Pedestrian estimation of their crossing time on multi-lane roads

Pedestrians' road-crossing behavior towards eHMI-equipped autonomous vehicles driving in segregated and mixed traffic conditions

Pedestrians' road-crossing behavior can be influenced by eHMIs (external Human-Machine Interfaces) on autonomous vehicles (AVs). In this research, we developed a novel eHMI concept that aimed to support pedestrians' risk evaluation by displaying predicted real-time risk levels. In a virtual reality environment, we measured pedestrians' road-crossing behavior when they encountered AVs with this eHMI and manual-driven vehicles (MVs) in the same lane. Results showed that pedestrians exhibited typical crossing behaviors based on gap size for both vehicle types. In segregated traffic conditions, compared to MVs, eHMI-equipped AVs made pedestrians more sensitive to the changes in gap size by rejecting more small gaps and accepting more large gaps. Pedestrians also walked faster and kept larger safety margins for smaller gaps. Similar results were observed for AVs in mixed traffic conditions. However, in mixed traffic conditions, pedestrians faced more challenges when interacting with MVs as they tended to accept smaller gaps, walk more slowly, and maintain smaller safety margins. These findings indicate that dynamic risk information could be conducive to pedestrians' road-crossing behavior, but the use of eHMIs on AVs might disrupt pedestrians' interactions with MVs in complex traffic conditions. This potential risk shift among vehicles also poses the question of whether AVs should use segregated lanes to reduce their indirect impacts on pedestrian-MV interactions.