Task-Based AR-HUD in Autonomous Driving: Enhancing Driver Agency, Engagement, Attention, and Takeover Performance

Abstract

In SAE Level 3 autonomous driving, prolonged passive monitoring presents the severe challenge of drivers falling 'out-of-the-loop'. To address this, we propose a 'task-based AR-HUD', an interface that transforms vehicle motion planning into actionable task opportunities requiring driver approval, thereby maintaining the driver's cognitive flow within the driving loop. This research investigates the psychological and cognitive impacts of such task-based interactions to enhance driver takeover performance. Utilising a high-fidelity driving simulator, we compared three progressive AR-HUD visual feedback modalities: linear, dynamic, and task-based. Subjective metrics concerning compliance, perceived control, comfort, and emotional experience were analysed. Results indicate that the task-based AR-HUD significantly enhances driver agency and engagement, effectively redirects attention to the road, and sustains the driver's 'in-the-loop' state. This study offers theoretical foundations and empirical evidence for future high-level autonomous driving human-machine collaborative designs that balance psychological needs with functional safety.

Keywords: Autonomous Driving, AR-HUD, Task-based Interaction, Sense Of Agency, Attention Maintenance

DOI: 10.54941/ahfe1007865