Effect of External Appearances on Pedestrian Fixation Behaviors with Mid-sized Vehicles: Virtual Reality-Based Research

Autonomous vehicles (AV) with different appearances are getting common. Some are equipped with conspicuous sensors, while most are similar to conventional vehicles (CV). Pedestrians may face difficulty in distinguishing AVs. This paper explored whether pedestrians can distinguish autonomous vehicles from CVs through external appearances. Eight mid-sized vehicles were chosen considering vehicle usages, vehicle types and automation levels and displayed different behaviors in a VR experiment. Thirty-one participants were involved and asked to stand on the sidewalk, watching the vehicle approaching or passing by in sixteen trials with eye movements recorded. Results showed that no significant difference in automation levels (AV and CV) on fixation duration was found, indicating that pedestrians may not be able to distinguish AVs from CVs. However, the scan rate of CVs’ rear was significantly larger than that of AVs’ rear, indicating pedestrians’ possible distrust in AVs. Moreover, areas of interest (AOI) that attracted pedestrians varied with vehicle types (SUV, taxi and minibus). The findings suggested the necessity of external human-machine interfaces (eHMI) for AVs and provided guidance for eHMI designers to choose proper positions for different vehicle types. Moreover, vehicle features with autonomous genes were summarized, which may help original equipment manufacturers design AVs that are both recognizable and pedestrian-friendly.