycle (n = 24) obtained a global Cronbach’s alpha of 0.90. Perceived competence (M = 4.38, d = 1.93) and autonomy (M = 4.14, d = 1.39) were significantly above the scale midpoint, whereas relatedness was not (M = 3.24, p = 0.125): the assistant substitutes for the informational function of mentorship but not for its social function. Industrial deployment remains a projected application.
Javier Izquierdo-Reyes, Miguel de Jesús Ramírez-Cadena, Olivia Calderón-Valderrama
Open Access
Article
Conference Proceedings
Cardiac and Sleep-Stage Responses to Rhythmic Auditory and Amplitude-Modulated Vibrotactile Pillow Stimulation During Daytime Naps
Rhythmic pillow stimulation combining mechanical vibration and sound may influence cardiac activity and sleep progression. We compared Control, heartbeat-like Tapping with sound and vibration at 45 beats per minute, and Vibrotactile stimulation produced by amplitude-modulating a 50-Hz pillow vibration at 0.25 Hz in a within-subject daytime-nap experiment. Twenty-three participants completed all conditions. Electrocardiography and polysomnography were analyzed over the first 55 min without sleep-duration exclusions. Mean heart rate was evaluated by one-way repeated-measures ANOVA and Holm-adjusted paired comparisons; sleep-stage proportions and latencies by Friedman tests. Mean heart rate was 60.69 bpm in Control, 59.94 bpm during Tapping, and 58.01 bpm during Vibrotactile stimulation. The condition effect was nonsignificant (F(2,44) = 2.08, p = 0.138). Vibrotactile stimulation was 2.69 bpm lower than Control in the unadjusted paired comparison (95% CI -5.36 to -0.01, p = 0.049, Cohen's dz = -0.43), but not after Holm correction (p = 0.147). Sleep-onset latency was 7.57, 8.54, and 6.61 min, and N3 proportion was 18.62%, 15.10%, and 21.15%, respectively. N2 was reached by 22, 20, and 23 participants and N3 by 18, 15, and 18. Among participants reaching each stage in all three conditions, N2 latency was 13.68, 12.76, and 12.61 min (n = 19), and N3 latency was 20.96, 20.17, and 18.04 min (n = 12); neither condition test was significant. Vibrotactile stimulation showed a descriptive pattern of lower heart rate, more N3 sleep, and earlier N3 occurrence, whereas Tapping did not shorten sleep onset. Exploratory findings require confirmation in prespecified, adequately powered studies.
Kazutaka Yamamoto, Keiichi Obara, Chinami Taki, Kohei Kishimoto, Shima Okada
Open Access
Article
Conference Proceedings
Unsupervised Clustering of Repetitive Task Trials via Tucker Decomposition of Joint-Angle Data Tensors
Movement patterns can vary across repetitions of the same task. This study aims to develop an analysis framework for comparing repeated movement trials in a shared feature space and describing the occurrence of movement patterns across repetitions. Time-series joint angle data was obtained from motion capture data recorded during a seated repetitive lifting task performed by twelve males. Partial Tucker decomposition provided temporal and joint-DOF bases shared across trials within each participant, while retaining trial-specific core coefficients. Gaussian mixture models were applied to these coefficients to group trials with similar kinematic features. Four to seven clusters were selected for each participant. In all participants, the dominant cluster changed in stages as the experiment progressed, although trial order was not used as a feature. Clusters dominant in early trials rarely regained dominance later, while later trials often showed repeated alternations in cluster membership probabilities. These results show that the framework can describe the occurrence and progression of movement patterns during repetitive work.
Akisue Kuramoto, Masaya Noguchi, Kosuke GOSHIMA, Daiki KOGA, Kazuki HIRANAI, Motomu Nakashima
Open Access
Article
Conference Proceedings
A Multitask Evolutionary Framework for Procedural Content Generation in Games
Procedural Content Generation (PCG) has emerged as a widely adopted technique used for creating content in various types of games. In recent years, numerous approaches have been proposed to address different PCG problems. Despite the diversity in their concept, most existing generators are designed for specific games, even though many games share structural similarities. In light of this reason, this paper investigates the application of Evolutionary Multitask Optimization (EMTO) for PCG. A Multifactorial Evolutionary Algorithm (MFEA)-based framework is introduced to systematically group PCG problems according to the compatibility of their search spaces, enabling multiple problems to be optimized simultaneously. To evaluate the effectiveness of the proposed framework, experiments were conducted on a range of PCG benchmark problems, along with comparisons to a single-task evolutionary approach. The results highlight both the potential and the challenges of applying EMTO to PCG, particularly with respect to task compatibility.
Hoang Long Nguyen, Taneli Nyyssönen
Open Access
Article
Conference Proceedings
The Evolution of ADAS Safety: A Human Factors Perspective
Safety discussions regarding Advanced Driver Assistance Systems (ADAS) have continually evolved as technological capabilities, market penetration, and empirical evidence have advanced. Human factors research on driver assistance technologies has included work on potential unintended consequences, including false alarms, driver annoyance, startle effects, diminished vigilance, behavioral adaptation, and overreliance. As systems have moved from concept to deployment, researchers have evaluated whether such concerns have materialized in real-world use conditions and whether ADAS deliver measurable safety benefits. This paper reviews the evolution of ADAS research using Forward Collision Warning (FCW) systems as a continually evolving example of ADAS features, and examines evidence supporting benefits from Level 1 and Level 2 ADAS. Prior studies have emphasized acceptance, warning design, and trust, as well as potential improvements in hazard perception and response and reductions in rear-end crash rates. Initial concerns that Level 1 and Level 2 ADAS technologies might reduce driver engagement and encourage secondary task use continue to be examined as technology proliferates. Evidence suggests that drivers generally adapt their behavior in response to system capabilities and driving demands, and maintain engagement with the driving task as needed while benefiting from reduced workload. Emerging crash-based evidence also suggests potential safety benefits associated with ADAS, though continued research is needed due to limitations related to data availability, market penetration, and uncertainty regarding system engagement during crashes. Collectively, literature demonstrates how ADAS safety cases have matured over time, evolving from hypothetical concerns and projected benefits to evidence informed by large-scale deployment and real-world performance.
Iiona Scully, Abby Swenson, Stephanie Reeves, David Cades
Open Access
Article
Conference Proceedings
Digital Fitting for Resource Optimization: Hybrid Practices for Sustainable Accessory Design
Digital fitting technologies have transformed fashion retail through virtual visualisation and consumer engagement, yet their potential in early-stage accessory product development remains underexplored. Accessories involve complex materials, specialised manufacturing processes and knowledge networks requiring collaboration between human expertise and digital systems. This study investigates digital fitting as a human-centred knowledge infrastructure supporting collaborative design and decision-making. The research integrates Service-Dominant Logic (S-D Logic), Service Science (SS) and the Viable Systems Approach (VSA), viewing value as emerging from the integration of resources, competences and knowledge across interconnected actors. The paper develops a conceptual framework through ecosystem mapping, relational knowledge modelling, practice-informed expert reflection, and scenario-based discussion. Ecosystem mapping identifies the relationships among designers, manufacturers, suppliers, and technical specialists, while relational knowledge modelling integrates material properties, manufacturing constraints, and production information within a Digital Twin perspective. Practice-informed reflections and representative design scenarios are used to examine recurring coordination challenges and discuss the conceptual applicability of collaborative digital fitting throughout accessory development, thereby strengthening methodological transparency, analytical coherence, and conceptual traceability. The findings indicate that the main obstacle to accessory innovation is knowledge fragmentation. Material intelligence, manufacturing expertise and technical constraints often become available only after physical prototyping, increasing iterations and uncertainty. The proposed framework reinterprets the Digital Twin as a relational socio-technical environment where digital fitting enhances human judgement through shared knowledge and earlier informed decisions. By strengthening collaboration and reducing uncertainty before production, the framework supports resource-efficient accessory development and contributes to Sustainable Development Goal 12 on Responsible Consumption and Production.
Tommaso Roviello, Benedetta Russo, Clara Bassano, Paolo Piciocchi
Open Access
Article
Conference Proceedings
Designing and Implementing a Total Pilot Performance Course for Undergraduate Aviation Students
Collegiate flight students are presented with a unique set of circumstances and challenges during their training. They must negotiate balancing their academic courseload while also completing several pilot certifications, which take hundreds of hours of flying to accomplish. In addition, students are often away from home for the first time and are exploring extracurricular and employment options, while also navigating new social settings. These competing obligations can create a high-stress and fatiguing environment for flight students. Ongoing studies have been evaluating the physical, cognitive, and emotional impacts on collegiate pilots (Henneberry et al., 2026). Metrics being observed before, during, and after training flights include heart rate, heart rate variability (HRV), electrodermal activity (EDA), electrocardiography (ECG), skin temperature, and psychomotor vigilance testing (PVT). Participants in the study have been provided with their results, which can be used to inform their preparation before a flight lesson. Further, this information may provide insight into health implications that can be used throughout the duration of one’s aviation career. The results of these studies are being used to design an experimental undergraduate course within a professional flight training program. Topics to be covered in the class include pilot readiness, sustainable health habits, career longevity, wearable technologies, and physiological self-monitoring. Students will be asked to utilize a wearable device throughout the semester and monitor their health and readiness both on the ground and in the air. Learning outcomes include the development of individual and team-based plans that support total pilot performance in collegiate aviation settings.
Debra Henneberry, Julius Keller
Open Access
Article
Conference Proceedings
Development of an AI Agent for Failure Monitoring, Diagnosis, and Remaining Useful Life Prediction of Capacitors in Nuclear Safety System Signal Processors
This study proposes an AI agent that performs failure monitoring, diagnosis, and Remaining Useful Life (RUL) prediction for the electrolytic capacitors of nuclear power plants (NPPs) safety system signal processors and explains the situation and evidence to operators. An unscented Kalman filter (UKF) applied to estimate the thermal aging degradation state and detects abnormal states from the residuals of its estimates. A Large Language Model (LLM) converts each event into diagnostic information classified by severity, cause, evidence, and recommended action. For verification, the artificial dataset generated from actual measured data of accelerated thermal aging test. The artificial date set created by applying the random error to measured value and the intentional fluctuation is used for abnormal stat. The normal state produced no false alarm over eight weeks, whereas the abnormal state at week 9 was detected immediately. A two week bias signal and a step shaped degradation were discriminated as an anomalous state and a sustained state change, using only the persistence of the residual. Twenty eight of the failed specimens were detected at 9 week, and the RUL prediction error converged to 3.4 weeks. In another test cases, the AI agent alarmed 46 cycles ahead of end of life and temporarily fluctuation was not reflected the RUL as it is not affect the degression factor of an thermal aging. These results suggest that providing operators with verifiable monitoring and diagnosis explanations, together with the limits of the monitoring system, can improve the operating reliability of NPP safety systems.
Sangjun Park, Chang Hwoi Kim
Open Access
Article
Conference Proceedings
From Emerging Technologies to Operational Capability: Human-Centred Transition Pathways for AI Implementation
Emerging technologies frequently demonstrate technical maturity before organisations develop equivalent operational capability. This paper presents a structured implementation methodology integrating digital twins, AI analytics, simulation, competency-based training and Human-AI Teaming within a unified transition pathway. Rather than evaluating technologies individually, the framework examines how emerging capabilities interact with organisational processes, workforce competencies and safety management. Aviation provides practical examples of AI-supported implementation, while the proposed methodology is transferable to multiple safety-critical sectors.
Debra Henneberry, Dimitrios Ziakkas, Ian Roth
Open Access
Article
Conference Proceedings
From Success Factors to Structured Judgement: A Preliminary Web-Based Decision Support System for Early-Stage Real Estate Readiness Assessment
Early-stage real estate development decisions commit substantial capital and define project parameters before objective performance data are available, yet they remain largely intuitive and unstructured in practice. This paper proposes the Real Estate Initiation Support System (REISS), a preliminary conceptual web-based Decision Support System (DSS) designed to support structured readiness assessment at the pre-commitment stage of commercial real estate development. Building on a preliminary framework of Critical Success Factors (CSFs) and Key Competitiveness Indicators (KCIs) established in the first phase of this research, REISS integrates both frameworks into a three-module evaluation architecture targeting CSF readiness, organisational performance, and individual capability. The Analytic Hierarchy Process (AHP) is employed to derive expert-based weights for each factor, which are combined with user self-assessment scores to generate readiness scores and ranked recommendations. The system is delivered through a three-tier web-based architecture comprising a wizard-style user interface, a model management system, and a database management system. The paper presents the conceptual design of REISS and discusses its potential contribution to more consistent and evidence-informed early-stage decision-making in the UAE real estate market. Empirical validation, AHP weight derivation, and prototype development are planned for subsequent research phases.
Alaa Abdou, Mohamed Zaghloul, Sahar Makky, Yousef Abdou
Open Access
Article
Conference Proceedings
Architecture and Early Software Validation for Smart Glasses: A Lean Based Approach
The expansion of smart glasses represents the next frontier of wearable computing. The main obstacle for teams developing emerging hardware is the gap between building, testing, and validating a connectivity and interface architecture on a device that is not yet accessible. This challenge is amplified in short development timelines, where final hardware becomes available late. Since mobile and wearable ecosystems operate under volatile requirements and intense time-to-market pressure, deploying adaptable process models is essential to mitigate scheduling risks (Vallon et al., 2015). This work presents a strategy for early technical structuring and validation, mapping the Android Jetpack and Jetpack Compose Glimmer frameworks (Google, 2026) to assisted reality displays, and testing low-latency protocols such as Wi-Fi Aware and Bluetooth between smart glasses and other Android devices. Using the Lean methodology, we mitigated the lack of hardware by using Android smartphones to simulate smart glasses, ensuring software was built under realistic constraints. This iterative process consisted of rapid delivery of Proofs of Concept (PoCs) for evaluation, followed by refactoring to consolidate architectural robustness. Jetpack Compose Glimmer was used to design optimized interfaces, generating PoCs to test connectivity and ensure low-latency data flow. The strategy resulted in a reusable module and a set of libraries, consolidating knowledge about the application lifecycle in Head-Mounted Displays and identifying UX bottlenecks. An intermediate test environment simulated the final user experience before integration. By adopting a hardware-agnostic approach, the code and architecture are ready for immediate integration once Android XR smart glasses are available.
Kinderman Barros, Aecio Jorge Macedo Braga, Marcos Silbermann, Randerson Queiroz
Open Access
Article
Conference Proceedings
The Integration of Eye Tracking and Electrodermal Activity Measures for Assessing Cognitive Load and Mental Fatigue in Engineering and Computing: A Systematic Review (2010-2025)
This systematic review investigates the combined use of eye-tracking and electrodermal activity (EDA) techniques to evaluate cognitive load and mental fatigue within engineering contexts between 2010 and 2025. The analysis highlights three predominant directions in recent research: (1) advancements in systems designed for real-time monitoring of cognitive states, (2) the adoption of machine learning methodologies for integrating multimodal physiological data, and (3) the empirical validation of these metrics across varied engineering education and task environments. Nonetheless, several challenges persist, particularly in mitigating the influence of environmental factors on physiological signals and achieving consistency in analytical approaches across studies. Based on a review of 101 scholarly publications sourced from major databases, including Scopus, Web of Science, Engineering Village, ASEE PEER, and ACM, this study consolidates the current understanding of integrating eye-tracking and EDA in assessing cognitive performance and fatigue. Evidence indicates that combining eye-tracking parameters - such as pupil dilation, blink frequency, fixation duration, and saccadic movement with EDA measures (tonic and phasic components) yields more consistent and dependable indicators of cognitive states than unimodal assessments. Research consistently demonstrates that variations in pupil size and EDA correspond strongly with levels of cognitive load, while blink dynamics and phasic skin conductance responses effectively capture the onset and progression of mental fatigue. Specifically, larger pupil dilations reflect heightened information processing and cognitive exertion. Future investigations should aim to design more advanced algorithms for real-time multimodal data analysis, establish standardized frameworks for measurement and interpretation, and further explore the complex relationship between physiological activity and task-specific cognitive demands.
Naqash Gerard, Idalis Villanueva
Open Access
Article
Conference Proceedings
Exposing Tool Use in AI-Assistant Interfaces: Inspectable Evidence for Industrial Assistance
AI assistants use tools to obtain information and perform actions, but users may still need to inspect the basis of an answer or action. This paper explores an interface for inspecting document retrieval and PDF navigation, with tool requests and results shown alongside a conversation and source PDF. Three author walkthroughs established concrete evidence conditions, including incomplete retrieval, in which inspection could help a user check an answer. Three moderated expert sessions on Meta Quest 3 then explored how participants responded to this support. When invited to check an installation answer, all pursued a source check; two identified omitted or additional details, while one could not verify the answer. Participants used different evidence routes and suggested clearer links, optional detail, and task-relevant visuals. The findings motivate further development of accessible evidence interfaces that support users in checking assistant outputs and deciding what to do next.
Andrea Alesani
Open Access
Article
Conference Proceedings
IMAGES-AI: Generative AI to improve products for the US Navy warfighter
Project IMAGES-AI uses generative AI (GenAI) to improve products for the US Navy warfighter. Recent advances in GenAI, and the ability to adapt it to a specific domain, have made it a useful tool for enhancing the interfaces that display information used by the warfighter. The overarching goal is to improve situational awareness and strengthen the decision-making process, primarily through visual GenAI image generation. GenAI image generation has become a fast-growing capability and has overcome many of the early errors and mistakes seen in the technology's early days. Much of this has been made possible with improvements to Large Language Models (LLMs). LLMs can be very large and cover a wide range of data to support correct image generation. However, for very specific domains, much of the data required, such as for the US Navy warfighter and related topics, can be very limited, which can impact the results of GenAI-generated imagery for this specific topic. Low-Rank Adaptation (LoRA) is a parameter-efficient fine-tuning technique that is used to customize pre-trained models (LLMs). It does this by training a small set of extra weights instead of altering the massive LLM model. The concept is to freeze the base weights of the original pre-trained model to preserve its coverage and capability. A domain-specific lightweight companion matrix or adapter layers can be added to inject domain-specific data into specific parts of the model with attention mechanisms. No retraining of the LLM is required; instead, a smaller add-on portion is used, where gradient updates modify only the much smaller set of adapter parameters. This produces a LoRA file, essentially a plug-in that adjusts the final output to better match the desired characteristics. In this case, applicable to US Navy applications. This technique is cost- and resource-efficient, small in file size compared to LLM data, and easy to swap out for different adapters to align with the desired output. IMAGES-AI has leveraged this technique to generate imagery that better illustrates Navy missions and courses of action. IMAGES-AI has been applied to applications to test and verify whether improvements are seen in warfighters through improved situational awareness and rapid decision-making. Domain and environmental information become embedded in the imagery. This has been examined through integration into applications and used in storyboarding workflows and scenario development. Initial prototype results indicate a valid return on investment.
Bryan Croft, Seana Rothman, Tai Nguyen
Open Access
Article
Conference Proceedings
Applying Formal Methods to an Augmented Reality Interface
This study applies formal methods utilizing probabilistic reasoning to integrated system models containing a human cognitive component. As the line between user, environment, and computational system becomes increasingly blurred by immersive technologies, being able to prove safety properties of the full, integrated system, instead of just the automation component increases in importance. We present a model of a human-machine system based on augmented reality and use it to reproduce results from human-centered studies of cognitive load. Specifically, we formulate a model of feature integration theory of visual search as probabilistic timed automata. The model satisfies properties related to task performance such as identify whole system. The properties, in turn, provide system design and implementation guidance. Our end-to-end modeling, verification, and validation presented here corroborates the applicability of probabilistic model checking to the human/environment/automation system in an augmented reality (AR) application. These formal models permit designers to guarantee system properties, as well as assess the impact of design choices without conducting full user studies for each decision.
Jeremy Gottlieb, Dan Bryce, Mieszko Muskala, Stephen Gilbert, Michael Dorneich, Shayama Khan, Maral Jafari Ranjbar, Eliot H. Winer, John Skovbekk, Robert Goldman, David Musliner
Open Access
Article
Conference Proceedings
Room-Scale Digital Twin Sensing for Quantifying Young Children's Gaze-Based Interest and Movement Behavior
This study proposes a room-scale digital twin sensing system for jointly quantifying young children’s gaze-based interest and movement behavior within the same three-dimensional environment. The system integrates skeletal data acquired from multiple Azure Kinect sensors, a three-dimensional point cloud model of the room, and stimulus-state data within a unified coordinate system. Gaze-based interest is estimated from head orientation and visualized as an interest heatmap, while movement behavior is quantified using cosine similarity between the child’s movement direction and each target direction. As a proof-of-concept study, the system was validated in a stimulus-controlled room where the brightness and color of visual targets could be electronically controlled. Experiments with two four-year-old children showed that the activated LED target achieved the highest interest score and movement-direction score under all tested conditions. These results demonstrate the feasibility of integrating gaze-based interest estimation and movement-direction analysis within a common room-scale digital twin, providing a technical basis for quantifying how children visually attend to and move within everyday environments.
Yuya Kawabe, Shihomi Kuribayashi, Koji Kitamura, Tatsuhiro Yamanaka, Yoshifumi Nishida, Mikiko Oono
Open Access
Article
Conference Proceedings
Architecting an Integrated Data Management and Governance Ecosystem for Enhancing Micro, Small, and Medium Enterprises (MSMEs) Promotion in Thailand
Micro, Small, and Medium Enterprises (MSMEs) serve as the cornerstone of Thailand’s socio-economic development, contributing significantly to national employment and Gross Domestic Product (GDP). However, public promotion mechanisms remain fundamentally impeded by institutional data silos, fragmented registries, inconsistent semantic standards, and predominantly output-based monitoring across more than thirty disparate government agencies. These structural deficiencies foster administrative redundancies, impede equitable resource distribution, and limit data-driven policymaking. To resolve these challenges, this study formulates a comprehensive conceptual framework and architectural model for an Integrated Data Management System, positioning the Office of Small and Medium Enterprises Promotion (OSMEP) as an ecosystem orchestrator rather than a transactional operator. Grounded in The Open Group Architecture Framework (TOGAF), Design Science Research (DSR), and public governance theories, this research aligns the target architecture with national strategic priorities, including the 20-Year National Strategy, the 13th National Economic and Social Development Plan, and the Digital Government Development Plan. The proposed technical design models Phase C (Information Systems Architectures) by delineating a modern Data Lakehouse repository (Medallion Architecture) and an event-driven, composable Application Architecture. Central to the data layer is the institutionalization of the "SME One ID" as an immutable primary key, enabling seamless record linkage across tax authorities, business registration databases, and alternative credit channels. The application layer introduces automated anti-duplication mechanisms, real-time eligibility verification, dynamic consent management complying with the Personal Data Protection Act (PDPA), and an artificial intelligence-driven engine for precision interventions and prescriptive service matching. To establish sustainable cross-agency collaboration, the study articulates a robust Data Governance Framework incorporating federated data ownership, zero-trust access controls, and strict quality assurance service level agreements (SLAs). Furthermore, to evaluate the multi-dimensional value proposition of public digital infrastructure beyond conventional financial metrics, the study formulates a rigorous Social Return on Investment (SROI) evaluation model.
Varit Intrama
Open Access
Article
Conference Proceedings
Integrating human factors with safety risk management through the Bowtie Analysis: Forming a commercial pilot mental health risk management system
Commercial pilot mental health has become a complicated issue in the United States. While a portion of the pilot population would like to seek help openly with their ongoing mental health problems, many of those pilots are unwilling to seek the help due to the risk it poses in jeopardizing their pilot careers. While pilots with current mental health issues are unwilling to visibly gain the help they need, many other pilots are potentially borderline dealing with oncoming mental issues due to being under tremendous pressure with fatigue and stress issues underlying their performance. The Federal Aviation Administration (FAA) has recently addressed the issue, but the FAA, the airlines, the pilots’ union and the pilots are still hard pressed to come up with a working solution. The researchers propose that such a complex aviation safety issue infused with human factors (HF) should be analyzed from a different perspective of risk management (RM) and HF through a safety technique called the Bowtie method that is designed to both deal with prevention and reducing the consequences through RM. By integrating realistic prevention programs, the researchers complete a risk assessment (RA) of the preventative left side of the Bowtie. The researchers then integrate realistic recovery mitigations to reduce the consequences followed by the RA of the right side of the Bowtie. In both cases of RA on the prevention and recovery sides of the Bowtie analysis, both RA matrix and mathematical probability support the risk reduction. The safety RM analysis of the HF interventions through the Bowtie method proves that the complex problem of pilot mental health may not ever be 100 % resolved but instead can be adequately managed by RM HF interventions through the Bowtie method integrated into a pilot mental health RM Safety Management System (SMS).
Mark Miller, Sam Holley, Aric Raus
Open Access
Article
Conference Proceedings
Trust Calibration and Surveillance Auditing in Human-AI Co-Ideation
As generative artificial intelligence becomes deeply embedded in computing and design education, instructors face the urgent challenge of teaching students to think critically about machine suggestions rather than trusting them uncritically. This paper evaluates a structured, classroom-ready Human-AI Co-Ideation Workshop designed to protect student creative agency and reduce automation bias. In the activity, students tackled a forward-looking design challenge focused on student wellbeing and academic success. We introduced a deliberate pedagogical delay by dividing the workshop into sequential stages: students first generated initial design ideas entirely on paper without computers, then used multiple commercial language models to expand their concepts, and finally completed a comparative gap analysis and reflection. Analyzing student design worksheets, conversational interaction logs, and reflective essays revealed four consistent patterns: students struggled to bridge personal lived experiences with broad institutional systems; articulate machine outputs created a "Smoothness Trap" that discouraged questioning; learners had to actively protect their design confidence against polished machine text; and students practiced "defensive curation" by identifying and stripping out invasive surveillance and behavioral tracking mechanisms introduced by automated models. This paper provides instructors with an adaptable workshop structure, an evaluation rubric, and practical teaching strategies to help students collaborate with AI safely and thoughtfully.
Akram Bayat
Open Access
Article
Conference Proceedings
How and Why Did the Lack of LCES Contribute to Fatal Outcomes in the June 1990 Dude Fire, July 1994 South Canyon, and June 2013 Yarnell Hill Fire?
This paper examines a controversial issue in wildland firefighting: institutional responses to wildland firefighter fatalities and the role of fire-shelter safety reform. The analysis focuses on three major fatal incidents spanning 1990 to 2013: the June 1990 Dude Fire in Arizona, the July 1994 South Canyon Fire in Colorado, and the June 2013 Yarnell Hill (YH) Fire in Arizona. Particular attention is given to recurring issues identified in official investigative reports, independent analyses, and subsequent institutional policy responses. Using human factors, situational awareness, communication failures, training adequacy, and organizational culture as analytical lenses, the study examines whether fundamental safety elements – including escape-route planning, fire-behavior prediction, and adherence to established operational protocols – were adequately addressed. The paper also explores the gap between prescribed safety procedures and actual field conditions, while comparing post-incident recommendations with the underlying causal factors identified in these events. The Yarnell Hill Fire serves as a contemporary reference point for assessing whether lessons from earlier tragedies were effectively incorporated into training, policy, and operational practice. In addition, the analysis considers the phenomenon of risk compensation and the possibility that perceived protection from fire-shelter technology may influence tactical decision-making under extreme conditions. This human-factors investigation does not dismiss the legitimate protective value of fire shelters; rather, it questions whether an institutional emphasis on equipment performance may have limited a more comprehensive examination of preventable causal factors. Although fatalities cannot be eliminated entirely in inherently hazardous occupations, their likelihood may be reduced through rigorous investigations, accurate identification of causal human factors, and effective organizational learning. The paper therefore proposes a broader framework for incident analysis that extends beyond equipment evaluation to include organizational culture, training effectiveness, situational awareness, communication, and systemic risk-management practices. Recognizing and addressing potential institutional biases in wildland firefighter safety investigations remains critical to reducing future tragedies and honoring those who lost their lives.
Fred Schoeffler
Open Access
Article
Conference Proceedings
Adjudicating the Imperceptible: Human Review of Concealed-Content Detections in AI Document Ingestion
Organizations adopting large language models for document-intensive work increasingly place an automated screening stage between untrusted source documents and the model. One such control detects concealed text: content present in a document's data layer but not visibly rendered, as white on white, zero height, beyond the page boundary, or sub-point type. Because a concealed passage can carry instructions to a downstream model while remaining invisible to any reader, detections are escalated to a human reviewer who decides whether each passage is released into machine-readable output or withheld. This paper contends that the resulting adjudication task is structurally unsound and presents field evidence of its failure. The reviewer must authorize content whose defining property is that it cannot be perceived, using a decision aid that presents an extracted string divorced from the visual context that would make the concealment meaningful. In production use of an offline ingestion system for United States federal contracting work, two unattended batches processed 20,346 pages and about 6.24 million source tokens. The completeness control withheld 13 documents over a shortfall of about 51 tokens, while the concealed-content scanner raised four detections, all released by reviewer override with recorded justifications, one stating of one-point type that no text was visible. The observations are situated in the trust calibration and compliance literature and contrasted with the completeness control, whose throughput cost operators relieve through a permissive mode. A controlled study comparing string presentation with rendered, geometry-annotated presentation is described, and the paper recommends presenting evidence of concealment rather than the concealed content and asking whether the concealment is explicable rather than whether the passage is safe.
Richard Stone, Michael Jenkins, Banyan Foust, Abagail Miller, Alexa Ramos
Open Access
Article
Conference Proceedings
The impact of human factors on counter-drone systems
Drones—particularly small and medium-sized systems—have become a major asset on the battlefield. Whether deployed individually or in swarms, coordinated or not, they have demonstrated significant impact in conflicts such as those in Ukraine and in regions like the Strait of Hormuz. The scale of damage they can inflict has become a growing concern. While systems have long existed to automatically detect and counter large drones, they are significantly less effective against small and medium-sized ones. Thus, due to late and limited detection, the human operator who makes the final decision on whether or not to engage a countermeasure has become central. In this paper, we explain why this is the case, highlighting its role and the mental load that is placed on him/her.
Serge Chaumette
Open Access
Article
Conference Proceedings
The FLOODPERCEP Dataset: Flood Hazard Maps and Imagery for Studying Human Perception of Flood Risk
Understanding public risk perception is essential as climate-related hazards, including floods, become more intense, frequent, and widespread. This perception varies across population groups, and is influenced by factors, such as sociodemographic characteristics, previous hazard exposure, and the format of risk information. Individuals process visual hazard information, such as flood hazard maps and photographic imagery, differently. We present FLOODPERCEP, a new dataset designed to examine public interpretation and perception of flood risk. The dataset integrates geospatial risk products from governmental and non-governmental entities, as well as web-sourced, geolocated photographic evidence of historical inundation events from 260 U.S. locations, complete with metadata, alongside contextual Google Maps views. The dataset also contains quantitative and qualitative responses from n = 796 study participants who viewed this visual data. This extensive set of information enables detailed analyses of the level of comprehension of multi-format flood risk information, as well as public risk perception and behavioral intentions. Ultimately, this dataset aims to support the design, development, and deployment of more effective, inclusive, and evidence-based hazard communication strategies.
Amir Behzadan, Subashree Dinesh, Armita Dabiri, Arunima Maitra
Open Access
Article
Conference Proceedings
Distance-Dependent Robustness of WiFi CSI-Based Movement Detection under Outdoor LoS and NLoS Conditions
Device-free Wi-Fi Channel State Information (CSI) sensing offers infrastructure-free motion detection for outdoor security applications such as agricultural perimeter monitoring, yet nearly all published systems are validated indoors under line-of-sight (LoS) and weather-controlled conditions, leaving the performance under non-line-of-sight (NLoS) conditions with vegetation-induced clutter and path offset unquantified. This gap matters because agricultural deployments routinely introduce lateral obstructions — tree canopies, crop rows, and irregular sightlines — that alter CSI signal structure in ways LoS-trained detectors cannot anticipate, making field-ready deployment guidelines essential rather than optional. We tested two detectors — short-time standard deviation (SD) and short-time Fourier transform (STFT)-based Doppler energy — using an ESP32 CSI platform (2.4 GHz, ~100 Hz) across a mango orchard (1–2 m tree spacing) and an open bare field, at NLoS offsets of 0.5–1.5 m, 1.5–2.5 m, and 2.5–3.5 m, against slow/normal/fast LoS baselines. The SD detector collapsed under NLoS attenuation, converging toward the fast-LoS baseline and losing threshold separability beyond 1.5 m, with F1 falling from 0.78 to 0.40 in the orchard and to zero in the open field, whereas Doppler energy remained compact and separable, sustaining F1 of 0.66–0.77 across all open-field distances and degrading more gracefully under vegetation. These results establish a 0.5–1.5 m NLoS reliability boundary for variance-based sensing. Critically, the two environments failed in opposite directions — the orchard degraded SD gradually through attenuation, while the open field caused abrupt threshold collapse — indicating that NLoS-robust CSI sensing requires Doppler-domain features whose reliability depends on scene geometry, not distance alone.
Prasanta Kr Sen, Mrigank Sharad, Ram Babu Roy
Open Access
Article
Conference Proceedings
Designing Integrated Human-AI Systems: A Socio-Technical Approach to Knowledge Management in Healthcare
The integration of artificial intelligence (AI) into healthcare knowledge management (KM) offers opportunities to improve knowledge flows, decision-making, and clinical workflows. However, current AI applications remain fragmented and often address isolated tasks rather than supporting system-wide integration. This conceptual paper examines AI-enabled KM through a socio-technical systems perspective and proposes a collective adaptive socio-technical framework comprising four interdependent dimensions: the KM process, social, technological, and relationship contexts. The framework emphasizes interactions among healthcare professionals, AI systems, organizational infrastructure, and regulatory environments, supported by continuous learning and feedback. The conceptual analysis suggests that effective AI integration requires more than technological capability; it also depends on usability, trust, human-AI collaboration, interdisciplinary cooperation, a knowledge-driven culture, and strategic leadership. By positioning AI as an embedded component of an interconnected healthcare system, the framework advances a human-centered approach intended to support collaborative intelligence and continuous adaptation.
Kristin Weger, Vineetha Menon, Jennifer Bail
Open Access
Article
Conference Proceedings
A Regulatory and Human Factors Approach to Overreliance in Legal Information Chatbots
Chatbots are rapidly transforming the way individuals from diverse social and economic backgrounds access legal information. This free access to legal knowledge can act as a corrective mechanism within society, reducing inequalities among social groups. Notwithstanding their benefits, the growing use of chatbots for legal matters raises significant concerns. One major issue is the accuracy of the information provided, since correctly applying a rule is a complex exercise, not a simple act of reading. Additional risks involve the processing of personal and sensitive data, which may be leaked or over-shared without users' deliberate consent or awareness. Concerns also arise regarding algorithmic bias, since AI systems may reproduce discriminatory patterns embedded in training or design data. Furthermore, chatbots may raise issues concerning the unauthorized practice of law. This article argues that legal question-answering chatbots, regardless of their pre-designed purpose, may endanger users' fundamental rights, and that classifying them as limited-risk systems undermines the goal of ethical, responsible and trustworthy AI. It further argues that the transparency framework under Article 50(1) of the AI Act is insufficient and proposes embedding ethics-by-design and ethics-by-default through a set of concrete interaction safeguards to actively support verification of chatbot-provided information before reliance: cognitive forcing functions, interaction friction, scaffolded verification, and calibrated uncertainty and limitation cues.
Niki Georgiadou, Maria Rigou
Open Access
Article
Conference Proceedings
Psychological Foundations of Epistemic Profiles for Misinformation Resilience: Implications for Defence-Oriented Agent-Based Modelling
This conceptual and methodological paper proposes a psychological and epistemological basis for agent-based modelling of misinformation resilience in defence-relevant information environments. We integrate bounded rationality, attention, threat appraisal, social identity and relational trust with intellectual humility, actively open-minded thinking and epistemic vigilance. Rather than classifying people into fixed susceptibility types, the proposed epistemic profiles distinguish enduring dispositions, context-dependent states and source-specific social relations. We specify separable behavioural stages—exposure, evaluation, belief updating, verification, sharing and correction—and offer a candidate mapping from theoretical constructs to agent parameters. Drawing cautiously on the Gateway Belief Model, we also propose credibility, relevance, efficacy and legitimacy as hypotheses concerning conditions for protective information behaviour, without assuming that climate-domain effects transfer to defence settings. Potential applications include testing defensive communication and verification procedures under time pressure, incomplete information and military organisational constraints. A staged validation agenda is outlined; no simulation results or validated psychological profiles are claimed. The proposed framework is developed in the context of Czech Technology Agency project TQ23000183.
Petr Zamecnik, David Černý, Vojtěch Juřík, Darina Zamecnikova
Open Access
Article
Conference Proceedings
Detecting Information Leakage in Predictive Process Mining: A Three-Layer Evaluation Framework
Predictive process monitoring aims to provide early warnings about process outcomes, yet published benchmarks frequently report near-perfect prediction scores on full process traces. This paper investigates whether those scores reflect genuine predictive ability or information already embedded in the event log. A simulated supply-chain dataset with 818,292 events across 25,000 linked cases was used to evaluate 276 model configurations under controlled conditions where ground truth causal labels were available. Three forms of information leakage were identified- activity names that encode outcomes, conformance features that overlap with prediction targets, and temporal variables that measure downstream effects rather than upstream causes. When these sources were retained, models achieved an AUC of 1.00. After leakage was controlled through prefix-based evaluation, scores fell to between 0.76 and 0.93. A diagnostic procedure with specific detection thresholds is proposed for each layer. The framework was examined on two external benchmark datasets from different domains, and all three leakage patterns were independently observed in both. These findings suggest that near-perfect classification scores in predictive process mining should be interpreted with caution, and that a systematic leakage audit should precede model comparison or deployment.
Fathima Lodi, Amir Schur
Open Access
Article
Conference Proceedings
Building Sustainable Business Value through Human-Centred AI Transition Pathways
Successful AI implementation depends as much on organisational readiness as on technological maturity. This paper proposes a business-oriented transition framework linking Human Factors, competency development, AI governance and organisational transformation. The model aligns investment decisions with operational performance, workforce capability and long-term resilience rather than technology acquisition alone. Aviation case studies demonstrate how organisations can generate measurable business value through structured implementation pathways that integrate AI, Human-AI Teaming and competency assurance into everyday operations.
Dimitrios Ziakkas, Debra Henneberry
Open Access
Article
Conference Proceedings
From Water Quality Data to Sustainable Innovation: Designing Learning Ecosystems for Systemic Transformation
This paper presents design principles and a transferable model for developing sustainability-oriented learning ecosystems that connect emerging technologies, business relevance, public engagement, and international collaboration. Sustainability challenges are often addressed as stand-alone educational assignments, despite requiring engagement with complex socio-technical systems involving data infrastructures, governance, stakeholders, communities, values, and organizational decision-making. The study examines how learning ecosystems can be intentionally configured to support systemic transformation while developing students’ capabilities in systems thinking, stakeholder engagement, responsible innovation, and cross-institutional collaboration. Using an Action Design Research approach, the paper reports the iterative development of a learning ecosystem built around the “Right to Know H2O” challenge, which focuses on access to, interpretation of, and action based on real-time water quality information. The learning ecosystem connects student teams, university-based innovation hubs, domain experts, mobility activities, and public-facing communication mechanisms. Building on insights from three iterative design cycles, the paper articulates nine design principles that synthesize into a transferable model for organizing sustainability-oriented learning ecosystems. The model comprises three interrelated components – challenge configuration, ecosystem amplification, and codification and assessment – that together provide practical and theoretically informed guidance for universities and partner organizations seeking to organize learning around authentic sustainability challenges. The study contributes transferable design knowledge for organizing learning ecosystems that integrate emerging technologies while generating stakeholder value, societal relevance, and learning experiences that extend beyond conventional course boundaries.
Jari Jussila, Andreea Cotoranu, John Cronin, Markku Mikkonen
Open Access
Article
Conference Proceedings
From Inquiry to Handoff: A Reference Architecture and Design Implications for Photorealistic Conversational AI Agents in First-Contact Services
In knowledge-intensive consulting services, the first contact with potential clients helps build mutual understanding and prepares for the next steps. However, this process is often managed through static forms or telephone hotlines that provide limited guidance and lead to repetitive intake tasks. Although conversational AI is established in customer service, its use as an end-to-end first-contact service remains underexplored. This paper presents the design, implementation, and evaluation of a photorealistic conversational AI agent in the real-world consulting environment of a premium kitchen manufacturer. The agent elicits customer requirements in dialogue, generates a structured summary, schedules the consultation, and transfers the result to a human consultant. Following a design science research approach, a reference architecture realizes the service workflow, and the artifact is evaluated with 19 prospective customers through a standardized questionnaire and semi-structured interviews, complemented by interviews with the organization. Results indicate high trust, perceived transparency, and data comfort; a majority would prefer starting with the agent over waiting for a human, whereas social presence and embodiment polarize across interaction styles. The findings are synthesized into five design principles.
Nick Tugarin, Christian van Husen
Open Access
Article
Conference Proceedings
Threshold-Based Key Management Across Heterogeneous Multi-Cloud Key Management Systems
As organizations increasingly adopt multi-cloud architectures, cryptographic key management has emerged as a critical security challenge. Although workloads may be distributed across multiple cloud service providers (CSPs), cryptographic authority often remains concentrated within a single provider's Key Management System (KMS), creating potential single points of failure, vendor lock-in, insider threat exposure, and operational dependencies. Existing threshold cryptography research demonstrates the benefits of distributed trust; however, many proposed solutions assume homogeneous environments or rely on centralized intermediaries that diminish the resilience advantages of multi-cloud deployment. This research introduces a threshold-based cross-cloud key management architecture that distributes cryptographic trust across heterogeneous cloud-native KMS platforms, including AWS Key Management Service, Microsoft Azure Key Vault, and Google Cloud KMS. The proposed architecture is embodied in the Benzene Ring Secure Cross-Cloud System (BRSCCS), in which a secure orchestration layer serves as the protected nucleus. At the same time, independent cloud providers serve as separate trust domains holding distributed key shares. Using Shamir’s Secret Sharing, a 256-bit AES master key is divided into multiple shares under a 2-of-3 threshold policy, ensuring that no individual cloud provider possesses sufficient information to reconstruct the complete key. The architecture incorporates several security-enhancing components designed for practical deployment in modern cloud environments. A confidential orchestration layer operating within a Trusted Execution Environment enables transient key reconstruction while minimizing key exposure. Federated identity services leverage short-lived credentials and workload identity federation to eliminate long-term secrets. Dynamic share rotation mechanisms periodically refresh distributed shares without modifying the underlying cryptographic key, reducing the impact of potential compromises.
Musa Adamou, Ivan Nkwawir, Amir Schur
Open Access
Article
Conference Proceedings
Evaluation of Models of Functional Automation and Driver SA in Automated Vehicle Control
The Society of Automotive Engineers (SAE, 2021) proposed a taxonomy of Levels of Driving Automation (LODA). Some research (e.g., Deng & Kaber, 2022) investigated effects of LODA on driver attention, performance and workload. The discrete nature of LODA definitions limits utility of results for systems design. We formulated an index of the rate of automation (AR) of vehicle driving functions based on elaborate task analysis and classification to stages of human information processing (Liu & Kaber, 2021). We also formulated a model of driver situation awareness (SA) based on AR index values as well as measures of driver expertise and attitudes towards vehicle automation. We showed goodness of fit of a scaled exponential function with shift to objective SA data from prior experiments and meta-analyses (Onnasch et al., 2014). Based on a new experiment, we tested the model for utility in predicting driver SA at LODA 0 (manual), 1 (driver assistance) and 4 (high driving automation). We found significant differences in SA responses among LODA (as quantified using the AR index) as well as significant effects of driver expertise. The new SA model showed superior fit to experiment data across the SAE LODA, as compared with simple linear statistical models with the AR or LODA as sole predictors. Next steps include use of the driver SA model as a design tool to prototype and test advanced driver assistance system variations to optimize driver performance based on the rate of vehicle automation (AR).
Yunmei Liu, David Kaber
Open Access
Article
Conference Proceedings
Visual Attention, Recognition, and Driver Response During Emergency-Vehicle Encounters in a VR Driving Simulator
Emergency-vehicle (EV) encounters are rare but safety-critical events, making naturalistic observation of driver attention difficult. This study examined visual-search behavior and response initiation during controlled EV encounters using a virtual-reality driving simulator with eye tracking and SUMO-based traffic simulation. Thirty-five participants completed five of eight possible combinations of traffic condition, siren availability, and in-vehicle warning (IVW), yielding 339 same- and opposite-direction EV encounters. Visual attention was quantified using area-of-interest (AOI)-based dwell allocation and normalized gaze entropy. AOIs included the EV, IVW, mirrors, surrounding traffic, front scene, speedometer, and other regions. Stable pre-encounter segments served as non-EV reference periods. First EV-related visual engagement was defined as the earliest sustained gaze to the EV, IVW, or mirrors, while first action was the earliest validated braking, steering, or turn-signal response. Non-EV and EV gaze measures were compared using within-run regression models, while experimental-factor effects were evaluated using generalized estimating equations (GEE). The findings demonstrate that EV warning conditions influence visual-search redistribution and response timing, particularly during same-direction encounters. During EV1 encounters, gaze entropy increased from 0.346 to 0.485, mirror dwell increased from 7.2% to 26.8%, and forward-roadway dwell decreased from 75.7% to 58.1%. EV2 encounters showed less overall visual redistribution but a marked increase in direct EV gaze. IVW availability was associated with larger increases in entropy and mirror dwell and with earlier EV-related visual engagement. Both IVW and siren availability were associated with earlier first-action timing during EV1 encounters.
Lanmin Liu, Robert Kluger
Open Access
Article
Conference Proceedings
Trust Calibration in Semi-Automated Rural Driving Under Reduced Visibility: Effects of Visibility and Collision-Warning Context on Driver Reliance
Advanced driver assistance systems (ADAS) increasingly support vehicle control while requiring drivers to monitor system behaviour and remain prepared to intervene as operating conditions change. Appropriate reliance is therefore particularly important in rural environments, where degraded visibility and unexpected roadway hazards may increase uncertainty surrounding automated support. This study examined scenario-level ADAS reliance and related system perceptions during repeated rural driving. Ten licensed drivers completed twelve counterbalanced simulator-based scenarios in a 3 × 2 × 2 within-subject design crossing visibility (low, medium, high), driving mode (manual, ADAS-supported), and collision-warning alarm condition (off, on). The present analysis focused on the six ADAS-supported conditions in which repeated ADAS-perception measures were collected. A linear mixed-effects model examined reliance as a function of visibility, alarm condition, and their interaction while accounting for repeated observations within drivers. Under low visibility, alarm activation was associated with substantially lower reliance (β = −1.70, 95% CI [−2.70, −0.70], p = 0.001). Visibility contrasts within the Alarm-Off reference condition were not statistically reliable, and the Visibility × Alarm interaction contrasts did not reach the conventional significance threshold. Multidimensional questionnaire profiles further showed that reliance, perceived safety, competence, predictability, explicit trust, and error awareness were related but non-equivalent aspects of ADAS perception. These findings highlight the importance of evaluating driver reliance as a dynamic and multidimensional response to environmental and interface context.
Amir Milad Moshref Javadi, Tekalign Mengistu, Natnael Takele, Steven Jiang, Ali Karimoddini
Open Access
Article
Conference Proceedings
From Driver Intent to Road-User Perception: A Narrative and Systematic Review of Perceived Aggressive Driving
Aggressive driving (AD) has no unified definition: some perspectives focus on the driver's intent, whereas others focus on measurable violations. Consequently, studies using divergent definitions measure fundamentally different phenomena, meaning behavior that surrounding road users perceive as aggressive is excluded if no harm was intended or no law was broken. This paper therefore proposes perceived aggressive driving (PAD), defined as any driving behavior that surrounding road users perceive as aggressive, regardless of intent, initiating agent, or legal status. A narrative review of prior definitions motivates a four-stage, process-oriented framework linking triggering factors, driver motivation, driving behaviors, and outcome impacts. A PRISMA 2020 systematic review, comprising 63 included studies, was then mapped onto this framework. Results were distributed unevenly across four stages, almost all studies documented triggering factors and observable behaviors (k = 59 each), while most reported driver motivation (k = 51) and only a minority examined outcome impacts (k = 21). The evidence supports features of the framework including broad triggering factors, dual motivational pathways, a substantial non-violative category invisible to official records, and explicit behavior-to-outcome links. Overall, these findings validate PAD framework as a workable construct while extending its application to mixed-traffic environments and informing clearer regulatory strategies for perceived aggressive behaviors.
Yan Chen, Moajjem Chowdhury, Yunmei Liu
Open Access
Article
Conference Proceedings
A Vision-Based Framework for Detecting Overtakes and Quantifying Novice Driver Responses in Continuous Simulator Traffic
Novice drivers struggle to anticipate and respond to unexpected aggressive maneuvers by surrounding traffic. Most simulator studies use isolated, scripted hazards, which do not capture overtakes that emerge unpredictably in continuous traffic. This study developed a computer-vision framework to identify overtaking events in driving-simulator recordings and examine driver behavior during each event. Eighteen drivers with less than three years of U.S. licensure completed 144 trials under non-overtaking and overtaking traffic conditions. Overtakes were not scripted, and their timing, location, and frequency varied across trials. YOLO11x and motion-based ByteTrack were used to detect vehicles and overtaking events by tracking their movement. Overtaking events were manually reviewed and synchronized with driving logs recorded at approximately 30 Hz. Sixteen trials were used to calibrate the detection framework. On the remaining 128 held-out trials, the framework achieved 94.97% precision, 78.70% recall, and an F1 = .861. Drivers’ behavioral analyses showed that in the overtake condition, mean speed was 8.26 mph lower and time above 45 mph was 40.00 percentage points lower than in the non-overtake condition. The framework provides an event-level method for studying traffic interactions in continuous simulator data and connects those events with measurable driver behavioral responses.
Udit Kumar Das, Yan Chen, Moajjem Chowdhury, Yunmei Liu
Open Access
Article
Conference Proceedings
Experiencing Autonomy: Passenger Perceptions of Self-Driving Shuttles Across Diverse User Groups
Self-driving shuttles (SDSs) are increasingly considered as a means of extending public transport connectivity, particularly for short-distance and first- and last-mile trips, yet their integration into multimodal networks depends on whether diverse users find them usable and relevant to everyday travel. Existing evidence on acceptance is predominantly on scenario-based studies, with participants who have not experienced the SDSs firsthand and rarely reflects the perspectives of marginalised users. This study investigates attitudes towards and perceptions of SDSs through a real-world trial in Milton Keynes, UK, using a mixed-methods approach. Twenty-six participants across five demographic groups including underrepresented populations completed pre- and post-ride surveys and took part in focus groups. Survey responses indicate that feelings towards the SDS and willingness to recommend and pay shifted upward after the ride. Focus group findings explained why these perceptions changed. Trust was supported by visible forms of reassurance, including onboard operators and information about the SDS’s detection and decisions. However, concerns centred on braking, decision-making at roundabouts, and personal security when travelling without a driver. These findings inform the vehicle design, service operation, and fare and integration policy needed to support the transition from SDS trials to driverless deployment.
Jiahui Qian, Guilhermina Torrao, Joshua Dsouza, Joseph Mundy, Anna Mizen, Jisun Kim
Open Access
Article
Conference Proceedings
A Real-Time Identification Method of Pilot’s Stress State Based on ECG Features
High stress states during flight operations impair pilots’ capabilities of decision-making and control, serving as a critical human factor contributes to flight accidents. Conventional methods for stress detecting mostly rely on post-hoc tests by salivary cortisol and subjective stress report, which fail to realize monitoring in real time and warning for pilots’ stress state. This paper proposes a real-time stress identification scheme for pilots based on dynamic electrocardiogram (ECG) signals. Following the psychophysiological equivalence principle, a ground VR simulation platform was constructed to induce altitude-related stress responses, and ECG signals, salivary cortisol concentrations and subjective stress reports were synchronously collected from 12 professional pilots. With the labels were annotated by combining salivary cortisol indicators and subjective report scores, and the heart rate was selected as the core input feature, a binary classification model distinguishing normal and high-stress states was established based on RBF-SVM, grid search and leave-one-out cross-validation were applied to enhance the generalization performance of the model. Furthermore, practical verification was carried out in the real flight scenario of Field Carrier Landing Practice (FCLP), by comparing the data of variation heart rate through novice pilots and experienced pilots, the results generated by the model were highly consistent with subjective stress assessments. This research findings provide a feasible technical support to management pilots’ stress in real time under airborne application scenario, and meanwhile reveal the pattern that pilots’ stress responses are alleviated with the growth of training proficiency.
Hua Zhao, Ran Peng
Open Access
Article
Conference Proceedings
Human Driver Temperament and the Safety Impact of a C-V2X Denial-of-Service Flooding Attack in Mixed-Autonomy Traffic
Cooperative and connected automated vehicles (CAVs) rely on Signal Phase and Timing (SPaT) messages to cross signalized intersections; a denial-of-service (DoS) flood that blocks SPaT forces CAVs into a fail-safe mode. Because human-driven vehicles share the intersection, the safety consequence depends not only on the attack and the CAV fail-safe policy, but on how the surrounding human drivers behave. We investigate this human-factors dimension with a coupled OMNeT++/INET (5G NR-V2X) and SUMO microsimulation of a signalized corridor, sweeping CAV market penetration (10–90%), four calibrated driver temperaments (cautious to aggressive) and two standards-based fail-safe policies, a minimal-risk maneuver (MRM) and an adaptive cruise control (ACC) keep-driving fallback, with each attack arm differenced against its policy-matched no-attack baseline. Temperament's effect on the attack is specific and modest rather than a blanket amplification. Aggressive surroundings worsen one metric, the hard-braking a keep-driving fail-safe forces on nearby drivers (p = 0.03), rising from near zero to +8 episodes/1000 veh-s. They appear to dampen rear-end conflicts, but only because the flood clears the queues aggressive drivers build, so the gain is in flow, not safety. On the attack's primary signatures, CAV red-light running and crossing conflicts, temperament has no detectable effect. It instead dominates baseline risk, producing a 13- to 18-fold cautious-to-aggressive gradient far larger than the attack itself, which acts through a channel already congested by CAV adoption. Human driver populations determine how dangerous the intersection is but do not systematically amplify this attack, so fail-safe design cannot assume a cautious test population bounds the risk.
Rasheed Bello, Varghese Vaidyan, Gurcan Comert, Vijay Bendigeri, Akinbobola Jegede, Judith Mwakalonge
Open Access
Article
Conference Proceedings
Continued Assessment of the Shortened Attention Control Tasks: Impact of Practice Effects, an Interfering Pre-cue, and Speed-accuracy Manipulation
Our previous research has demonstrated the viability of three-minute long (“squared”) double conflict attention control tests for predicting early flight training outcomes in U.S. Naval flight students. The present study was an effort to continue exploring and refining these tests by assessing both learning and the impact of two experimental manipulations (an interfering pre-cue and a speed-accuracy manipulation) in the Double Stroop task, a shortened variant of the traditional color Stroop task. We administered 12 Double Stroop tasks, two Double Flanker tasks, three processing speed tasks, and a survey assessing video gaming experience to a final sample of 131 Naval flight students and 64 enlisted Sailors (N = 195). We found the pre-cue interval and order of administration had a substantive impact on performance, as did the speed-accuracy manipulation. Further, learning seemed to plateau after seven attempts of the Double Stroop task (i.e., after around 10 minutes). Processing speed was moderately related to performance on the double conflict tasks. Action video gaming engagement was mostly unrelated to improvements, although there was weak evidence that gamers showed slightly larger gains with practice. These results may help understand the nature of performance and processes at play in these double conflict tests more generally.
Christopher Draheim, Ciara Sibley, Kaylin Strong, Michael Kukenberger, Joseph Coyne
Open Access
Article
Conference Proceedings
Reducing Nursing Communication Burden Through a Digital Health Passport Platform: A Patient-Centered UX Design Approach
During outpatient visits, patients are often unable to fully describe their symptoms due to limited consultation times. Furthermore, the complexity and highly specialized nature of medical information frequently lead to comprehension difficulties and cognitive overload for patients post-consultation. This, in turn, increases the workload and stress of nursing staff, who must repeatedly explain procedures and provide health education. To address these challenges, this study proposes a "Digital Health Passport" service framework centered on personal health data linked via the National Health Insurance (NHI) card, integrating authorized medical data. Through an AI-assisted pre-consultation feature, the system helps patients organize their symptoms and health information prior to their appointments. Post-consultation, the system transforms diagnostic results, medication instructions, and care recommendations into easily understandable digital content, enabling continuous tracking and management for both patients and caregivers. Ultimately, this study aims to develop a smart healthcare App platform grounded in UI/UX design, integrating functions such as health data management, AI consultation assistance, medical information visualization, and caregiving collaboration. By doing so, it seeks to improve patients' health literacy, optimize the efficiency of doctor-patient communication, and alleviate the care burden on nursing staff and caregivers, thereby serving as a vital reference for the innovative design of future smart healthcare services.
Zun-hwa Chiang, Cheng-Yun Wu, Wei-wei Su
Open Access
Article
Conference Proceedings
An Entrustment-Based Human Factors Framework for Integrating Artificial Intelligence into Healthcare and Medical Education
Artificial intelligence (AI) is increasingly transforming medical education and clinical practice through image interpretation, risk prediction, diagnostic decision support, and automated documentation. As learners encounter these technologies, medical education must prepare them not only to understand and use AI, but also to critically evaluate when to trust, question, or override AI outputs. Existing approaches to AI education commonly emphasize digital literacy, technical understanding, and ethical use, yet provide limited guidance regarding how supervisors should determine whether a learner can use AI responsibly with increasing independence. This work proposes an entrustment-based human factors framework for integrating AI into healthcare and medical education. The framework brings together three elements: human factors principles, which emphasize the interaction among clinicians, learners, technologies, tasks, teams, and clinical environments; the patient-centered CARES framework (Communication, Autonomy, Respect, Equity, and Safety) which defines the responsibilities that should guide AI-supported care; and entrustment, which links these responsibilities to observable learner behaviors, supervision decisions, and progressive independence. Within this framework, Communication requires clinicians to explain the role, limitations, and uncertainty of AI-supported decisions. Autonomy requires preservation of meaningful patient and clinician choice. Respect includes attention to privacy, dignity, and individual context. Equity requires recognition of biased or poorly generalizable performance. Safety requires active monitoring, appropriate human oversight, and thus the prevention of harm. We have observed that AI introduces a second-order entrustment challenge into medical education. Supervisors must determine not only when learners can be trusted to perform a clinical task, but whether they can be trusted to judge when, why, and to what extent an AI-enabled system should itself be trusted. This requires a metacognitive capacity for trust calibration. Learners must recognize the limitations of both human and algorithmic judgment, communicate uncertainty, and seek guidance when the limits of safe independent action are reached. As learners progress toward independent practice, they increasingly assume a learner-supervisor role in which they are responsible for overseeing the contribution of AI to patient care. Within this framework, entrustment decisions can help supervisors determine when learners are ready for greater independence in using AI-supported tools while maintaining appropriate human oversight, continuous learning, and patient safety. We recommend integrating CARES-informed competencies into entrustment decisions in order to provide a practical foundation for curriculum development, assessment, and the safe incorporation of AI into healthcare and medical education.
Bryan Johnston, Scott Adams, Kiven Erique Lukong, Jay Kalra
Open Access
Article
Conference Proceedings
ConoCancer: A RAG-Based Chatbot for Breast Cancer Health Literacy
Breast cancer remains one of the most commonly diagnosed cancers among women worldwide, yet gaps in health literacy and language access continue to limit early detection and informed decision-making, particularly among low-literacy Hispanic/Latina populations. Conversational AI offers on-demand health education at scale, but general-purpose large language models deployed without grounding frequently produce fluent yet unsupported or outdated responses, a failure that is especially costly in medical contexts. A further complication is that many patients in this population communicate by code-switching between English and Spanish rather than in one language exclusively. To address these gaps, we developed a bilingual intelligent tutoring system for breast cancer education, constituting short instructional videos with interactive quizzes and a chatbot for answering foundational breast cancer questions. However, this had some limitations that are targeted in this study. This paper focuses on improving the Quiz Scoring module and the Chatbot. We replaced the existing chatbot's architecture with a Retrieval-Augmented Generation (RAG) pipeline and keyword-based quiz grading with semantic grading and natural language inference (NLI). We evaluate the performance across four language variants (English, Spanish, and code-switched variants). Generation quality is assessed with Ragas (RAG Assessment), and retrieval performance is measured against the source passage from which questions were drawn. Generation was found comparable across language variants, but retrieval was not. The Spanish and code-switched queries could retrieve the corresponding passage, but failed to rank it as first option. Semantic grading outperformed the keyword baseline for only English, while NLI outperformed both approaches for all variants.
Alexa Rivera, Anika Bokka, Francisco Iacobelli, Renu Balyan
Open Access
Article
Conference Proceedings
Utilization of Multi-Stakeholder Visualization in Medical Management Strategy Development: A Case Study of the BSP Cardiovascular Detection Project
In the wake of the rapid advancement of medical informatization, the diversity of stakeholders involved in medical scenarios has expanded, rendering traditional management models inadequate for addressing the needs of multi-party collaboration. Against the backdrop of cardiovascular health detection management, this study delves into the requirements of five distinct stakeholder groups: testers, customer service representatives, nurses, doctors, and management personnel. It aims to design a suite of visualization tools to bolster management decision-making processes. The implementation of the tool drew upon educational methods and tools, such as problem-based learning and collaborative learning. By comprehensive user research, we pinpoint gaps in existing medical management systems concerning stakeholder inclusion and visual decision support, subsequently proposing a web- and mobile-based visualization solution. The present study utilizes semi-participatory observation and in-depth interviews to gather requirements and devises a visualization system encompassing data statistics, workflow optimization, and information architecture. Practical implementation reveals that the tool substantially improves collaboration efficiency among various roles, offers intuitive data support for management, and refines resource allocation decisions. This present study provides new visual design ideas and practical references for multi-stakeholder collaborative management in medical scenarios.
Shunjia Zhang, Yuan Liu
Open Access
Article
Conference Proceedings
A Task Selection Method for Alzheimer's Disease Patient Prediction to Reduce Patients’ Burden
Alzheimer’s disease (AD) is one of the leading causes of dementia, and its diagnosis can impose a burden on patients due to examinations requiring specialized equipment and sustained effort. Handwriting-based assessment has attracted attention as a simplified diagnostic approach; however, such assessments may still require older adults suspected of having AD to perform multiple tasks. This study proposes a task selection method to reduce the number of handwriting tasks while maintaining classification performance. For each task-feature combination, measurements from 165 participants were compressed into a one-dimensional latent representation using an Autoencoder (AE). The resulting representations were organized into a task-by-feature matrix, and Principal Component Analysis (PCA) was applied to characterize the tasks. Tasks considered to contribute little to discrimination were identified based on their principal component scores and removed. Classification performance was evaluated using Random Forest (RF), Support Vector Machine (SVM), and Multilayer Perceptron (MLP). The results showed that accuracy decreased for SVM and MLP as the number of tasks was reduced, whereas RF maintained relatively stable accuracy even with fewer tasks. These results suggest that the proposed method can identify tasks that contribute little to discrimination while reducing the number of tasks required for AD prediction. However, the reduction in examination burden was inferred from the reduction in the number of tasks and was not directly evaluated in terms of task difficulty. Future work should therefore develop and validate objective indices of task difficulty to further evaluate the effectiveness of the proposed task reduction method.
Soma Furukawa, Masaomi Kimura
Open Access
Article
Conference Proceedings
Toward Robust Patient-Level Parkinson’s Disease Classification: Evaluating Stability Across Participant Compositions Using Wearable-Sensor and Questionnaire Information
Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized byheterogeneous motor symptoms that vary across individuals and over time. Wearableinertial sensors provide a low-burden means of capturing motor characteristics, but reliableclassification requires consideration of performance variability across participantcompositions. Building on our previous temporal–frequency integrated deep learningframework, this study evaluated the robustness of patient-level PD classification andexamined whether complementary questionnaire information improves classificationstability. Using the Parkinson’s Disease Smartwatch dataset, participants with PD andhealthy controls (HC) were evaluated by stratified five-fold participant-level cross-validation.Movement signals were classified using a model integrating temporal and wavelet-basedfrequency features, and segment-level probabilities were averaged to obtain patient-levelpredictions. A 30-item non-motor symptom questionnaire was independently classifiedusing logistic regression, and movement-based and questionnaire-based outputs wereintegrated with equal weighting to obtain probabilities of PD. The movement-based modelachieved an accuracy of 0.854±0.029, while the questionnaire-based classifier achieved0.864±0.020. Integration increased accuracy to 0.927±0.027. In particular, Recall(HC)improved from 0.791±0.107 for movement-only classification to 0.896±0.036 afterintegration, indicating reduced fold-level variability, while the change in PD classificationperformance was comparatively modest. Evaluations using alternative participant partitionseeds showed similar overall tendencies without substantial increases in variability. Theseresults suggest that integrating complementary motor and non-motor information canimprove both classification performance and stability across participant compositions,supporting more robust patient-level PD classification.
Ruka Hashimoto, Masaomi Kimura
Open Access
Article
Conference Proceedings
Artificial Intelligence-Assisted Rapid On-site Evaluation: A Human Factors-Informed Patient Safety Framework for Cytopathology
Artificial intelligence (AI) in cytopathology has largely focused on image classification rather than integration into real-time clinical workflows. Rapid on-site evaluation (ROSE), including endobronchial ultrasound-guided transbronchial needle aspiration and endoscopic ultrasound-guided sampling of pancreatic and other lesions, presents a more complex implementation environment. These procedures involve sequential sampling, specimen preparation, adequacy assessment, multidisciplinary communication, and immediate decisions regarding next steps in procedural management. Errors or delays may contribute to unnecessary needle passes, inadequate material, repeat procedures, and potentially delayed diagnoses. Although emerging AI studies demonstrate potential applications in adequacy assessment and malignant cell detection, current ROSE and cytopathology guidance provides limited direction regarding validation, usability, human oversight, or integration into procedural work. To address this gap, we conducted a structured, literature-informed conceptual synthesis of AI-assisted ROSE, procedural cytopathology, digital cytology, and human-factors research. We used this literature to illustrate a purpose-driven approach: define the clinical purpose, determine what would need to be established to judge whether it was achieved, identify the evidence required, and determine what must therefore remain observable before deployment. Three clinical purposes were examined: confirming representative lesional material, preserving and triaging material for molecular or ancillary testing, and determining whether additional sampling is needed. Drawing on these findings, we suggest that the same broad workflow requires different evidence depending on the clinical purpose and that model accuracy alone does not establish safe performance. We therefore recommend treating AI as a component of the procedural cytopathology work system rather than as a stand-alone image classifier. This approach may help move AI-assisted ROSE from isolated image analysis toward safer and more operationally integrated clinical systems.
Bryan Johnston, Jay Kalra, Heather Ruff, Janine Benoit
Open Access
Article
Conference Proceedings
Health Psychology in the Digital Age: Cyberchondria, Health Anxiety, and Online Health Information-Seeking Behaviors
The widespread availability of digital health resources has improved access to health information; however, repetitive and anxiety-driven online searching may be associated with cyberchondria. This cross-sectional correlational study examined the relationships among cyberchondria, health anxiety, and online health information-seeking behaviors in 400 adults who self-reported having both a physical disability and diabetes. Data were collected using a Personal Information Form, the Health Anxiety Inventory, and the Cyberchondria Severity Scale–Short Form (CSS-12). Regular online health searching was reported by 75% of participants, repetitive searching by 40%, trust in online health information by 55%, and searching to feel reassured or reduce worries by 65%. Cyberchondria was moderately and positively associated with health anxiety (r = .48, p < .001). Health anxiety was also positively associated with regular online health searching (r = .41), repetitive searching (r = .44), and reassurance seeking (r = .39; all p < .001). In hierarchical regression analysis, demographic variables explained 8% of the variance in cyberchondria. Adding diabetes duration, perceived health status, and trust in digital health information increased the explained variance to 24%, while the inclusion of health anxiety increased it to 32%. The final model, which additionally included regular online health searching, repetitive searching, and reassurance-seeking behavior, explained 38% of the variance in cyberchondria, R² = .38, F(10, 389) = 23.84, p < .001. These findings support the relevance of a multidimensional health psychology perspective on cyberchondria and suggest the potential value of integrating human factors and health psychology perspectives in the design of digital health environments.
Özlem Özden Tunca, Deniz Sarlak, Ayse Altunkaya Erdogmuş, Özlem Koç Özden, Özlem Melek Mumcuoğlu
Open Access
Article
Conference Proceedings
The Disclosure Paradox: A Human Factor Framework to Medical Error Disclosure
Disclosure of adverse events is essential to managing the consequences of medical error, yet the period immediately following an error may be when healthcare providers feel least prepared to communicate openly. Fear of blame, litigation, reputational harm, and damage to the physician-patient relationship may discourage disclosure. However, failure to disclose can compromise patient autonomy and trust while increasing provider isolation and distress. We propose a quadrant model in which two independent factors shape responses to medical error: transparency of disclosure and support for patients, families, and healthcare providers. Their interaction produces four states: Silence, when both transparency and support are lacking; Insulation, when providers receive support without appropriate disclosure; Exposure, when disclosure occurs without adequate support; and Restoration, when honest disclosure is accompanied by meaningful support. To promote Restoration and reduce recurrence, we introduce the seven-domain “RESTORE” Framework: Recognition of the adverse event and its impact; Ethical, patient-centred disclosure; Support for patients, families, and providers; Transparency about known and uncertain information; Opportunity to restore trust and professional integrity; Reflection on individual and systemic contributing factors; and Evaluation and improvement. A restorative approach keeps patients and families central while recognizing that disclosure without support and support without disclosure are incomplete responses. Integrating transparency, support, reflection, and quality improvement can move healthcare beyond blame and isolation toward healing, learning, and safer, higher-quality patient care.
Zoher Rafid-Hamed, Jay Kalra
Open Access
Article
Conference Proceedings
Analysis of Mechanical Loads on the Handrail During Stair Ascent at the Initiation and Completion Phases
Japan has one of the most rapidly aging populations in the world, with 29.5% of its people aged 65 years or older, making it a super-aging society. Previous biomechanical studies on handrail use have focused mainly on lumbar and joint moments, electromyography, and sensory evaluation. In the present study, we focused on the mechanical loads applied to handrails by users. To our knowledge, no previous studies have investigated the mechanical loads at the initiation and completion of stair climbing. To address this gap in the literature, we measured and analyzed the mechanical loads exerted on a handrail at the initiation and completion of stair climbing. Participants were two healthy adult men (age, 29.0 ± 11.3 years; height, 174.5 ± 0.7 cm; weight, 67.4 ± 14.5 kg). Informed consent was obtained prior to the start of the experiment. The movement in this experiment was stair ascent, following the Japan’s traffic rule of keeping to the left. The starting foot was the right foot. The analysis revealed that the gripping time on the handrail was approximately 1.30 s at the initiation of stair climbing and 1.72 s at the completion of stair climbing. The average maximum grip force was approximately 40.9% of the participants’ maximum voluntary grip force at the initiation of stair climbing, and 15.7% at the completion of stair climbing. In future work, we will increase the number of participants in order to improve data accuracy; acquire data on gripping movements under steady-state conditions; and refine the experimental system.
Runa Narita, Tomoya Hasebe, Daisuke Morioka
Open Access
Article
Conference Proceedings
AI-Assisted Augmented Reality System for Home-Based Upper-Extremity Stroke Rehabilitation
Stroke is a leading cause of long-term disability worldwide, frequently resulting in upper-extremity motor impairments that restrict daily activities. While repetitive, task-oriented rehabilitation is essential for recovery, home therapy faces major hurdles: limited access, transportation barriers, high costs, reduced motivation, and an absence of real-time progress feedback. To address these limitations, this project investigates how artificial intelligence (AI) and augmented reality (AR) can be integrated into home-based upper-extremity stroke rehabilitation.A human-centered design methodology was employed, encompassing a literature review, analysis of current rehabilitation technologies, and consultations with clinical professionals. The research identified key user requirements: enhanced motivation during repetitive exercises, immediate movement feedback, personalized guidance, progress tracking, and seamless continuity between clinical and home care.These findings directly informed ReachGuide, an AI-AR interactive companion designed for home rehabilitation. ReachGuide pairs evidence-based exercises with real-time visual feedback, interactive guidance, and remote connectivity features to bridge patient-clinician communication. Designed as a conceptual framework to complement conventional therapy, this system is intended to support independent exercise and extend rehabilitation beyond clinical settings. As a human-centered design and prototype study, ReachGuide has not yet been evaluated with stroke survivors. Future work will involve empirical usability testing with a stroke-survivor cohort and clinical evaluation to validate system interaction feasibility and therapeutic utility.
Sejal Shah, Ryan Mohammadian, Elham Morshedzadeh, Amin Alipour, Pei-Fen Chang
Open Access
Article
Conference Proceedings
Speech analysis integration for assistance in identification of potential Alzheimer’s Disease early symptoms
The number of Alzheimer's disease cases is increasing worldwide. Diagnosing this disease at its early stages remains a major challenge, particularly in low-resource countries. Early detection is essential to help patients and caregivers prepare for future care needs. Therefore, the development of affordable screening tools is of significant interest. In this study, a preliminary framework based on machine learning and speech analysis is explored as a non-invasive approach for identifying features associated with Alzheimer's disease. The Pitt Corpus from the DementiaBank database was used to extract linguistic and acoustic features. The analysis was conducted using the associated text transcripts. The proposed pipeline involved segmenting the transcripts and analyzing the resulting text data to evaluate temporal speech patterns and discourse structure. Key features, including speech pauses, repetitions, and grammatical complexity, were examined to assess differences between healthy controls and individuals with dementia. Subsequently, a neural network model was trained using the filtered feature set to identify early signs of cognitive decline. Hyperparameters were optimized to improve model performance and efficiency. These results demonstrate that subtle speech alterations associated with Alzheimer's disease can be identified despite the significant class imbalance present in the dataset.
Lili Marlene Camacho Bustamante, Alejandro Díaz Ruiz, Sergio Alberto Navarro Tuch
Open Access
Article
Conference Proceedings
Attention to Medications and Related Information During Pharmacist Verification: An Analysis of Experts and Novices Using AI-Based Gaze Classification
Pharmacists perform prescription verification to assess the appropriateness of prescriptions and medication verification to confirm that dispensed medications match those prescriptions. Understanding how pharmacists gather information during these processes is important for improving verification practices and training. Our previous eye-tracking study showed differences in gaze behavior between experienced and novice pharmacists; however, that analysis focused mainly on gaze behavior within prescriptions and did not capture gaze directed toward other information sources throughout the verification task. In the present study, we reanalyzed eye-tracking data from a simulated verification task to examine gaze toward dispensed medications and related information, including prescriptions and medication bags. Gaze targets were classified using a deep-learning-based image classifier, and relative gaze duration toward medications and related information was evaluated using a log-transformed ratio. Data from 11 pharmacists (five experts and six novices) were analyzed. No significant difference in the log-ratio was found between experts and novices, and the effect size was small. Substantial inter-individual variability was observed, particularly among experts. These findings indicate that the relative gaze time toward medications and related information varies considerably among pharmacists and cannot be explained by professional experience alone. Further investigation of the relationship between these gaze patterns and error-detection performance may help identify effective verification strategies and inform future pharmacist training.
Yoshitaka Maeda, Kota Miyanishi, Kosuke Oiwa, Yukikazu Murakami
Open Access
Article
Conference Proceedings
Acceptance and User Experience of a Serious Game for Independent Home-Based Rehabilitation in Older Adults
The success of home-based digital rehabilitation depends not only on clinical effectiveness but also on older adults’ willingness and ability to use such technologies independently. This study examined older adults’ perceived usability of Vida no Campo (“Country Life”), a serious game for autonomous rehabilitation that uses the Gripwise handheld sensing device, together with acceptance-related responses (motivation, intention to continue or repeat, and perceived autonomy) during an initial session. Sixteen older adults (M = 71 years, SD = 8.0), recruited from a senior university, took part individually at the university, in a familiar, controlled setting; participants’ homes were not evaluated. Data combined unstructured observation with spontaneous verbalisations (a lightweight think-aloud approach) and a post-session questionnaire comprising the UMUX (Usability Metric for User Experience) and complementary items on instructional clarity, ease of learning, confidence, motivation and time perception. Perceived usability was high (mean UMUX score = 86.2), with favourable motivation and intention-to-repeat responses. Half of the participants (50%) perceived time as passing faster than usual and none as passing slowly. In total, 81.3% reported feeling fully capable of using the game independently, and none reported needing supervision. Unstructured observation identified interaction issues across four core usability categories: visual and haptic affordances, instructional content, system pacing, and system logic, suggesting targeted design improvements. The findings provide preliminary evidence of favourable perceived usability and autonomy among a digitally active sample of older adults. They do not establish safe or sustained independent use at home.
Rosane Sampaio, Alexandre Carrança, Joao Oliveira, Joana Andrade, Emanuel Sousa
Open Access
Article
Conference Proceedings
Human and psychological factors and seasickness in transitions at sea
Seasickness is a natural physiological response to motion, yet it remains a significant Human Factors and Ergonomics (HFE) challenge in offshore operations (Fenn & Froholdt, 2025; Ramic & Lützen, 2024). Research suggests that between 34% and 80% of maritime personnel experience seasickness, depending on occupational setting, vessel type, and environmental conditions (Arnardottir et al., 2022; Chan et al., 2006; Fenn & Froholdt, 2025; Le Cloirec et al., 2024). Beyond nausea and vomiting, seasickness can impair cognitive functioning, reduce situational awareness, contribute to fatigue, and negatively affect safety, well-being, and operational performance. Despite these consequences, seasickness is often perceived as a minor or unavoidable aspect of offshore work.This paper disseminates current evidence and practical guidance regarding the acknowledgement, management, and social support of seasickness among offshore workers. The paper further examines a range of contributing risk factors and reviews evidence-informed coping strategies.Recommended approaches include dietary measures such as moderating food intake before voyages. Behavioural strategies include resting or sleeping, maintaining visual focus on the horizon, remaining in the vessel's centre, seeking fresh air, practising mindful breathing, using acupressure bands, and allowing adequate recovery time before and after offshore work. The findings also identify environmental factors to avoid, including strong odours, excessive heat, overheating, heavy clothing, and visually demanding activities such as reading or prolonged use of electronic devices during transit.Importantly, the review highlights the critical role of social support in managing seasickness. Open communication, peer monitoring, practical and emotional support, management engagement, and the normalisation of seasickness through supportive workplace cultures may facilitate coping, particularly among newcomers. These findings suggest that effective seasickness management extends beyond physiological interventions and should also address the social and organisational context in which symptoms occur. By framing seasickness as both a health and systems issue, this review argues for a human-centred approach that integrates individual, social, environmental, and organisational interventions to improve worker well-being, resilience, and safety in offshore operations.This broader perspective aligns with the concept of Performance Influencing Factors (PIFs), which is widely used in high-risk industries such as offshore, maritime, aviation, nuclear, and healthcare to describe the conditions that influence a person's ability to perform tasks safely and effectively. Common PIFs include fatigue, stress, workload, environmental conditions, competence, communication, and physical condition. The findings of this review suggest that seasickness should not only be considered a health condition, but also a PIF in its own right. Moreover, by contributing to fatigue, impaired cognitive performance, stress, reduced communication, and diminished situational awareness, seasickness may also exacerbate several other PIFs, with important implications for individual performance and operational safety.Viewing seasickness through a PIF framework shifts the focus from treating it solely as an individual medical problem to recognising it as a systems issue that interacts with organisational culture, leadership, operational demands, and human performance. This perspective provides a more comprehensive foundation for developing interventions that improve both worker wellbeing and operational safety.ReferencesArnardottir, N. Y., Jonsdottir, S. S., & Petersen, H. (2022). Seasickness among Icelandic seamen. PLOS ONE, 17(8), e0273477. https://doi.org/10.1371/journal.pone.0273477Chan, G., Moochhala, S. M., Zhao, B., Wl, Y., & Wong, J. (2006). A comparison of motion sickness prevalence between seafarers and non-seafarers onboard naval platforms. International Maritime Health, 57(1–4), 56–65.Fenn, A., & Froholdt, L. L. (2025). Social Constraints and Seasickness: Coping Behaviours and Attitudes Among Offshore Windfarm Workers—A Mixed-Methods Study. In Innovations in Sustainable Maritime Technology—IMAM 2025: Ship Operations, Autonomy, Safety, Digitalization, Human Factors, Logistics, Shipyard Organization (pp. 332–349). https://doi.org/10.1007/978-3-032-02102-1_26Le Cloirec, M. J., Lucas, D., Loddé, B., Pougnet, R., Maffert, A., & Jégaden, D. (2024). The prevalence of seasickness in a population of French civilian sailors. International Maritime Health, 75(3), 147–154. https://doi.org/10.5603/imh.99481Ramic, E., & Lützen, M. (2024, December 1). Addressing Seasickness of Technicians on board Working Vessels.
Lisa Loloma Froholdt, Simone Madsen, Anna Udh Bøge
Open Access
Article
Conference Proceedings
A Distributed Local LLM and Centrally Governed RAG Architecture for Privacy-Preserving Cancer Registry Support
Hospital-based cancer registration requires registrars to interpret detailed and frequently updated coding rules, disease-specific manuals, staging classifications, and historical question-and-answer records. Retrieval-augmented generation (RAG) can provide natural-language access to these resources, but cloud-based implementations raise concerns about transmitting clinical information outside the hospital. Moreover, retrieval of the correct document does not guarantee correct application of a registration rule. This paper presents a prototype architecture for deploying a local large language model and vector database within each hospital, while centrally governing and distributing approved reference documents, metadata, retrieval policies, and selected deterministic validation rules. The prototype uses FastAPI, Ollama, ChromaDB, multilingual embeddings, and locally executable language models. Questions are routed to prioritized document groups including the standard registration form, ICD-O materials, staging and site-specific resources, and historical Q&A records. For closed, explicitly enumerated, or high-risk rules, generative processing is supplemented by deterministic safeguards. A formative evaluation involving prostate-specific antigen illustrates the limitation of conventional RAG: although PSA is clinically associated with prostate cancer, it is not one of the four combinations permitted as a site-specific tumor marker for diagnostic-basis coding in the Japanese cancer registry. The case illustrates that semantic plausibility and registry correctness are not necessarily equivalent. The proposed functional allocation assigns natural-language interpretation, retrieval, synthesis, and explanation to the LLM; exact membership checking to deterministic components; and final confirmation to the registrar. This hybrid design is intended to improve privacy, traceability, maintainability, and calibrated human reliance.
Mitsuhiro Takasaki, Kazumi Sasaki, Narue Nakabayashi, Megumi Hori, Tomohiro Matsuda
Open Access
Article
Conference Proceedings
Perceived Walking duration across Street-tree Spatial Patterns: Toward a Route Recommendation System
Walking-route recommendations prioritize distance and travel duration, but streetscape differences can affect the perceived walking duration even on routes with similar travel times. Accounting for this variation could support recommendations tailored to walking experience; however, the relationship between perceived walking duration and streetscape structure remains unclear. In this study, we investigate how tree density and spacing regularity are related to perceived walking duration. We conducted a virtual reality walking experiment with 21 participants in six streetscapes combining three tree density levels with regular or irregular spacing. Coniferous or broadleaf tree shapes were assigned between participants. After each trial, the participants reported the perceived walking duration, and eye-tracking data and first-person HMD-view videos were recorded. After excluding a single outlier, 120 trials from 20 participants were analyzed. Linear mixed-effects models related the duration estimation error, i.e., the difference between the perceived and actual walking duration, to the tree layout conditions. The results showed that walking duration was perceived as longer with increasing tree density. For broadleaf trees, the duration tended to be perceived as shorter under regular spacing than irregular spacing. Correlation and cluster analyses showed that longer mean fixation durations on street trees accompanied longer estimates in the tree-focused and scene-exploration clusters (but not the road-focused cluster). Video features quantified visible streetscape elements. Edge-pixel measures captured density, and ACF- and FFT-based measures captured spacing regularity; however, neither explained the variation in the walking duration estimation error. The findings indicate that walking-route recommendations based on perceived walking duration should consider streetscape structure and user-specific gaze behavior.
Kenta Namiki, Yuichiro Kinoshita
Open Access
Article
Conference Proceedings
Streamlining Remote Care: Computer Access Solutions for Older Adults with Cognitive Impairment
With a growing aging population with cognitive impairment, there is increasing emphasis on developing caregiving interventions that support aging-in-place. In this paper, we focus on a video call feature designed specifically for older adults living alone with cognitive impairment to support connection, remote assistance and intervention. A custom Zoom Workplace Pro application was integrated within a caregiving platform displayed in kiosk mode on a computer within participants' homes. The application was configured for easy connection, screen sharing, and remote control of the users’ screen. Our feasibility pilot study included 2 older adults with mild-to-moderate cognitive impairment and low-to-moderate technology experience. To assess the accessibility of the video call feature, we evaluated the accuracy and time it took participants to complete specific tasks prompted by an examiner (e.g., answer calls, make calls, and share screen). Post testing, participants completed a questionnaire to evaluate Perceived Ease of Use, Perceived Usefulness, and Intention to Use. Regarding results, users completed each task with 100% accuracy within approximately one minute and provided high ratings on Perceived Ease of Use. By eliminating the need to navigate between external applications or web browsers, we observed successful and rapid task completion. Participants identified screensharing as a high-value tool for both technical support and family communication. These findings provide a compelling framework for future caregiving interventions, particularly in rural or remote settings where in-person training and troubleshooting are limited.
Priyanka Koppolu, Amey Gohil, Tejas Patil, Preyash Yadav, Michelle Cohn, Chen-Nee Chuah, Shijia Pan, Sarah Tomaszewski Farias, Alyssa Weakley
Open Access
Article
Conference Proceedings
Embedding Self-Administered Dementia Monitoring into a Remote Caregiving Platform: A Pilot Usability and Data- Sharing Preferences Study
Older adults living alone with cognitive impairment desire tools that support aging-in-place, while remote family caregivers desire awareness of their loved one's cognitive health to inform care. To meet these needs, we embedded Self-Administered Tasks Uncovering Risk of Neurodegeneration (SATURN), a validated self-administered cognitive screener, into Interactive-Care (I-Care), a remote caregiving web-application, to enable monthly cognitive check-ins. A pilot study (Phase 1, n=4) with two care receiver-caregiver dyads evaluated the usability, acceptability, and data-sharing preferences arising from this integration. Care receivers (ages 84-88, mild dementia, MoCA 19-20) completed the assessment with no observed confusion, and both reported high usability and usefulness of monthly check-ins. Both wanted to see results in multiple formats and be informed of score changes. Caregivers reported high usefulness and identified the importance of dynamic consent and shared decision-making for result-sharing, as well as data security concerns including encryption and role-based access. These findings directly informed the design and development of a caregiver-facing SATURN Results Dashboard within I-Care, enabling score visualization, longitudinal trend tracking, per-session contextual notes, selective release of results to care receivers, and PDF/JPEG export. A follow-up dashboard evaluation (Phase 2, n=4 caregivers, n=2 care receivers) revealed high usability, strong support for contextual notes and trend visualization, and nuanced views on result disclosure across caregiver and care-receiver perspectives. Cumulatively, findings demonstrate the feasibility of integrating longitudinal cognitive screening into a remote caregiving platform and surface important human factors design considerations around result display, consent management, and data access control.
Amey Gohil, David Bissig, Hao-Chuan Wang, Priyanka Koppolu, Tejas Patil, Preyash Yadav, Shreyas Aswar, Shijia Pan, Michelle Cohn, Chen-Nee Chuah, Sarah Tomaszewski Farias, Alyssa Weakley
Open Access
Article
Conference Proceedings
Evaluating integrated multimedia communication designed for cognitive accessibility within a remote caregiving platform
Communication applications can impose memory, sequencing, and navigation demands that are difficult for people with cognitive impairment. We conducted a proof-of-concept field evaluation of Interactive-Care (I-Care) Messaging, an integrated text, image, video, and video-call module within a remote caregiving platform. Two family caregiver–care receiver dyads (N = 4) completed structured usability tasks and exploratory post-task questionnaires. The study was designed to expose workflow breakdowns and guide refinement, not to estimate population-level usability. Care receivers completed 11 of 12 tasks independently and one with guidance. Caregivers completed 10 of 12 tasks; both incomplete tasks involved ending an externally launched video call and returning to messaging. Forced-choice responses favored or equated familiar methods, but those comparisons bundled platform familiarity with communication modality, device, and workflow reliability. The findings identify practical promise in the integrated text and visual-media functions and support continued evaluation of I-Care as a complement, rather than a replacement, for familiar communication methods.
Tejas Patil, Priyanka Koppolu, Amey Gohil, Preyash Yadav, Chen-Nee Chuah, Shreyas Aswar, Sarah Tomaszewski Farias, Alyssa Weakley
Open Access
Article
Conference Proceedings
Designing Clinical Research Participation as a Sociotechnical System: The TRIO Human-Centered Design Studio Method
Clinical and translational research recruitment and retention are often treat-ed as isolated operational problems, although participation is produced by interacting people, tasks, technologies, organizational processes, community relationships, and environmental constraints. We conducted a retrospective cross-case analysis of 10 Trial Recruitment Innovation Office (TRIO) Design Studios held in 2021–2022 through Zoom, with Mural as the virtual white-board, to help different clinical trials and health researchers who were facing recruitment and retention problems. Each studio had available data includ-ing presentation from researcher discussing the problem, a post-Studio report and visual artifacts from Mural. Guided by sociotechnical systems and SEIPS concepts, we examined changes in problem representation, human-factors mechanisms, and redesign propositions. Across cases, TRIO Studios shifted at-tention from participant motivation or single recruitment tactics toward par-ticipation work systems. Artifacts externalized distributed knowledge, ex-posed participant and study-team work, expanded system boundaries, re-vealed handoff and feedback failures, and supported reframing. Recommen-dations coordinated changes across roles, workflows, technologies, organiza-tional structures, and community relationships.
Santosh Basapur, Raj Shah
Open Access
Article
Conference Proceedings
Development of a Game Application Using Texture-Related Onomatopoeia with a Focus on Declined Texture Recognition in Dementia
Texture recognition is an important component of everyday cognition, supporting the ability to infer material properties such as softness, roughness, smoothness, and slipperiness from visual information. Previous research has demonstrated that the visual texture recognition ability of patients with dementia is significantly lower than that of healthy individuals across various materials. These findings suggest that changes in texture recognition ability may serve as a useful indicator for the early detection of dementia. Our previous studies have shown that the Sound Symbolic Words Texture Recognition Test (SSWTRT) is a simple cognitive screening method that imposes a low burden on participants and is designed to evaluate cognitive decline using sound-symbolic words (SSWs). The objective of the present study was to develop an SSW-based game application that focuses on the decline in texture recognition ability associated with dementia and supports early screening through enjoyable interaction. We developed a two-phase game. The first phase consists of a falling-object collection task in which fruit items with different visual textures fall from the top of the screen, and players collect the items that match a target texture. The second phase consists of a sorting task in which players select the SSW that best matches each collected fruit from eight alternatives: fuwa-fuwa (fluffy), gotsu-gotsu (rugged), zara-zara (rough), tsuru-tsuru (smooth), sara-sara (dry/silky), sube-sube (sleek), deko-boko (uneven), and nuru-nuru (slimy). A scoring method was designed to evaluate responses in the sorting phase not simply as correct or incorrect, but according to the empirical distribution of SSW choices obtained from young, healthy participants in previous studies.
Shunsuke Miyauchi, Masaki Sakanaka, Yuji Nozaki, Madoka Nakajima, Chihiro Kamohara, Chiyoko Kameue, Koichi Hirota, Koichiro Iwami, Maki Sakamoto
Open Access
Article
Conference Proceedings
Interactive Mattress Balancing Body Pressure Distribution and Maintaining Sleeping Posture: Optimized Control of Variable Spring Constant
This study addresses the support of the recumbent human body by simultaneously considering posture maintenance and the redistribution of interface pressure, with prolonged recumbency and pressure ulcer prevention as the target application. Prior work has focused on reducing contact pressure without incorporating body alignment into the optimization. The body is modeled as a discrete set of support points, each carried by an independent spring and each associated with a contact area, and the problem is posed as a multi-objective optimization combining shape preservation and pressure redistribution. A closed-form optimum is obtained in which the stiffness at each point is proportional to its contact area and inversely proportional to its required displacement; equalizing reaction forces, the form usually assumed, is recovered only when the contact areas are equal. The two objectives are therefore not inherently conflicting. Introducing the stiffness range attainable with conventional pocket springs, however, shows that the cervical and popliteal regions fall short of the required stiffness by factors of 1.1 and 1.4, so that the exact optimum is not realizable in a passive structure and a genuine trade-off reappears. Variable stiffness, and hence active support integrating sensing and actuation, is required to track the optimum. This work is a feasibility study establishing the analytical framework rather than a validated physiological model.
Jun Ito, Shin Usuki
Open Access
Article
Conference Proceedings
Development and Evaluation of Applications for Oral and Cognitive Rehabilitation: A Facial Rock–Paper–Scissors Program with Psychological Effects
We developed a “Facial rock-paper-scissors” program for oral and cognitive rehabilitation by incorporating mouth and cheek movements, and vocalizations used in swallowing training into the facial expressions and vocalizations of “rock,” “paper,” and “scissors.” In basic research, we investigated psychological evaluations when multiple programs with different facial-expression setting rules were implemented. The results showed improved mood when using a program in which the “scissors” expression involved stretching the mouth horizontally and raising the corners of the mouth to form a smiling face. Based on these results, we established a facial-expression setting for the program expected to produce positive psychological effects. Next, we developed two applications with different difficulty levels that allow users to easily perform the program. Both applications consist of a practice session for experiencing facial expressions and vocalizations, and a practical session. In the practical sessions, the system uses the tablet’s camera to recognize the user’s facial expressions and enables judgments of “win,” “lose,” and “draw,” thereby encouraging users’ motivation and promoting continuous engagement in rehabilitation. Practical evaluations of the applications conducted with a community-dwelling older adult suggested that implementing the program may have positive psychological effects such as increased vitality, pleasure, and arousal, may have cognitive function effects such as improved concentration and attention, and may contribute to improvements in orbicularis oris muscle strength. The developed applications were suggested to provide an engaging experience that contributes to users’ physical and mental functions.
Kiyomi Yoshioka
Open Access
Article
Conference Proceedings
When the model card is the hazard: a silent preprocessing failure in a CT foundation model and a label-free check for trustworthy deployment
Clinical teams increasingly build decision support on frozen foundation models (FMs), preparing inputs as the public model card shows. We report a case in which the documentation itself was the hazard. A linear probe on CT-FM embeddings was trained to predict COVID-19 outcome in 1,205 patients and applied unchanged to 309 patients at another hospital. Following the model card, external mortality AUC (area under the receiver operating characteristic curve) was 0.64; restoring the resampling used in pretraining, documented only in the training code, gave 0.71 (0.73 at 2 mm; Holm P = .10). In 319 examinations reconstructed both thin and thick, the model-card pipeline raised predicted risk by 1.76 standard deviations (SD) and scrambled patient ranking (Spearman 0.34): a thicker slice looked like a sicker patient. The corrected pipeline was stable (−0.02 SD; 0.96). The failure was reproduced by removing resampling from two other FMs on 1,110 CTs from a third cohort, and model-side repairs did not fully fix it. A Korean cohort (255 patients) analysed under a prospectively specified plan added kernel-driven instability and, post hoc, inflated predicted risk. We frame the omission as a latent condition that is invisible to clinicians and unequal across sites, and propose a three-step pre-deployment check that clinical teams can run: reproduce the pretraining preprocessing, test stability on paired reconstructions without labels, and recalibrate on about ten local cases.
Woo Hyun Shim, Chae Ri Park, Sang Cheol Kim, Sukyoung An
Open Access
Article
Conference Proceedings
System Safety Analysis Considering Human Factors in Aircraft Development
In conventional system safety analysis approaches, human factors were often overlooked by aircraft designers, as traditional methodologies such Preliminary System Safety Analysis (PSSA) and Functional Hazard Analysis (FHA) primarily focused on system-level failures. With advancements in human factors and ergonomics research, the aviation industry has increasingly recognized the critical role of human elements in safety outcomes. Recent aviation accident statistics indicate that human factors contribute to approximately 70% of incidents, serving as fundamental and complex causal factors. Classic theoretical frameworks such as the Swiss Cheese Model and the REASON model were initially proposed to elucidate the mechanisms of human error occurrence. Subsequent systematic classification systems, including the Skill-Rule-Knowledge model and the Human Factors Analysis and Classification System (HFACS), were developed to standardize human error categorization. Typical pilot performance errors encompass attention allocation failure, cognitive overload, and loss of situation awareness. To quantify the impact of human errors on safety, this study proposes a novel risk assessment methodology integrating human factors into system safety analysis. This approach, termed Human Safety Assessment, enables the identification of potential human error modes and the formulation of mitigation strategies. It is recommended for implementation during the early stages of aircraft development to detect latent human-factor risks and design deficiencies. A human-centered design philosophy is adopted to analyze error propagation through cockpit human-machine interfaces under specific flight scenarios. Furthermore, Human Safety Assessment must be fully integrated with system safety requirements analysis and human factors engineering within the aircraft development lifecycle.
Wu Xu, Shuai ma, Wanchen jia, Lin ding, Chongchong miao, Ruiming zhang
Open Access
Article
Conference Proceedings
Designing sustainable business models for industrial service providers – concepts and challenges
Sustainability is becoming increasingly important for companies. However, there is a discrepancy between symbolic and genuine sustainability implementation. A key aspect of how genuine sustainability can be achieved in companies is the organizational embedding of sustainability – that is, the extent to which there are clear responsibilities for sustainability within the corporate structure. Our own empirical analysis demonstrates precisely this relationship: Clear responsibilities are associated with a higher level of green service maturity. This raises the question of how sustainability can be systematically embedded in companies. The literature on sustainable business models provides approaches to this end. Building on this, we further develop our own framework for designing business models and supplement the five building blocks of success it contains with sustainability dimensions.
Michaela Friedrich, Christian Schiller
Open Access
Article
Conference Proceedings
Runway Incursions at U.S. Airports, 2017-2026: How Airfield Geometry, Operations, and Human Factors Shape Safety
Runway incursions are the most closely watched measure of surface safety at U.S. airports, and the serious close calls of 2023 put them back under scrutiny. Federal Aviation Administration (FAA) runway incursion totals by type, month and region for fiscal years 2017 through 2026, are divided by FAA Operations Network airport operations and tested for trend and seasonality with Poisson regression. The coded fields and keyword-coded narratives of 1,528 NASA Aviation Safety Reporting System (ASRS) runway incursion reports from 2015 through 2025 are examined separately. Incursions fell from 34.7 per million operations in FY2017 to 28.6 in FY2025, an overall decline of approximately 2.0% per year. Pilot deviations showed a statistically significant decline and accounted for approximately 85% of the overall rate reduction. No significant trend was detected for controller operational incidents. Incursions per operation are about 16% higher in June than in January. Situational awareness (70%), communication breakdown (58%) and confusion (35%) dominate the coded factors. Controller reports more frequently include distraction and workload codes than pilot reports. Hold-short instructions and readbacks, signs, markings and lighting, construction and expectation bias recur in the narratives, and the airport is a contributing factor in 36% of reports. The results show where airfield geometry changes, phraseology discipline, vehicle-operator training and surface-detection technology are most likely to pay off.
Jules Roa, Lilian Annan
Open Access
Article
Conference Proceedings
Human Factors and Propulsion Failures in U.S. General Aviation Accidents: What a Decade of NTSB Findings Implies for Electric Training Aircraft
Battery-electric training aircraft are entering U.S. flight schools without the fuel, induction, ignition, exhaust and lubrication systems used in piston aircraft. The extent to which this architecture eliminates, changes or introduces accident mechanisms has not been assessed against recent accident data. Every U.S. 14 CFR Part 91 accident from 2015 through 2025 in the National Transportation Safety Board (NTSB) database (12,052 accidents, 2,038 fatal) was examined, counts were normalized by FAA survey hours, and the 10,970 accidents with coded findings for 2015–2024 were classified by human-factors and propulsion-related finding groups. The accident rate fell from 5.47 to 4.13 per 100,000 hours and the fatal-accident rate from 1.04 to 0.64. Poisson trends give declines of about 3% and 4% per year. Personnel findings appear in 79% of accidents and pilot aircraft-control findings in 53%, rising to 60% on instructional flights. A propulsion or fuel finding is present in 22% of accidents. An elimination-and-substitution screen shows that 8.3% of accidents carry a finding naming a component absent from an electric powertrain. A further 11.4% involve components that persist in electric form. Fuel exhaustion and starvation have energy-management counterparts, while internal engine failure has counterparts in motor and inverter failure, at rates not estimated here. The other four accidents in five, and most fatal accidents, involve aircraft control, decision making, weather and terrain, and they require continued attention to human factors regardless of propulsion type.
Jules Roa, Mehdy Emad Araim
Open Access
Article
Conference Proceedings
Operationally Conditioned Transit Injury Risk: A National Machine Learning Analysis of Safety, Mobility, and Socioeconomic Exposure.
Transit injury risk may vary with operating conditions and community context, but the usefulness of these factors for safety screening must be assessed against agencies' prior injury experience. We examined whether operations, socioeconomic conditions and ambient mobility improve injury prediction beyond ridership exposure and history, while distinguishing prediction from moderation of associations. National agency-mode-month records linked transit safety and service data to headquarters-county context. The broad sample comprised 84,721 observations from 2019–2025. Poisson regression, negative binomial regression and gradient-boosted trees used identical samples within each comparison, with 2019–2023 training, 2024 validation, retrospective evaluation in January–June 2025 and a reserved July–December 2025 evaluation after model settings were fixed. History reduced validation Poisson deviance by 37.5% beyond exposure, mode and calendar controls. A smoothed historical-rate benchmark remained competitive, with close calibration of reserved-period national totals but variation across subgroups. Socioeconomic variables improved boosted validation deviance by 0.0219; gains in both 2025 evaluations were positive but uncertain (0.0279 and 0.0177; 95% confidence intervals spanning zero). Headquarters-county mobility showed no reliable incremental improvement in either available period; earlier exploratory tests did not establish mobility–vulnerability moderation. A secondary bus analysis found no clear difference between collision and non-collision service-hour associations: the ratio of rate ratios for 10% greater hours was 0.985 (95% confidence interval 0.936–1.035). These findings support historical benchmarking and subgroup calibration before local safety review. They neither identify causal service effects nor establish the absence of mobility-related vulnerability
Timothy Kangogo, Oladotun Adeola, Barbara Asamoah, Anthony Dontoh, Judith Mwakalonge, Gurcan Comert, Varghese Vaidyan, Vijay Bendigeri
Open Access
Article
Conference Proceedings
Human-Centered Creative Support for VE Idea Realization through Contradiction Analysis
In product development, value Engineering (VE) has been widely adopted in manufacturing as a standardized framework based on function-oriented thinking approach to meet increasingly diverse and sophisticated customer demands. However, the assessment of the feasibility of ideas generated through VE and the resolution of design conflicts (contradictions) still tend to depend heavily on the tacit knowledge of experienced designers. To address this challenge, this study proposes a systematic method that utilizes the concept of physical contradictions from TRIZ (a Russian acronym meaning “Theory of Inventive Problem Solving”) into the development phase of VE in order to support the realization of innovative ideas that often face obstacles due to conflicting requirements. A case study was conducted on the liquid end parts of a metering pump manufactured by TACMINA CORPORATION. As a result, concrete solutions based on TRIZ inventive principles were derived for several improvement proposals that had previously been considered infeasible due to contradictions when relying solely on conventional VE activities.
Masataka Kanzaki, Akagi Koichi, Hikaru Takashima, Manabu Sawaguchi
Open Access
Article
Conference Proceedings
Integrating ergonomics, lean manufacturing, and work study to improve overall system performance within Industry 5.0
This paper develops a conceptual framework for integrating ergonomics, lean manufacturing, and work-study methods within an Industry 5.0-oriented work system. Although these approaches address related aspects of work-system performance, they commonly use different measures, analytical perspectives, and improvement routines. The resulting separation can make it difficult to evaluate productivity, operational flow, and worker-related outcomes simultaneously. The proposed framework addresses this integration problem by defining four operational integration points: process observation, value-stream mapping, standard-work documentation, and the Plan-Do-Check-Act (PDCA) cycle. At each point, ergonomic exposure, lean flow and waste information, and work-study method and time measurements are retained as distinct evidence types and incorporated into a shared improvement output. The resulting framework comprises a unified observation record, an exposure-and-flow-augmented value-stream map, an integrated standard-work sheet, and a joint improvement log with multidimensional acceptance criteria. The paper does not claim empirical effectiveness; rather, it establishes a conceptual architecture and specifies measurable outcomes for future field validation. The proposed validation should assess productivity, quality, ergonomic exposure, and worker-related outcomes before and after implementation and compare the results with conventional improvement practices. This provides a basis for testing whether integrated improvement can achieve operational gains without transferring costs to worker health or workload.
Erman Çakıt
Open Access
Article
Conference Proceedings
Hindrances to understanding and identifying hazards in the workplace: A Systematic Literature Review
Industries and various sectors of the economy continue to record workplace incidents. Understanding and identifying workplace hazards is crucial for ensuring employee safety and health. However, various hindrances can impede effective hazard recognition and risk assessment. This systematic literature review weaves recent research to identify the factors that contribute to these barriers. The central purpose of this review is to systematically examine journal articles from the past ten years that examines the difficulties encountered by both employees and organizations in identifying and comprehending workplace hazards, and to offer a thorough summary of the key issues involved. The review identified several key barriers to hazard identification, including cognitive biases, insufficient training and education, organizational culture, and inadequate communication of risks. Personal factors such as risk perception and prior experiences significantly influence hazard recognition. There is an opportunity to improve standardized procedures and tools for hazard assessment, which its insufficiency often results in underreporting of potential risks. This systematic literature review highlights the multifaceted challenges in understanding and identifying workplace hazards. It examines the findings of studies that focused on the challenges employees and management face in understanding and identifying workplace hazards. Addressing these hindrances is critical to effective workplace incident prevention. This will require better training, a stronger safety culture, improved communication, and standardized hazard assessment. Future research should evaluate the effectiveness of these strategies. The review highlights key themes, identifies gaps in occupational health and safety literature, and offers recommendations to strengthen safety protocols.
Olumide Adewumi, Jeremy Gouzd
Open Access
Article
Conference Proceedings
A replicable two-phase field protocol for assessing fatigue risk in aircraft line maintenance: design and operational applicability
Fatigue is a recognised contributor to error in aircraft maintenance, yet operators lack a practical, standardised way to assess it, and fatigue research has concentrated on flight crew rather than on maintenance personnel. This paper presents the design of a replicable two-phase field protocol for assessing fatigue risk in aircraft line maintenance and sets out the operating conditions under which it is most valuable. The protocol was developed in an Omani line-maintenance context and is designed to transfer to other operators; this paper presents the method, not country-specific data. Phase 1 is an environmental field-measurement procedure: twenty-one pre-flight inspection checkpoints on the Airbus A320, drawn from the Aircraft Maintenance Manual and grouped into three zones, are assessed for sound pressure level and illuminance using calibrated instruments against recognised occupational limits. Phase 2 is a structured self-report survey pairing two validated instruments, the Epworth Sleepiness Scale and the Fatigue Severity Scale, with exposure items across four domains (roster, sleep/wake, environmental and psychosocial), scored into a fatigue-likelihood index. The protocol is low-cost, tool-light and reproducible by an operator's own safety team. We describe the instruments, checkpoint mapping, scoring logic and data-handling steps, and set out where the approach applies best: hot-climate, high-daylight line operations implementing safety management systems. The protocol has since been executed in full across ten aircraft and 236 maintenance workers, demonstrating operator-scale feasibility; the empirical results are reported separately in companion manuscripts. It offers maintenance organisations a first step toward a fatigue risk management system for line maintenance.
Mundhir Nasser Ali Al Alawi, Ashish Kumar Singh, Noraisikin Sabani, Jundika Candra Kurnia
Open Access
Article
Conference Proceedings
Improving Sitting Posture with Back-Pressure Stimulation from a Carbon-Fiber-Reinforced Plastic Chair
A carbon-fiber-reinforced plastic (CFRP) lumbar-support chair was developed to provide localized back pressure as a proposed sensorimotor stimulus. Two separate pilot studies evaluated pressure delivery and external sitting geometry. In Study 1, five male students tested two commercial chairs and two CFRP prototypes in individually randomized order. Active-cell counts (pressure > 5 g/cm²) were lower for the prototypes (121.2–127.8) than for the commercial chairs (246.6–249.6); a Greenhouse–Geisser-corrected repeated-measures ANOVA indicated an overall chair effect, F(1.318, 5.271) = 39.50, p = 0.000957. These findings show that the CFRP prototypes apply pressure to a smaller, targeted area of the back. Study 2 analyzed eight selected Office–CFRP scan pairs from an office-work project targeting end-of-workday posture. Texture-free overlays and an 80 × 50 mm collar-referenced upper-back region were evaluated. Mean local inclination was 31.60° for Office and 26.98° for CFRP; six participants showed a smaller angle with CFRP. The mean difference was −4.62° (95% CI −10.07 to 0.83), t(7) = −2.003, p = 0.085; Cohen’s dz = −0.708 and Hedges’ gz = −0.629. Exact sign-flip analysis yielded p = 0.102. Although not statistically significant, the data showed a numerical tendency toward a more vertical local upper-back surface with CFRP. The findings support staged evaluation, not an established muscular mechanism or reduced spinal loading. Small samples, clothing, retrospective scan selection, and separate-day assessment limit interpretation.
Shigekazu Ishihara, Hiroyoshi Ohara
Open Access
Article
Conference Proceedings
A Cryptoagility-Driven Architectural Model for Post-Quantum Cryptography Migration in Edge and Access Networks
The migration to Post-Quantum Cryptography (PQC) represents a systemic transformation of cybersecurity architectures, requiring not only algorithm replacement but also continuous cryptographic lifecycle governance across heterogeneous infrastructures. While standardization of PQC algorithms is progressing, organizations face significant challenges in deploying quantum-safe security across heterogeneous edge and access infrastructures characterized by constrained resources, long device lifecycles, and multi-vendor ecosystems. This paper presents a cryptoagility-driven model for PQC migration that supports controlled adaptation of cryptographic mechanisms while preserving operational continuity across heterogeneous infrastructures. The proposed model integrates continuous cryptographic inventory, centralized crypto-policy governance, algorithm abstraction, modular cryptographic services, hybrid classical-PQC operation, communication security, interoperability support, lifecycle management, and long-term data protection. Cryptographic agility is incorporated as a transversal architectural capability spanning governance, lifecycle management, communication security, and trust management functions, enabling runtime adaptation to evolving standards, emerging vulnerabilities, and changing operational requirements. The model addresses both technical and organizational aspects of migration while supporting scalable deployment across distributed edge and access environments. By combining layered architectural principles with dynamic cryptographic governance and lifecycle management, the proposed model provides a practical blueprint for achieving sustainable quantum-safe transformation in complex infrastructures.
Reijo Savola
Open Access
Article
Conference Proceedings
Comparison of Horizontal Moment Arm Measurements for the Revised NIOSH Lifting Equation and the Lifting Fatigue Failure Tool
The Revised NIOSH Lifting Equation (RNLE) and Lifting Fatigue Failure Tool (LiFFT) manual handling risk assessment methods require measuring horizontal load moment-arms, but from different body landmarks. Since it is unclear if these different measurement methods are interchangeable the research objectives were to determine if differences exist between RNLE and LiFFT moment-arms and determine the impact of moment-arm differences on assessment method risk interpretations. Twenty subjects lifted loads from ten vertical (floor-to-mid shin, mid shin-to-knuckle, knuckle-to-shoulder, shoulder-and-above) and horizontal (close: 30cm; intermediate: 55cm; far: 75cm) zone combination locations. Moment-arms were quantified by a motion capture system by determining the horizontal distance between reflective markers placed on the third knuckle and markers placed on the hips (LiFFT method) and ankles (RNLE method). ANOVA techniques assessed moment-arm differences within each vertical zone, and risk interpretations (low-, moderate-, high-risk) were assessed using both RNLE and LiFFT moment-arms in each assessment method. RNLE moment-arms were larger than LiFFT moment-arms in the shoulder-and-above and knuckle-to-shoulder vertical zones. LiFFT moment-arms were larger than RNLE moment arms in the mid shin-to-knuckle and floor-to-mid shin vertical zones. When LiFFT moment-arms were input into RNLE and LiFFT assessment methods there was no difference in resulting risk categories. When RNLE moment-arms were input into RNLE and LiFFT methods there was no difference in resulting risk categories. For sagittally-symmetric lifting it appears RNLE and LiFFT moment-arm measurements may be used interchangeably reducing multiple measurements.
Michael Jorgensen, Daniel Chikwendu, Nils Hakansson, Juan Hernandez, Hai Nguyen
Open Access
Article
Conference Proceedings
An Institutional Federated Readiness Assessment Framework for Multidisciplinary Research Organisations
Research organisations increasingly need to access and use data held across institutions and research infrastructures. Federated research environments support this by allowing data and resources to be accessed across organisational boundaries while remaining under the responsibility of the organisations that hold them. In Europe, these environments have developed around different scientific communities, types of data and purposes. A multidisciplinary research organisation may therefore need to consider different federated environments across its research activities, together with their requirements for data management, interoperability, governance, security and access. Assessing readiness for a single environment may give only a partial view of what is needed across the organisation. This study presents the Institutional Federated Readiness Assessment Framework (IFRAF) to address this problem. We reviewed scientific literature, standards and documentation from European federated research environments and compared the requirements identified from these sources. The analysis resulted in six areas of institutional capability. It also showed that some requirements are common across several environments, while others depend on the data, the environment and the role of the organisation. Systems engineering principles were used to organise the requirements and maintain traceability to their sources. We initially applied IFRAF at the Department of Chemistry and Biotechnology (KBI) at Tallinn University of Technology by mapping its scientific and data domains to potentially relevant European federated environments. The mapping showed that different research areas within the same organisation may be associated with different federated environments. The next stage of the work covers the evidence-based capability assessment and gap analysis, followed by further validation of the framework.
Jayden Khakurel, Pirjo Spuul
Open Access
Article
Conference Proceedings
Human-Centred Transition Pathways for AI-Enabled Safety-Critical Systems: An Industrial Systems Engineering Perspective
Artificial intelligence is transforming safety-critical industries from technology-centred automation towards integrated human-AI work systems. However, many organisations continue to implement AI through isolated technical projects rather than coordinated socio-technical transformation. This paper proposes a Human-Centred Transition Pathway that combines systems engineering, Human Factors Integration, Safety-II, STAMP and Human-AI Teaming into a structured implementation framework. The proposed methodology aligns operational objectives, competency requirements, organisational processes and AI capabilities throughout the system lifecycle. Aviation provides the principal case study through recent developments in EASA AI 3.0 and competency-based operations, while lessons are transferable to rail, maritime and healthcare systems. The framework demonstrates how industrial organisations can move beyond regulatory compliance towards resilient, human-centred AI ecosystems that support sustainable operational performance.
Dimitrios Ziakkas, Debra Henneberry
Open Access
Article
Conference Proceedings
Detecting Silent Automation Reliability Degradation - A Signal Detection Theory Analysis
Silent reliability degradation provides no explicit alert, requiring operators to distinguish a genuine system-state change from ordinary stochastic errors. Using the previously reported first-report dataset (N = 60; 720 participant-scenario trajectories), this secondary signal detection theory (SDT) analysis formally decomposed terminal failure-report behavior into discrimination sensitivity (d-prime) and relative decision criterion (c). Reliability losses of 5% and 10% were treated as signal-present states, 0% loss as signal absent, and any first failure report within 100 formal monitoring trials as a positive response. Larger reliability loss was associated with higher d-prime. Higher initial reliability was associated with both higher d-prime and a more conservative c. Thus, the slower yet more accurate pattern previously observed under high initial reliability is consistent with the joint contribution of better terminal state discrimination and more conservative reporting. This analysis quantitatively complements prior time/accuracy and psychometric-threshold analyses of the same dataset.
Cheng Tan, Liang Ma
Open Access
Article
Conference Proceedings
Experimental investigation of damping and mass parameters of an adjustable physical hand-arm model for reproducing human mechanical impedance
Hand-arm models enable reproducible investigation of physically coupled human-machine systems. They can substitute for the human user in experimental investigations by mechanically representing the human hand-arm system. This enables the dynamic interaction between the hand-arm system and an actual machine to be studied under controlled, reproducible, and systematically variable conditions. Adjustable hand-arm models also make it possible to represent different anthropometric characteristics and force combinations. Many existing models are either nonadjustable or show deviations from mechanical impedance responses measured in humans, particularly in the frequency range from 350 to 500 Hz. The adjustable hand-arm model investigated here is implemented as a three-mass oscillator. To address the observed deviations, two studies were conducted using the adjustable hand-arm model. In the first study, the damping behavior of the three-mass oscillator was systematically investigated. In the second study, the mass parameters of the adjustable three-mass hand-arm model were systematically varied and adapted to the measured impedance responses of the human hand-arm system. Based on these investigations, an enhanced adjustable physical hand-arm model with adapted damping and mass parameters was developed to achieve closer agreement with the measured mechanical impedance of the human hand-arm system over the investigated frequency range from 10 to 500 Hz. The new hand-arm model provides an adjustable physical representation whose damping and mass parameters are specifically adapted to the measured mechanical impedance of the human hand-arm system over the frequency range from 10 to 500 Hz. The investigated parameterization therefore, represents a frequency-dependent trade-off rather than a uniform improvement over the entire frequency range.
Simon Saurbier, Johannes Roggenstein, Moritz Engelmann, Lukas Bunk, Dieter Krause, Sven Matthiesen
Open Access
Article
Conference Proceedings
Predicting Ergonomic Risk and Worker Performance Using Neural Networks with EMG and Kinematic Data
Ergonomic performance is fundamentally governed by the interaction between postural configuration and muscle activation; however, frontier based methodologies such as Data Envelopment Analysis (DEA) yield relative efficiency scores rather than predictive estimates. This study investigates the extent to which supervised neural networks (NNs) can generate predictive estimates of DEA derived ergonomic efficiency using electromyography (EMG) and kinematic inputs. In this study, ten participants completed five sweeping task trials with bilateral EMG on biceps brachii and flexor carpi radialis, and elbow angles captured via motion tracking. Trial features were aggregated into six inputs per participant to compact feedforward NNs trained for regression of left/right DEA efficiencies and classification of efficient frontier participants. Leave one out cross validation and probability calibration were applied. As results, left side regression achieved R² = 0.57 and right side performance was weaker with R² = 0.11. The classifier reached 80% accuracy with high distinguishing power of efficient frontiers about 75% (F1 accuracy) and high ability to rank-order efficient vs. inefficient frontiers about 92% (AUC accuracy). Key predictors included left elbow angle and bilateral biceps activity. Two NN–DEA disagreements reflected differences between normative and predictive modeling. As a conclusion, neural networks accurately predicted ergonomic efficiency and frontier membership from the same EMG/kinematic data used in DEA, demonstrating their value for prospective, real time screening and future extension to sequence-based models.
Sohyung Cho
Open Access
Article
Conference Proceedings
The Process of Information Integration in Children's Spatial Cognition: An Analysis of Newly Acquired Knowledge and Mapping Rates via Fieldwork
In Japan, children may need to evacuate independently when natural disasters occur while adult support is unavailable. Effective evacuation may depend in part on children's internal representations of the environment, or cognitive maps. This study investigated the development of children's cognitive maps through fieldwork, focusing on qualitative changes from route-map to survey-map representations and the factors associated with individual differences in spatial learning. Two fieldwork sessions were conducted with 94 fifth-grade students in Tokyo as part of disaster prevention education. Students drew hand-drawn maps of an assigned area before and after the fieldwork and completed a landmark recall list after the second session. Sixty-four valid responses were quantitatively analyzed by comparing students who changed from route-map to survey-map representations (Group A) with those who remained at the route-map stage (Group B). The results indicated that the two groups recalled a similar number of landmarks acquired during the fieldwork. However, Group B represented fewer of these landmarks on their post-fieldwork maps and showed a high concentration of landmarks along a single road. The average mapping rate was 26.8% in Group B, compared with 76.0% in Group A. Group A also expanded the number of roads represented on their maps and distributed landmarks across multiple interconnected roads and intersections. These findings suggest that the qualitative development of children's mental maps may depend not only on the amount of information recognized during fieldwork but also on how that information is spatially integrated within a network of roads and intersections. For disaster prevention education, it may therefore be useful to encourage children to use landmarks as spatial anchors for connecting multiple roads and intersections.
Makoto Oka, Hirohimo Mori
Open Access
Article
Conference Proceedings
Use Case Analysis for Designing an Integrated Monitoring Center for Nuclear Power Plants
An integrated monitoring center (IMC) is a system that remotely monitors systems or facilities at separate sites. It brings site information together so operators can view conditions from one location. For nuclear power plants (NPPs), an IMC receives operating and safety information from several plants and provides a common view of plant conditions. NPPs have many connected systems, strict safety limits, and large amounts of operating data. When a problem occurs, the center helps operators understand the situation, share information, and respond quickly. The purpose of this study is to design use cases for an IMC for NPPs. We designed the use cases based on four goals of the center. First, the center should show the operating conditions of several plants in a common view. Second, it should help users identify early signs of problems. Third, it should help users share information, track response actions, and prepare reports during emergencies or disasters. Fourth, it should manage the data, records, AI models, and security needed for these activities. We then defined 13 use cases based on these goals. To satisfy the four goals of an IMC, we defined 51 top-level functions for the 13 use cases. These functions specify the support required for each use case and connect monitoring, review, response, and reporting. The output of one function can be used as the input to the next. Therefore, the proposed use cases can guide the definition of system functions for future NPP IMCs.
Junhyeong Bang, Jonghyun Kim
Open Access
Article
Conference Proceedings
Designing an Experimental Environment for Training Stakeholder Orchestration by the Non-Expert One-Person SIRT in Manufacturing SMEs
A cyber incident that stops a small or medium-sized enterprise (SME) in a Japanese manufacturing supply chain can stop its customers' lines within days. In many such firms the response falls to one non-technical employee, who cannot act on contact rules and decision criteria under pressure even when these exist. We call this person the one-person SIRT. This position paper describes the socio-technical design of an experimental environment for drilling that person. The competency we target is orchestration, a program-management term for aligning autonomous programs by presenting and selecting the criteria under which each judges its options, not by command. In incident response it means aligning several stakeholders' conflicting criteria into one defensible course of action under time pressure, which the one-person SIRT must do alone. Since no colleague is available, the drill gathers three to five one-person SIRTs from different firms into one team, a scaffold in which reasoning normally done alone is spoken and compared. AI stakeholder personas stand in for the missing departments, each holding private facts derived from a functional model of the firm, an affective profile, and a bounded authority to act on instructions. A state manager fires the events that change the situation. Reflection combines automated extraction, an expert retrieval system, and a human facilitator. The primary outcome is transfer to the individual, measured with a solo post-test on a new scenario. We present the design, its rationale, and a narrowed verification plan.
Tomoyuki Kato, Isao Kato
Open Access
Article
Conference Proceedings
Cyber Resilience Act Reporting Obligations in Vulnerability Handling: An IDEF0 Analysis
Article 14 of the Cyber Resilience Act (CRA) requires manufacturers to report actively exploited vulnerabilities once they become aware of them, even while investigation and remediation development continue. We examine what information CRA reporting requires from existing vulnerability handling activities and which internal responsibilities manufacturers must assign. Using IDEF0, we map selected CRA provisions onto the six activities of ISO/IEC 30111, with ISO/IEC 29147 as a complementary reference. FIRST’s PSIRT Services Framework v1.1 and IEC 62443-4-1:2018 inform the role analysis. Reporting may be required before remediation is complete, while subsequent reports require updates from development and release. The examined documents describe technical and communication roles but do not assign CRA reporting and user-information tasks to specific internal roles. This study identifies these information dependencies and the responsibilities that manufacturers must assign internally. Their operational feasibility remains to be evaluated in organizations.
Jumpei Tahara, Kenji Watanabe
Open Access
Article
Conference Proceedings
A Cross-Functional Value Engineering Approach to Transforming Existing Factories into Advanced Factories
In recent years, the labor force in Japan’s manufacturing industry has been declining significantly, increasing the importance of reforming existing factory infrastructure and operations. To address this challenge, this paper proposes value engineering (VE) activities that enable the transformation into advanced factories that promote automation and labor-saving measures while maximizing the use of existing infrastructure. Specifically, based on a comprehensive analysis of the target factory, issues are identified and a project team including stakeholders is established. Additionally, a function analysis incorporating Industrial Engineering (IE) principles is conducted by all team members to develop and implement concrete improvement plans, considering the constraints of existing infrastructure. This paper introduces this approach.
Akio Tsuchiuchi, Kengo Takezawa, Ryosuke Takahashi, Shinji Amano, Takashi Kido
Open Access
Article
Conference Proceedings
A Real-Time Edge-Based Multimodal Safety Assistance System for Forklift Operations
Forklifts are essential to industrial and warehouse operations, but their use in confined and dynamic environments creates safety risks associated with restricted visibility, unsafe maneuvering, delayed operator response, and improper load handling. Conventional manual safety checks and road-vehicle telematics are not sufficient for forklift environments, which involve short-range interactions, narrow aisles, reversing Maneuvers, and continuously changing operating conditions. Existing forklift safety systems have demonstrated the usefulness of cameras, radar, LiDAR, and wireless identification for visibility improvement and collision avoidance. However, these systems commonly focus on individual sensing functions and provide limited integration of surrounding hazards, operator attention, and vehicle state.To address this limitation, this study proposes a real-time edge-based multimodal safety assistance system for forklift operations. The system integrates four-side LiDAR sensing, operator viewing-direction and attention information from an in-cabin monitoring module, and vehicle-state information including travel direction, gear-related state, and speed. These inputs are processed by an embedded edge device that applies context-aware decision logic and generates audio warnings. Validation in six short warehouse runs showed that multimodal fusion increased final-decision accuracy from 83.1% to 97.0% and reduced audio-warning candidates by 84.4% compared with module-only proximity warning logic. These initial results indicate the potential of the proposed system to improve warning relevance and support safer forklift operation under the evaluated warehouse conditions.
Thasnimol Valuthottiyil Shajahan, Babji Srinivasan, Rajagopalan Srinivasan
Open Access
Article
Conference Proceedings
Acoustic Latency and Timekeeping in Large Co-Located Ensembles: A Human Factors Analysis of Marching Band Coordination
Sound propagates at approximately 343 m/s, imposing delays of 175–292 ms between performers separated by 60–100 m. Marching ensembles manage these delays through selective listening, predictive timing, and visual conducting. This paper develops a document-informed systems and task analysis and a reproducible computational audit of five synthetic formations, each containing 100 modeled performers, at quarter-note tempi of 120–200 BPM. It distinguishes simultaneous sound production, alignment to locally heard references, and coherent arrival at a specified audience position. The audit enumerates 4,950 unique performer pairs per formation and evaluates nearest-neighbor delays, benchmark exceedance, geometric connectivity, and audience-arrival dispersion. Mean nearest-neighbor delays are 5.3–6.7 ms, whereas 56.4–87.1% of all pairs exceed a 25 ms comparison benchmark and maximum pair delays range from 67.9 to 254.2 ms. Simultaneous emission produces 48.9–124.2 ms of arrival dispersion at the reference listener. A specified college back-hash battery requires a 93.8 ms lead over the front ensemble, equivalent to 1.13 sixteenth notes at 180 BPM. A geometric derivation explains when listening back naturally compensates distance and when lateral routing introduces additional delay. These results characterize propagation constraints rather than establish behavioral failure thresholds. They support explicit listening assignments, staged rehearsal, audience-specific calibration, and formation-audit tools, while identifying priorities for subsequent field validation.
Sean Ge
Open Access
Article
Conference Proceedings
End-to-End MBSE Framework for Lifecycle Traceability, Digital Engineering, and System-of-Interest Realization
The increasing complexity of modern engineered systems, systems-of-systems, and digitally enabled enterprises has intensified the need for Model-Based Systems Engineering (MBSE) approaches that provide comprehensive lifecycle integration, architectural continuity, and end-to-end traceability. While existing MBSE frameworks offer valuable modeling methodologies and architectural viewpoints, many remain concentrated on system definition and development activities, providing limited guidance for maintaining traceability across implementation, integration, verification, validation, deployment, sustainment, maintenance, modification, and retirement. This paper presents novel ongoing research findings on E2E MBSE Framework that establishes structured lifecycle architectures spanning the complete realization of a System of Interest (SoI), from evolution to extinction. This research’s previous publications introduced conceptual fundamental constructs and formalized the operational architecture. Building upon these foundations, the present work completes the original research objective by integrating all twenty-four lifecycle packages into a coherent architectural structure that bridges both the left and right sides of the Systems Engineering (SE) Vee Model and establishes end-to-end lifecycle continuity. The framework is supported by a traceable architecture of 166 modeling constructs and collectively address evolution, mission analysis, stakeholder analysis, requirements analysis, system analysis, operational analysis, architecture development, implementation, integration, verification, validation, qualification, digital twin, deployment, sustainment, production scalability, in-service maintenance, and retirement or extinction. The principal contribution of this research is the establishment of a lifecycle-integrated framework that bridges longstanding gaps on both the left and right sides of SE Vee model by providing digital continuity through traceability, verification evidence, validation outcomes, and digital engineering artifacts. The paper discusses each package and demonstrates how the framework enables E2E digital engineering. A forthcoming dissertation presents a detailed SoI case study that applies all twenty-four packages to validate the framework's practical utility, completeness, and scalability for complex systems and systems-of-systems environments. The domain-agnostic E2E MBSE Framework can be utilized by any domain for any SoI.
Joshua Adelabu, Bhushan Lohar, John T Wade, Sean Walker, Carlos Montalvo
Open Access
Article
Conference Proceedings
A Structured Migration Framework for Brownfield Digital Transformation in Manufacturing Systems
The digital transformation of manufacturing enterprises poses significant challenges, especially in existing brownfield production environments Legacy production infrastructures are often characterized by heterogeneous automation systems, proprietary communication interfaces, and inconsistent data models, limiting interoperability between machines, IT systems, and digital services. At the same time, companies must maintain the operation of existing production assets while gradually integrating them into modern, data-driven value creation networks. Although numerous Industry 4.0 reference architectures, standards, and technologies—including the Reference Architectural Model Industry 4.0 (RAMI 4.0), the Industrial Internet Reference Architecture (IIRA), the Asset Administration Shell (AAS), OPC UA, Companion Specifications, and Digital Twins—provide essential building blocks for digitalization, they are typically considered in isolation. This paper investigates how these complementary concepts can be combined into a structured migration framework for Brownfield production environments. The proposed approach analyzes relevant reference architectures, communication standards, and semantic information models, categorizes them according to their functional roles, and describes their interaction within a stepwise migration process toward interoperable Smart Factory architectures. To illustrate the framework, the Smart Factory research and demonstration environment at Bochum University of Applied Sciences is used as a case study. The implementation examines how established Industry 4.0 reference architectures, standards, and enabling technologies can be systematically combined to support the structured digital transformation of existing Brownfield production systems.
Haris Karic, Arockia Selvakumar, Daniel Schilberg
Open Access
Article
Conference Proceedings
Personal Protective Equipment Use and Occupational Health Risks Among Farmers in Trinidad and Tobago: Preliminary Findings from an Agricultural Safety Study
Trinidad and Tobago's agricultural sector comprises approximately 32,000 farmers facing unique occupational hazards, including year-round pesticide application, sustained manual labor in tropical heat, and economic pressures limiting safety investments. Despite widespread PPE availability, compliance rates remain low due to cost, heat discomfort, limited access, and inadequate training. No comprehensive occupational safety and health data exists for Trinidad and Tobago's farming population, representing a critical knowledge gap. This study examines PPE usage patterns and their relationship with pesticide symptoms, musculoskeletal disorders, and mental health outcomes among Caribbean farmers. A cross-sectional survey was administered to n=170 farmers in January 2026, assessing PPE usage, pesticide exposure symptoms, musculoskeletal disorders with work limitations, and mental health symptoms. Regular PPE use was reported for 87.5% (foot protection) to 55.0% (hand protection), with an overall mean PPE score of 4.11 (SD=0.90). Pesticide symptoms were prevalent: 24.3% eye irritation, 20.6% headaches/dizziness. Musculoskeletal disorders were high: 48.8% lower back pain, with 35-52% reporting work limitation. Mental health burden was substantial: 51.2% depression symptoms, 42.9% anxiety. Spearman correlations revealed that higher PPE scores significantly negatively correlated with pesticide symptoms (ρ=-0.174, p=0.024) and MSD severity (ρ=-0.216, p=0.005), but not with mental health (ρ=0.135, p=0.081). Pesticide symptoms and MSDs were positively correlated (ρ=0.479, p<0.001). These findings inform the development of culturally appropriate, climate-adapted safety interventions for Caribbean farming populations.
Barry Parasram, Avishek Choudhury
Open Access
Article
Conference Proceedings
Exploring Spatial and Psychological Factors Associated with Creativity in Office Workers
Activity-Based Working (ABW) and hot desking have spread, but which spatial features promote employee creativity, and how, remains unclear. We surveyed 1,000 Japanese office workers on spatial/environmental features, psychological variables, workplace satisfaction, and creativity, using regression and path analysis. Psychological variables (R² = .481) and satisfaction items (R² = .285) explained considerable variance in creativity, with communication, personal control, and openness as key predictors. An exploratory path model (CFI = .816, RMSEA = .115) showed communication as the strongest direct predictor (β = .499, p < .001), followed by openness (β = .246) and personal control (β = .097, p < .05). The number of shared facilities had the largest indirect effect on creativity (β = .115, via communication), and seating arrangement a smaller indirect effect via personal control (β = .043). These results suggest spatial features may be associated with creativity indirectly, via communication and personal control.
Chika Nakatani, Toshihisa Doi, Shushi Doi, Rion Saijyo, Shiori Miyao, Shunya Shimizu, Takayuki Kitakata
Open Access
Article
Conference Proceedings
Eye-Tracking Factors During Spatial Ability Task Predict IRT-Based Item Difficulty and Naval Flight Training Performance
Since WWI, spatial reasoning abilities have demonstrated strong predictive validity for aviation training success. Eye-tracking offers millisecond-level indexing of underlying cognitive mechanisms: visual attention, visual search, and cognitive load. 228 U.S. Naval flight students performed the Terrain Orientation Task (TOT) while eye-tracking was recorded. Exploratory factor analysis reduced eight eye-tracking metrics into three factors: Visual Search, Visual Attention, and Cognitive Load. Objective difficulty was calibrated via a two-parameter logistic Item Response Theory model. Harder TOT items induced significantly larger peak pupil dilations (b = 0.06, p < .001) and greater pupil trace area under the curve (b = 0.02, p < .001), confirming pupillometry as an online cognitive load gauge. With escalating difficulty, cognitive load rose (b = 0.04, p = .027) while visual search narrowed (b = -0.29, p < .001), reflecting load-induced scan tightening. At the trial-level, broader visual search conferred a ~30% increase in the odds of a correct response (p < .001). At the participant level, the interaction between visual search and visual attention predicted Naval Introductory Flight Evaluation (NIFE) academic performance (β = -0.189, p = .013). whereas flight performance was predicted by interaction between attention and cognitive load (β = -0.26, p = .006). Adding eye-tracking factors to traditional selection metrics improved prediction of NIFE academics (ΔR² = +7%) and flight performance (ΔR² = +10%), while markedly enhancing attrition prediction (ΔAUC = +15%). These findings demonstrate that eye-tracking provides valuable cognitive workload and strategic signatures to enhance aviator selection, difficulty titration, and targeted coaching.
Alejandro Albizu, Laura Jamison, Kaylin Strong, Kayla Russell, Ciara Sibley, Shannon McGarry, Noelle Brown, Christopher Draheim, Cyrus Foroughi, Rebecca Nesmith, Sarah Melick, Joseph Coyne
Open Access
Article
Conference Proceedings
A Controlled Comparison of MARL Training-Execution Paradigms for Cooperative Multi-UAV Fire Detection
This paper presents a controlled empirical comparison of three fundamental multi-agent reinforcement learning (MARL) paradigms, the Centralized Training and Execution (CTE), Decentralized Training and Execution (DTE), and Centralized Training with Decentralized Execution (CTDE), in a cooperative Unmanned Aerial Vehicle (UAV) fire-detection environment under identical task conditions. In MARL, the training and execution architecture shapes what the agents can see, what they can coordinate on, and how stable the learning problem is. Our study addressed a gap where paradigm-level comparison under controlled settings remains scarce. Results across four benchmark scenarios and a 6 to 15 agents scalability test show that CTDE may be the most reliable paradigm, winning Level-1 mission completion in three of four scenarios and Level-3 path quality in all four. CTE provides a measurable advantage over CTDE in Hard-Dynamic, where execution-time global state conditioning enables faster response to mid-episode fire spawning, but its joint action space of 4^n renders it infeasible beyond thirteen agents. DTE fails systematically due to non-stationarity, with lower completion rates across all configurations and apparent safety gains that are artefacts of coordination failure rather than genuine collision avoidance. These findings confirm that paradigm choice is a first-order design decision in cooperative MARL and that CTDE may be a promising architecture for coordination-intensive multi-agent deployments at moderate to large scales.
Tanvi Chanchalani, Fuhua Lin
Open Access
Article
Conference Proceedings
Embodied Empathy as Pedagogy: A Consumer Tribe Field Immersion Activity for Customer Experience Professional Formation
A structural gap persists in higher business education between declarative knowledge of consumer behavior theory and the empathic competencies required for customer experience (CX) design practice. This paper presents the Tribal Disguise Activity, a structured six-phase experiential learning intervention in which undergraduate students design and embody consumer archetypes from diverse tribal identities, visit high-traffic commercial environments, and systematically document the service treatment they perceive while embodying that characterization. Implemented across eight consecutive academic semesters at two campuses of a major Mexican business school, the activity operationalizes Kolb's (1984) experiential learning cycle and is grounded in Cova, Kozinets and Shankar's (2007) consumer tribe framework. Drawing on reflexive thematic analysis of 115 reflection reports and 613 store interactions (2018–2025), with 25% independently second-coded, results show consistently perceived service differentials, emotional disequilibrium as a transformative learning catalyst, and a mindset shift from consumer to tribe-sensitive CX practitioner. Counter-patterns and limitations are discussed. The activity offers a replicable, low-cost, and scalable pedagogical instrument for business educators in international higher education contexts.
Mariana Alfaro Cendejas
Open Access
Article
Conference Proceedings
Teaching Human–AI Collaboration: A Task-Based Framework for Inclusive Design Education
Design education is absorbing generative AI without a shared view of what to teach and when. Two responses dominate: tool-by-tool training, which ages with each model release, and general “AI literacy” modules, which connect poorly to studio practice. Both leave questions of access, bias and skill formation implicit. This paper proposes a task-based framework that separates teaching objectives from specific tools. The framework crosses two established distinctions: Task Nature (Routine/ Abstract), from task-based labor economics, and AI Role (Automation/ Augmentation), from research on human–AI collaboration. The four quadrants, Efficiency Engines, Analytical Partners, Creative Sparks and Co-Creation Studios, describe different task patterns, and each is mapped to a characteristic cognitive mode: convergent, divergent, or an oscillation between the two. We show the framework at work by decomposing one published studio brief, a chair design project run with immersive hardware but without AI, into tasks across the quadrants, and by placing published teaching cases within it. We then read the framework through an inclusive-design lens and identify three tensions that concentrate unevenly: hardware access in Co-Creation Studios, aesthetic bias in Creative Sparks, and skill-base erosion in Efficiency Engines. For each we state the pedagogical implication rather than a single curricular fix. This paper draws on prior empirical studies; curriculum studies are the necessary next step. The contribution is a way for educators to decide what to teach that stays stable as tools change, and that shows where inclusion problems are likely to arise.
Dan Li, Lei Xia, Mudasir Ahmed, Ling Fan
Open Access
Article
Conference Proceedings
How startups communicate their positioning in the EdTech and corporate training market: a topic modeling analysis
The workforce transformation in established industries and organizations is driven by dynamics in technologies and changes in demographics. Consequently, organizations' skill requirements are shifting, and upskilling is becoming an essential practice to address this transformation. At the same time, this environment opens new opportunities for entrepreneurs to develop business models and creative solutions to overcome these challenges. Accordingly, a new industry has recently developed that offers corporate training and EdTech solutions at the intersection of technology, learning, and upskilling. Although these startups play a critical role in workforce transformation from a practical standpoint, there is a gap in the research landscape. It is therefore also essential for stakeholders, such as potential customers and investors, to gain a better understanding of these startups. For this reason, this study explores 1,106 startups from the Crunchbase database using biterm topic modeling (BTM). In doing so, this analysis reveals five clusters, in which startups position themselves: exam preparation, coaching and consulting, digital learning platforms, enterprise skill software, personalized learning apps. Furthermore, the results suggest that the startups in this sample focus primarily on customer benefits in their positioning and in their Crunchbase description. Finally, based on this exploration, the study then proposes a positioning map of startups in the EdTech and corporate training market.
Manuel Kaiser, Tim Beichter
Open Access
Article
Conference Proceedings
A Bloom's Taxonomy-Based Cybersecurity Training Curriculum for Air Traffic Controllers
Advancements in computer technology have improved the efficiency of air traffic management (ATM) systems while introducing new cybersecurity vulnerabilities. Cyberattacks targeting ATM infrastructure may disrupt air traffic control (ATC) services and compromise aviation safety. To strengthen the cybersecurity and resilience of ATM systems, air traffic controllers (ATCOs) need appropriate knowledge and skills to identify and respond to cyber threats. This paper presents a cybersecurity training curriculum designed to enhance ATCOs’ cybersecurity awareness and cyber threat identification skills. Guided by Bloom’s taxonomy, the proposed curriculum consists of three stages: lecture-based training, simulation-based training, and evaluation. The curriculum follows a sequential learning approach, progressing from foundational cybersecurity knowledge to the practical application of threat identification skills in simulated ATC operations. The instructional design specifies learning outcomes, training activities, and assessment methods to support the development of these competencies. Training effectiveness will be evaluated through pre- and post-training assessments and a follow-up questionnaire. The proposed curriculum provides a structured approach to developing ATCOs’ cybersecurity competencies and supporting the resilience of ATM operations.
Hui Wang
Open Access
Article
Conference Proceedings
System Awareness, Context Awareness, and Situational Trust in Human-AI Decision Making for Military Command and Control
Artificial intelligence (AI) is increasingly integrated into military command and control (C2), creating a need to better understand the human factors that shape the use of AI-generated recommendations. Existing research has largely examined awareness, trust, and AI-supported decision-making as separate constructs. This paper proposes an integrated Human-AI Decision Competence framework linking three complementary competencies: System Awareness, Probabilistic Context Awareness, and Situational Trust.System Awareness refers to understanding available capabilities, resources, and constraints, while Probabilistic Context Awareness extends traditional awareness concepts by incorporating uncertainty, feasibility, and risk into the interpretation of operational environments. Situational Trust represents the context-dependent judgment that determines whether AI-generated recommendations influence decisions.The framework is informed by recent research on system awareness, probabilistic information fusion, and AI-supported military planning. Findings from a judge-advisor study indicate that situational trust is the strongest predictor of AI advice acceptance, while experience contributes to trust formation and objective AI knowledge is associated with more calibrated skepticism.The paper argues that effective human-AI teaming depends on the interaction between awareness, uncertainty management, and trust calibration. Together, these competencies support adaptive decision-making and organizational agility in AI-enabled military environments. The framework contributes a unified perspective on Human-AI Decision Competence with implications for military education, training, and future decision-support systems.
Rune Stensrud, Sigmund Valaker
Open Access
Article
Conference Proceedings
Measuring What Generative AI Actually Saves in Instructor-Led Training Development: A Matched-Deliverable Time-and-Quality Comparison
Instructor-led training (ILT) is a core delivery mode for workforce development, safety, equipment familiarization, and procedural instruction, and producing it is expensive. Generative Artificial Intelligence (AI) promises to cut that cost, and organizations want a number they can trust. The obstacle is measurement. The metrics that are easiest to capture, software-recorded editing time and time per slide, are unreliable, and the apparent advantage of AI can appear or vanish depending on what is counted. ILT slides are deliberately sparse because the instructor carries the content, so word count per slide says little about completeness, and recorded editing time counts only work done inside the slide file. This paper fixes the deliverable instead: a complete teachable module of defined instructional scope, including slides, instructor talking points, activities, and assessment items, sufficient for a qualified instructor to deliver. Development time is measured to that standard and normalized per learning objective and per minute of instruction. Three workflows produced the same module from the same source material: an experienced designer, a junior designer, and a designer working with generative AI followed by Subject Matter Expert (SME) review. The AI-assisted workflow reached a complete module in 5 person-hours against 32 and 56, but drafting fell 94 percent while review and revision rose to 65 percent of what remained. The contribution is a matched-deliverable method that lets organizations state AI productivity gains defensibly rather than rhetorically, with guidance for safe adoption in safety-critical domains.
Richard Stone, Anthony Lavazza, Michael Jenkins, Claire Carter, Lixsa Barraza, Meagan Ault
Open Access
Article
Conference Proceedings
FULL PAPER ONLY: Beyond Tools: A Task-Based Human-AI Studio Scaffold for Critical Thinking in Architectural Education
Generative AI is increasingly present in architectural education, yet studio use often remains tool-centric, privileging visual acceleration over design reasoning. This paper reframes AI integration as a Human-AI interaction and UX-for-learning problem: educational value depends less on access to image generators than on interaction structures that require students to compare alternatives, justify selections, and reflect on consequences. Drawing on Human-AI interaction guidelines, creativity support tools research, architectural pedagogy, and design-cognition literature, the paper proposes a six-exercise studio scaffold covering conceptual exploration, iterative development, material and environmental simulation, diagram-to-atmosphere visualization, final visual portfolio enhancement, and AI-generated video presentation. Each exercise embeds lightweight process evidence through comparative outputs, captions, storyboards, and short written reflections. The paper is theoretical and design-oriented, but it also discusses some preliminary anonymized student exemplars as illustrative cases of framework applicability rather than empirical proof of learning gains. The main contribution is a repeatable scaffold that turns AI use into visible decision-making episodes and links ordinary studio outputs to critical-thinking dimensions such as interpretation, analysis, evaluation, inference, explanation, and self-regulation. The paper concludes by outlining a future empirical design for controlled comparison within parallel studio groups.
Silvia Albano
Open Access
Article
Conference Proceedings
Evaluation of ChatGPT-Generated vs. Self-Created Study Reviewers: A Case Study Comparing Student Test Scores, Efficiency, and Cognitive Workload
This study compared ChatGPT-generated and self-created reviewers in terms of engineering students’ test scores, time efficiency, and perceived cognitive workload. Using a within-subjects mixed-methods design, 30 engineering students participated. Each subject experienced using both a ChatGPT-generated reviewer and a self-created reviewer. Quantitative data were analyzed using descriptive statistics, ANCOVA, and correlation analysis, while qualitative responses were examined thematically. Results showed that ChatGPT-generated reviewers were associated with lower perceived cognitive workload and shorter average study time, while self-created reviewers produced slightly higher average test scores. ANCOVA results showed that the reviewer method significantly affected test scores, NASA-TLX workload, and time efficiency, even after controlling for covariates such as AI usage frequency, note-taking frequency, reviewer-making habits, mental exhaustion, and reviewer satisfaction. Note-taking frequency was also a significant predictor of test scores, with students who reported taking notes more frequently generally achieving higher test scores regardless of the reviewer method used. The reviewer-making habit was also a significant predictor of NASA-TLX scores. Participants who regularly create reviewers as part of their study routine reported higher perceived workload across both methods, showing differences in participants’ study habits. For time efficiency, none of the covariates were significant. Qualitative findings supported these results, showing that ChatGPT was preferred for speed and reduced effort, while self-created reviewers were valued for deeper understanding and retention. Recommendations for future studies include increasing the sample size, involving a wider population, and exploring a hybrid study approach.
Aliyah Maxine Mallari, Yngrid Bless Cadena, Aleeza Fiona Montinola, Lorelie Grepo-Jalao
Open Access
Article
Conference Proceedings
A Conceptual Framework for Modeling Team Adaptation in Cooperative Games through Ludic Knowledge
With the increasing importance of teamwork skills for modern workplaces, development of teamwork training programs has received substantial attention. Game-based teamwork training is one promising approach that is engaging, cost-effective, and well-suited to increasingly decentralized workplaces. However, design of effective game-based teamwork training requires understanding how a game elicits specific desired teamwork behaviors. Significant progress has been made in characterizing these relationships. However, despite its critical importance, little work has examined how a game's design influences team adaptability behaviors. This paper presents a preliminary framework for analyzing adaptability in cooperative games by conceptualizing adaptive stimuli as retrieval or disruption of players' ludic knowledge. We illustrate this framework through a qualitative case study that applies interaction analysis methods to gameplay videos of Overcooked!, a cooperative cooking game. We examined instances where game events led to players altering their behavior and connected the game's design features that resulted in each event with three adaptive stimulus cue categories. Although exploratory and limited to a small case study of a single game, the proposed framework is grounded in established theories across teamwork research and game studies, and it offers an initial vocabulary for describing how cooperative games can be designed to create demands for team adaptation. With this continued development, the framework may provide an analytic tool to help inform the design and evaluation of purpose-built game-based teamwork training environments.
Caleb Vatral
Open Access
Article
Conference Proceedings
Sensing the Seasons: Exploring Preschoolers' Seasonal Cognition through Multisensory Interaction
Current preschool cultural education mainly focuses on visual recognition. It lacks embodied and multisensory learning experiences, which makes it difficult to support children's deep understanding of culture. This study aims to explore how multisensory interaction can support preschool children's embodied understanding of seasonal knowledge and to examine its role in children's attention, exploratory behavior, and conceptual understanding. To achieve this, we grounded our design in embodied cognition theory and designed a multisensory seasonal toolkit themed on the twenty-four solar terms. Using both quantitative and qualitative analyses, we empirically examined the behavior and cognitive performance of 16 preschool children under different sensory conditions. Results showed that tactile-centered multisensory combinations significantly enhanced children's active exploratory behavior. Children's understanding of seasonal order was positively correlated with attention duration and spontaneous verbal expression. Qualitative findings revealed that multisensory experiences connect abstract cultural concepts with children's concrete perceptual understanding. These findings provide design implications for embodied seasonal learning experiences within preschool cultural education.
Yi Li, Yu Gao, Yanfei Xu
Open Access
Article
Conference Proceedings
Terrain Orientation Test Predicts Navy Flight Training Outcomes in High and Low Stakes Settings
This study evaluates the predictive validity of the U.S. Navy's Terrain Orientation Task (TOT), a spatial ability test for pilot selection introduced in 2023. The new test was developed after a previous one lost its effectiveness, partly due to applicants learning test-taking "tricks." However, a replica of the new TOT appeared online shortly after its official rollout, raising concerns about its continued validity. To investigate this, researchers studied 200 flight students who had already taken the TOT as part of their official Aviation Selection Test Battery (ASTB). In a low-stakes lab setting, the students completed the TOT again, practiced it extensively, and then took an alternate version. The results showed that scores from the official high-stakes test and both low-stakes lab attempts were all significantly correlated with the students' flight training grades. While practice did improve scores, the test remained a valid predictor of training outcomes. The study concludes that the TOT is a robust predictor of pilot performance in both high- and low-stakes settings, even with replicas available online. The authors recommend continued monitoring and leveraging the TOT's ability to generate new items to maintain its long-term integrity.
Joseph Coyne, Ciara Sibley, Kaylin Strong, Michael Kukenberger
Open Access
Article
Conference Proceedings
Detection Without Speed Isn’t Defense: Human-AI Coordination Against Cognitive Cyberattackers
Accurate attacker identification improves defense only if defenders can act in time. Because an attacker’s cognitive profile can dictate their network maneuvering, we evaluated whether an AI defender that profiles attacker cognition outperforms a human-inspired defender, and if pairing them improves defense. Using an agent-based simulation of an enterprise network, we tested two cognitively distinct attackers (aggressive vs. cautious) against three defender configurations (human-analog heuristic, AI, and human-machine team) under partial observability in a 2×3 factorial design. Counterintuitively, the human-machine team underperformed both individual defenders. The AI defender performed best, yet its identification accuracy remained at chance, indicating its success stemmed from strong default actions rather than accurate profiling. Improving the AI’s observability raised identification accuracy across three interventions but failed to improve defense. Only a structural intervention trapping attackers without identification reduced breaches, leveraging a compact topology with centralized hubs. These findings show that rapidly stopping a threat is distinct from identifying it. In many instances, effective human-AI cyber defense may rely less on profiling accuracy than on response speed and seamless coordination
Ayla Santos, Stephan Carney, Rob Groome, Nikolos Gurney
Open Access
Article
Conference Proceedings
The Vigilance Decrement in the Wild: How to Study and Design for Long-Term Process Monitoring
Several industries have made shifts towards highly automated or autonomous operations, with driving, aviation, and maritime operations serving as the most common examples. These concepts of operations shift the human from active operator to supervisory controller. Supervisory control requires consistent system monitoring to ensure success if human intervention becomes necessary. New build commercial nuclear power plants are being developed to leverage significant automation and autonomy, using sophisticated distributed control systems to manage and control the plant while human monitors continuously observe ongoing processes and trends. Process monitoring requires vigilance, the ability to remain alert and attentive while observing to detect a low likelihood event (e.g., abnormal plant conditions). Vigilance decrement is the degradation on monitoring performance over time and has been a known human performance issue since 1948 (Mackworth, 1948) with over 3,000 publications specifically on vigilance decrement. Despite the size of the literature, this research has translated into limited practical interface and job design guidance. This paper reviews and summarizes the primary theoretical frameworks posited as explanations and outlines a simulator-based method for studying and characterizing vigilance in a scientifically rigorous way that can be directly translated into actionable design insights. The study methodology proposes a multidimensional measurement approach, using physiological, survey-, and performance-based measures to capture the magnitude and timeline of the vigilance decrement over the course of a 2-hour observation interval.
Kelly Dickerson, Courtney Thomas, Ronald Boring
Open Access
Article
Conference Proceedings
OpenIN X: A Dual-Path Wearable Auditory Interface for Divided-Attention Support in Unilateral Conductive Hearing Loss
People with unilateral conductive hearing loss frequently perform two listening tasks at once: extracting dense factual content from a talker in front of them, and remaining responsive to events outside that conversation. Devices that optimise the first task — amplification, remote microphones, closed in-ear receivers — can make the second harder, because the ear carrying the improved speech signal is also the ear that would otherwise monitor the room. This paper presents OpenIN X, a dual-path wearable auditory interface that treats the situation as an information-allocation problem rather than an amplification problem. Semantically dense target speech is routed to a closed in-ear module on the intact side; brief environmental events are routed, after one fixed category-blind processing chain, to an open bone-conduction module worn on the affected side, where the ear canal may be unavailable to an in-ear fit. We report the design rationale, the integrated wearable design, and a within-participant feasibility protocol in which normal-hearing adults under acute unilateral attenuation complete twelve continuous service-counter scenes across four output conditions. Speech comprehension and environmental-event detection are reported as separate outcomes and never combined into a composite score. A worked analysis of simulated data shows that, at pilot scale, per-condition rates move in coarse steps, so the pilot can establish feasibility but not effect sizes.
Yutong Cai, Lei Xia
Open Access
Article
Conference Proceedings
I've Got an Idea: A Study on the Decision to Share
The decision to share an idea is a fundamentally human factors challenge — requiring an individual to assess cognitive risk, calibrate social trust, evaluate available protections, and make a behavioral choice under uncertainty. Yet design education has largely treated idea sharing as a given, embedding critique and collaboration into studio culture without addressing the factors that cause students to withhold their best thinking. When ideas stay locked inside the person who conceived them, they never live. This study applies a human factors lens to examine why professionals are reluctant to share ideas and proposes pedagogical mechanisms that design educators can integrate. A survey instrument was distributed via LinkedIn for its broad, cross-industry professional user base. A total of 71 responses were collected, with 52 completed surveys informing the analysis. Respondents ranged from 18 to 65+, with the largest cohorts in the 25–34 (37%) and 35–44 (25%) ranges. The survey examined willingness to share ideas with trusted friends, strangers, and on social media, and assessed relative trust in four protection mechanisms: patent protection, non-disclosure agreements, members-only access controls, and tracking systems. Findings reveal a pronounced trust gradient across sharing contexts. Patent protection earned the highest trust level but, notably, 29% agreed that no safeguard would make sharing feel secure. These findings inform four pedagogical interventions: structured critique formats with the startup community; IP literacy programming; founder lecture series; and team-building through disclosure. Together, these practices build the conditions under which sharing becomes the more rational choice.
Steven Doehler
Open Access
Article
Conference Proceedings
Continuous Competency Assurance through Human-AI Teaming: Reimagining Instructor Development
Competency-based training relies upon consistent instructor judgement, yet instructor calibration remains one of the weakest elements of many training systems. This paper proposes a continuous competency assurance framework integrating instructor recruitment, onboarding, calibration, recurrent standardisation and AI-enabled grading analytics. Artificial intelligence identifies grading trends, inter-rater reliability and behavioural divergence, while instructors retain responsibility for assessment and feedback. The proposed methodology transforms instructor standardisation from periodic compliance activity into continuous organisational learning. The framework supports airline operations, Advanced Air Mobility and future AI-enabled training ecosystems.
Debra Henneberry, Dimitrios Ziakkas, Konstantinos Pechlivanis
Open Access
Article
Conference Proceedings
Urban Context and Travel Experience Events: An Exploratory Comparison of Two German Cities
The presented study investigates associations between events influencing public transportation experience in both urban (Hamburg) and rural (Tuttlingen) areas in Germany, with the aim of identifying events associated with travel experience, which may in turn influence travel behavior. Using a mobile application, 21 participants in Tuttlingen and 70 participants in Hamburg tracked everyday trips, providing real-time evaluations of travel experiences along with situational data. Multi-level regression analyses were applied to assess associations of events such as punctuality, capacity offer, information about public transportation and others and the ontrip experience. Results indicate that a sufficient public transportation capacity offer shows the strongest positive association in Tuttlingen, whereas a lack of punctuality and low personal well-being show the strongest negative association. In Hamburg, a lack of punctuality and a negative information event seem to have the largest impacts. These findings provide a basis for prioritizing measures to be tested in local public transportation.
Marie Güntert, Esther Bosch
Open Access
Article
Conference Proceedings
Beyond Formal Likeness: A Four-Tier Evaluation Framework for Stylistic Fidelity in AI Generated Chinese Ink Videos
This study presents a four-tier evaluation framework for assessing AI-generated video style transfer in the context of Chinese ink animation. Developed through a case study of one contemporary artist, the framework draws on iterative co-creation, model training, and reflexive thematic analysis of expert interviews. It identifies four key dimensions: (1) visual genre and structural fidelity, (2) alignment with artistic intent, (3) brush-and-ink consistency, and (4) stylized emphasis of signature features. By integrating perspectives from the artist, art scholars, and AI practitioners, the framework emphasizes the expressive and cultural functions of style beyond surface texture. It serves as a practical tool for evaluating stylistic fidelity in AI-assisted video creation, supporting both artists and developers in achieving more meaningful and expressive outcomes. By operationalizing traditional Chinese aesthetic principles into computational contexts, this research bridges digital humanities and computational aesthetics. The framework also provides a culturally grounded basis for discussing responsible style reproduction in generative media.
Yilin Wang
Open Access
Article
Conference Proceedings
Designing a Gamified Hawker Centre Simulation for Language and Cultural Learning in Singapore
Gamification has emerged as a growing method in the field of education, utilising game elements to sustain student engagement and active learning to address ineffective traditional methods found in formal education. This study aims to utilise this technique in the context of language and cultural education in multicultural Singapore. Currently, Singaporeans utilise our unifying language of English to communicate between communities, thus introducing a gap between formal language education and real world interactions with local communities. This study addresses this gap through the development of an educational video game that immerses players in a culturally relevant scenario with the objective of enhancing Singaporeans' ability to learn their non-mother-tongue language and thereby improve cross-cultural understanding. In the context of a food establishment, the game tasks players with taking orders, preparing food and serving customers that only speak their mother tongue. By simulating authentic yet consequence-free interactions, the game aims to build vocabulary, listening comprehension and cultural familiarity. Data will be collected through in-game performance metrics, tracking players' playthroughs and scores over time. The project seeks to contribute to ongoing discussions regarding the role of gamification beyond academic education and its potential to cultivate practical life skills, intercultural competence and social understanding.
Jamie Johann Tan, Kenneth Y T Lim
Open Access
Article
Conference Proceedings
Actuation Feasibility and Cost Floor for a Low-Cost Active Knee–Ankle–Foot Orthosis: A Parametric Biomechanical Analysis
Powered lower-limb exoskeletons cost roughly USD 80,000–110,000, motivating maker-built active knee–ankle–foot orthoses (KAFOs) whose components are often chosen on price before the demand is known. We ask what minimum actuation an active KAFO needs for sit-to-stand (STS) transfer and what it costs. An illustrative low-cost design (6061-T6 frame, cable transmission, a 55 kgf·cm hobby servo, IMU/sEMG intent detection) is audited with a reproducible parametric model built from published biomechanics, manufacturer ratings and September 2026 prices; no human participants were involved. Full-assistance STS requires about 46–99 N·m of peak knee-extension torque per knee for 60–90 kg users, against 5.39 N·m servo stall torque: a gap of 8.5–18.4× (43–92× against a 20 % continuous rating). Because gearing cannot add power, even a lossless transmission limits knee speed under this load to 0.13–0.28 rad/s, 7–25× below natural STS knee speeds. Closing the torque gap through an assumed 40 mm knee pulley puts at least 1.15–2.48 kN in the cable, 2.7–5.8× the safe working load of 1/16-in stainless cable, and no bend-radius-compliant cable at that radius is admissible above 32.7 N·m. Integrated brushless actuators, several marketed for exoskeletons, already contain the reduction, making the cable stage redundant. A bilateral minimum-viable design built this way is priced at USD 2,345–3,459 (actuators 47–63 %), 24–47× below the Medicare fee schedule for powered exoskeletons but 4.3–8.4× above a hobby-component baseline, before validation, certification and service. We contribute a reusable feasibility-gate method, an explicit torque–power–tension–cost budget and bench prerequisites for human-loaded testing.
Cathy Chen
Open Access
Article
Conference Proceedings
Effects of Avatar-Mediated Helping and Gratitude Feedback on Users’ Prosocial Behavioral Intentions
Helping others and receiving gratitude may promote positive affect, well-being, and prosocial behavior. However, interpersonal helping may be inhibited by fear of rejection or uncertainty about receiving appreciation from others. We developed a web application that framed solving Sudoku puzzles as helping an in-app avatar and examined whether visual avatar responses and verbal gratitude affected prosocial behavioral intentions, affective states, depressive symptoms, avatar attachment, and application use. Thirty university students (26 men and 4 women; mean age = 21.63 years) completed a within-participant study with three counterbalanced 12-day conditions: Task-Only, Avatar-Response, and Avatar-Gratitude. The mean changes in prosocial behavioral intention from baseline to Day 12 were -1.467, 1.700, and 3.633, respectively. Although the omnibus Friedman test did not reach significance (p = .089), the prespecified within-participant comparison showed a greater increase in the Avatar-Gratitude condition than in the Task-Only condition after the Holm correction (adjusted p = .042, Cohen’s d = 0.53). The Avatar-Gratitude change was also greater than zero after correction (adjusted p = .015). No significant difference emerged between the two avatar conditions, and no consistent condition effects were found for affect, depressive symptoms, attachment, use, or self-reported effectiveness. Thus, combining avatar responses with gratitude may strengthen prosocial behavioral intentions relative to presenting a task without an avatar, but the independent contribution of verbal gratitude has not been established. Because participants used each application only about four times on average, avatar-based interventions may require immediate, salient, and individually adaptive feedback from the beginning.
Akihiro Ogino
Open Access
Article
Conference Proceedings
Translating User Needs into Telemedicine Training: A Mixed-Methods Approach to Designing an Augmented Reality System
Telemedicine carts (TCs) are being used more often in telehealth systems, which they allow remote clinical examinations. However, successful examinations require well-trained operators who can effectively coordinate equipment operation, communication, and patient/provider interactions. This study investigated the challenges faced by TC operators during pediatric visits using a mixed-methods approach that combined contextual inquiry, interviews, observations, eye-tracking, and user journey mapping. The findings showed that while the current hands-on training improves the familiarity with technology and equipment of TCs for novice users, it does not fully prepare the operators for the dynamic cognitive, procedural, and visuospatial communication demands of real telemedicine encounters. The identified needs and priorities, synthesized via user journey mapping, guided the design and development of an Augmented Reality Assistive Training System (ARATS). The proposed prototype provides frequent access to an adjustable, context-rich and immersive training within simulated clinical scenarios while preserving the values of hands-on learning. This study showcases a design-driven approach for translating user research into technological interventions for telemedicine training.
Elham Morshedzadeh, Wallace Lages, Shruti Shukla, Shiva Ghasemi, Andre Muelenaer
Open Access
Article
Conference Proceedings
Exploring User Affect in Virtual Reality Cultural Heritage Learning Using EEG
This study examined affective responses during guided learning and active interaction with a virtual Davula, an Assyrian goblet drum traditionally used at weddings, New Year celebrations, and festivals. We developed an immersive heritage-learning application using Three.js and deployed it on a Meta Quest 3. Eight participants completed the three-phase study, and seven provided complete synchronized datasets for the EEG-interaction analysis. The three phases are an initial baseline, guided learning about the Davula’s history and controls, and free exploration involving controller-based drum hits with audio feedback. A five-channel Emotiv Insight headset recorded six EEG-derived affective indicators: stress, engagement, focus, interest, excitement, and relaxation. We compared these measures across phases to determine whether guided learning and embodied artifact interaction produced distinguishable affective patterns. This work contributes an interactive VR platform and an exploratory methodology for integrating EEG-based affect sensing into immersive cultural-heritage experiences.
Maryann Godje, Javier Gonzalez Sanchez, Elizabeth Minor
Open Access
Article
Conference Proceedings
The Immersive Power of Binaural Sound: Reproducing the Experience of a Music Festival
Immersive technologies overwhelmingly prioritize visual fidelity and leave the auditory channel comparatively under-exploited, even though spatial sound is a powerful driver of presence. This study examined whether binaural (3D) sound can reproduce the experience of a live music festival in a digital setting, and how auditory and visual information combine to support immersion and emotional response. Performances at the 2025 Fono Festival in Québec City were recorded on site using a Neumann KU 100 dummy head paired with a head-mounted camera, which yielded binaural and standard stereo versions of the same performance at capture. Because spatial encoding occurs acoustically in the dummy head rather than being synthesized in post-production, the binaural signal preserves the complete auditory scene with its natural localization cues. Forty-eight adults, half of whom had attended the festival in person, experienced songs in binaural or stereo, with or without visual support, in a counterbalanced within-subject design. Without visual support, binaural sound elicited significantly greater presence and immersion than stereo, and binaural audio alone matched the immersive quality of audiovisual stereo. Adding video raised stereo to binaural levels but produced no further gain for binaural. The audio format had no effect on perceived clarity or readability of the sound. Visual support heightened arousal without affecting valence, and emotional responses were positive and moderately intense whether or not participants had attended the festival. Attendees judged the reproduction authentic in ambience and emotion, though not as immersive as being there. Because binaural delivery requires only headphones, it offers an accessible route to immersive experience.
Coralie Bureau, Jeanne Nicole, Cindy Chamberland, Alysun Paradis, Calypso Cases, Sébastien Tremblay
Open Access
Article
Conference Proceedings
Wearable Multimodal Sensing for Pilot Stress and Fatigue Monitoring Across Real-World Flight Phases
This exploratory study looks at patterns of stress and fatigue among four student pilots during flight training, using a mix of physiological recordings and self-reported measures. Physiological data, including heart rate (HR), electrodermal activity (EDA), skin temperature, and acceleration, were collected continuously during flight sessions, while perceived stress and fatigue were also recorded through questionnaires before and after each flight. The signals were then examined across flight phases to identify trends in arousal. EDA and skin temperature were seen to gradually increase across flights. Heart rate responses were more event-specific, with noticeable peaks during more demanding flight segments. Together, these patterns may indicate a gradual buildup of workload-related fatigue, as all four pilots reported greater fatigue post-flight. However, these observations do not establish that the signals specifically measured stress or fatigue; heat, movement, and differences between flights may also have contributed. The findings are exploratory and require confirmation in larger, more controlled studies.
Atandrila Chowdhury, Sudip Vhaduri, Julius Keller, Debra Henneberry, Mark Wilson
Open Access
Article
Conference Proceedings
Fabric-Based Wearable Sensors for Multimodal Motion and Pressure Tracking
We present a pilot study of a customized wearable sleeve prototype with sensing elements for the assessment of multimodal motion and pressure tracking with activity classification. The system combines APDM OPAL inertial measurement units (IMUs), textile-embedded flexible pressure-sensing matrices, and a video camera used as an external kinematic reference. The pressure sensors capture localized variations in body–garment interaction during movement, while the IMUs provide complementary measurements of joint motion and segment orientation. Integrating these sensing modalities enables the characterization of movement patterns, pressure distributions, and their relationship to different activities. The proposed framework offers a portable and human-centered approach for multimodal wearable sensing, with potential applications in ergonomics, smart garments, rehabilitation, and activity monitoring.
Ganit Goldstein, Amit Schechter, Skylar Tibbits, Dava Newman
Open Access
Article
Conference Proceedings
Evaluating Occupational Heat Stress Using Wearable Physiological Monitoring: A Field Study of Environmental and Physiological Response
Occupational heat stress is a significant safety hazard for workers performing physically demanding tasks in hot environments. Current heat stress management relies predominantly on environmental metrics, such as Wet Bulb Globe Temperature (WBGT), which measure ambient exposure but do not account for individual physiological responses. This study examined relationships among estimated Wet Bulb Globe Temperature (eWBGT), heart rate, and core body temperature, and assessed each predictor's relative contribution in an occupational setting. We collected data using the Slate Safety Band V2® wearable device, monitoring ten workers performing heat-exposed occupational tasks over a one- to two-week period. We used a linear mixed-effects modeling approach to evaluate eWBGT and heart rate as predictors of core body temperature, accounting for repeated measures nested within individual workers. Results showed that heart rate and eWBGT were significant independent predictors of core body temperature (p < .001). Heart rate emerged as the more direct predictor of physiological heat strain. Fixed effects explained 51.3% of the variance, and worker-level random effects increased the explained variance to 69.2%. An ICC of 36.8% indicated substantial variability between workers under comparable environmental conditions. These findings support integrating wearable physiological monitoring into occupational heat stress programs alongside environmental measurements. This study provides evidence relevant to emerging heat stress management regulatory frameworks and highlights the need for worker-specific physiological thresholds in elevated-heat occupational environments.
Brandon King, Jeremy Gouzd
Open Access
Article
Conference Proceedings
Blast and Physiological Wearables: Hardware Testing and Voice of the Customer Analysis
A literature review conducted in early 2026 identified various biosignals that have the potential to be affected by blast overpressure (BOP) exposure. These key physiological markers were identified as heart rate variability (HRV), electroencephalography (EEG), and electrooculography (EOG), gait, and balance, with further investigation needed into respiratory rates. For the physiological and BOP wearable systems to be utilized in tactical military scenarios, hardware reliability and end-user acceptance must be evaluated. Methodology: Two surveys were distributed in May 2026, the first End-user Voice of the Customer (VoC) survey was distributed to U.S. Army Soldiers to determine initial wearable preferences for comfort in sensor placements, willingness to wear, form factor, and data usage. The second was a Leadership VoC survey focused on how leadership would want to receive, view, and utilize the data from BOP and physiological sensors to make informed decisions. Preliminary hardware evaluation was conducted using commercial-off-the-shelf (COTS) devices such as physiological sensors, BOP wearable sensors, and Pencil Probes. An airgun was used and the previously mentioned devices were set up in various configurations to determine if the incident pressure wave creates any noise or interference with the hardware. The level of quality/reliability was also determined between the wearable sensors and the gold-standard Pencil Probe. Results and impact: The end-user survey results identified concerns of snagging and size, preferences towards a non-intrusive, lightweight sensor, a device that can be taken on/off. The leadership survey results identified a preference for a numerical Soldier readiness score that could be displayed on a computer or tablet. During bench testing the physiological device used for testing was equipped with the capability to record EKG, HRV, IBI, respiration, temperature did not experience interference or noise from exposure to the incident pressure waveform when mounted near the blast, not connected to a person. Path Forward: Future field testing will outfit soldiers with synchronized COTS devices for physiology and BOP wearable devices. A wearable survey instrument has been developed to gather data on Soldiers’ perspectives on the comfort of the wearables during their training, which will inform the results from the Voice of the Customer survey. The results from the operational tests should provide a better understanding of the effects BOP has on the readiness and performance of the operators of Army systems.
Alexis Cady, Kelly DiCristina, Robert DeMarco, Lucas Hess, Saniyah Bullock, Michael Menegus, Dominic Cheng, Florence Chua, Jose Rodriguez
Open Access
Article
Conference Proceedings


AHFE Open Access