Pedestrian-Safety Perceptions in Automated-Vehicle Crossing Scenarios: An Immersive Virtual-Reality Experiment

Understanding how pedestrians perceive roadway conditions and vehicle-state communication is important for examining pedestrian interactions with automated vehicles (AVs). This study used an immersive virtual reality experiment to examine three trial-level perception outcomes—perceived crash likelihood, perceived environmental safety, and perceived collision severity—across urban local-street crossing scenarios. The primary analysis included 949 trial-level observations from 104 participants across scenarios varying in site context, roadway-design configuration, vehicle speed, and vehicle condition. The three vehicle conditions represented a conventional vehicle, an AV without an external human–machine interface (eHMI), and an eHMI-equipped AV. Linear mixed-effects models were used to account for repeated observations within participants. Curbside buffer presence was associated with lower perceived crash likelihood and perceived collision severity and higher perceived environmental safety. Relative to the conventional-vehicle condition, the eHMI-equipped AV condition was associated with more favorable ratings across all three outcomes, whereas no statistically significant differences were detected between the AV-without-eHMI and conventional-vehicle conditions. Direct comparison of the two AV conditions showed lower perceived crash likelihood and perceived collision severity for the eHMI-equipped AV. The 50-km/h vehicle-speed condition was associated with higher perceived crash likelihood and perceived collision severity but not with perceived environmental safety. Overall, the findings highlight the importance of distinguishing multiple dimensions of pedestrian safety perception and support the use of immersive VR as a controlled platform for examining perceptual responses to roadway and vehicle conditions.