Over time and under pressure: a longitudinal virtual reality study exploring the effects of time pressure and road environment on pedestrian-automated vehicles interactions

Research suggests that Automated Vehicles (AVs) may improve the safety of Vulnerable Road Users (VRUs). Over time, VRUs might adapt their behaviour as interaction with novel technology increases. To safely incorporate AVs into current traffic, we must understand how VRUs interact with AVs, e.g., at road crossings.
We conducted a longitudinal between (time-pressure), within (repeated-exposures) Virtual Reality (VR) experiment to analyse the crossing behaviour of 30 pedestrians when exposed to various types of AV (small car vs lorry), road gradient (uphill vs downhill), AV-pedestrian distance (40 vs 80 m) and weather (sunny vs misty).
A novel finding was that AV moving downhill elicited more cautious crossing behaviours. Participants under time pressure made decisions approximately 47% faster, walked across 30% quicker and fixated 91% less on AV compared to no-time pressure conditions. Behavioural adaptation was observed; participants initiated crossing 31% earlier and showed 59% less fixation on AV in second session compared to first. Differences in arousal, which are positively correlated to perceived risks, were evident through reduced skin conductance response events by 56% in the second session. Significant relationships also existed between personality traits and crossing behaviours.
The pedestrians' behavioural, psychophysiological and self-reported measures give insight into factors that elicit risky crossing behaviours. As pedestrians adapt their behaviour through repeated exposures to AVs, early-stage AV design should prioritise building user trust to support smoother long-term integration. The findings can inform policies to improve pedestrian safety, especially for those who perceive more risks from AVs or exhibit riskier road crossing behaviours.