Should I stay or Should I go? Using a Cooperative Sneak Peek Interface in Highly Automated Vehicles
Authors: Jürgen Pichen, Nikol Figalová, Martin Baumann
Abstract: In the future, highly automated vehicles (HAVs) will be capable of performing aspects of driving safely and without human assistance. However, even HAVs may occasionally reach a system boundary and not perform efficiently, especially in complex traffic situations. A human driver could outperform the HAV in such situations, given that human eyes and cognitive processes are likely to be more capable in certain tasks than advanced sensors. Therefore, allowing the human driver to cooperate with an automated system may improve driving efficiency and safety. Implementing the driver-vehicle cooperation approach, the HAV could suggest a manoeuvre that can either be accepted or ignored by the driver. To ensure the driver has enough information prior to making the decision, we proposed a “Sneak Peek” function. This cooperative feature allows drivers to adjust the lateral position of the HAV prior to accepting the manoeuvre request. This study aims to explore how drivers use the Sneak Peek in scenarios when an overtake manoeuvre is requested, but the view from the ego vehicle is obstructed by a lead vehicle. Method. We conducted a driving simulator study with 27 participants (15 female, age M = 25.0, SD = 3.2). The ego vehicle drove in a highly automated mode (SAE level 4) with a velocity of 100 km/h. Participants experienced nine scenarios in which they approached a slow-moving lead vehicle (LV) driving at 70 km/h. The position of the LV was either right, the centre, or left of the lane, which changed how obstructed the view from the ego-vehicle was. At a distance of 45 meters from the LV, the HAV suggested an overtaking manoeuvre. Participants could (a) approve the overtake request straight away, (b) adjust their lateral position using the Sneak Peek function prior to deciding whether to accept the requested manoeuvre, (c) continue slowly following the LV. Results. In total, participants accepted 97.7 % of overtake requests. The Sneak Peek function was used in 83.1 % of trials. We found a significant association between the LV lateral position and the Sneak Peek usage (X2 (2, 486) = 65.69, p <.01). Participants used the Sneak Peek significantly less when the LV was positioned on the left of the lane (63.6 %) than when it was in the center (93.8 %) or right (92.0 %) of the lane. Moreover, the duration of the Sneak Peek was longer when the LV was positioned to the left side of the lane (M = 9.6, SD = 10.0) than when on the right of the lane (M = 6.0, SD = 6.4), (F(2, 399) = 5.21, p < .05). Furthermore, higher visual obstruction caused by the LV on the left lane lead to a greater lateral position adjustment (F(2, 444) = 19.044, p <.01), which implies that when the LV was positioned to the left of the lane, participants moved the lateral distance of EV further to the left. All participants confirmed that the Sneak Peek was useful in helping them decide whether to overtake.Conclusion. The Sneak Peek function helps drivers to decide whether to overtake or not and is a feasible cooperative interaction concept.
Keywords: Cooperative driving, assistant systems, human factors, cooperative interface, driving efficiency, highly automated driving
Cite this paper: