Is eHMI always helpful? Investigating interference of eHMI on pedestrians interacting with autonomous vehicles on multi-lane streets: A virtual reality-based experiment
Date:
Li, Z., Che, Y., Zhou, Z., Liang, H., & Ye, Y.* (2025, July 22-25). Is eHMI always helpful? Investigating interference of eHMI on pedestrians interacting with autonomous vehicles on multi-lane streets: A virtual reality-based experiment [Poster Presentation]. The 25th COTA International Conference of Transportation Professionals (CICTP2025), Guangzhou, China.
Abstract: Addressing the issue of right-of-way negotiation between pedestrians and vehicles is a key factor in ensuring traffic safety. Currently, the external human-machine interfaces (eHMI) of autonomous vehicles are considered an effective solution and can be categorized into two types: allocentric communication and egocentric communication. Although the effectiveness of eHMI has been widely studied, its potential interference with pedestrian crossing behaviors in complex scenarios remains uncertain. This study conducted a Virtual Reality-based experiment to investigate pedestrian crossing behaviors when interacting with autonomous vehicles equipped with different eHMIs in multi-lane streets. The results indicated that when pedestrians interact with autonomous vehicles in multi-vehicle scenarios, the co-existence of different eHMIs increases pedestrians’ cognitive load, thereby interfering with their crossing behavior. Additionally, when the same type of eHMI was used consistently across lanes, pedestrians exhibited significantly longer gaze time and waiting time, as well as shorter post-encroachment time (PET), under allocentric eHMI compared to egocentric eHMI. Higher cognitive load was also reported for allocentric eHMI in terms of subjective cognitive load assessment. This study provides valuable insights for the future application and standardization of eHMI design.