A High-Fidelity VR Simulation Study: Do External Warnings Really Improve Pedestrian Safe Crossing Behavior?

Abstract

To better communicate with pedestrians, adding external displays to autonomous vehicles (AVs) has been proposed as a potential communication method to encourage safe crossing behavior by pedestrians. Whereas, most researchers have conducted intercept interviews, lab studies, or simulation studies to explore the efficacy of these displays, these approaches only studied crossing intention but did not explore crossing behavior. We developed a high-fidelity virtual reality scenario where participants could demonstrate actual crossing behavior within an adequately replicated real-world street. We simulated a local street with scalability of the real world in a VR environment, conducted an experiment in an empty space large enough for participants to move across the road in the VR environment. A mixed-method approach assessed attitudinal and behavioral interactions with potential warning patterns. The results showed that the warning patterns contributed significantly to pedestrians’ perceptual vigilance, as in past studies, but safer crossing behavior was not observed. This suggests that crossing intention measures may not be an adequate substitute for behavioral measures of crossing.