Pedestrians Receptivity in Autonomous Vehicles: Exploring a Video-based Assessment

Deb, S., Hudson, C. R., Carruth, D. W., & Frey, D. (2018). Pedestrians Receptivity in Autonomous Vehicles: Exploring a Video-based Assessment. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. Philadelphia, PA. 62(1), 2061-2065.

Abstract

Pedestrian receptivity toward autonomous vehicles (AVs) usually depends on the extent to which they receive indication of the vehicle’s intended action. Previous studies have typically used overt subjective measures (trust measures, ratings, etc.) and few objective measures (walking speed, waiting time, etc.) to identify external features that can improve pedestrians’ receptivity toward AVs. The current study aims to evaluate pedestrians’ behavioral measures of receptivity based on their body (head and foot) movements as they experience an AV in a virtual traffic environment. Videos of pedestrians at a virtual crosswalk, interacting with an AV that was equipped with an external feature indicating different operator statuses were coded. The operator statuses used in this study included: no driver, attentive driver, and distracted driver. The external features used were: no feature, upraised hand, stop sign, walking silhouette, walk in text, music, and a verbal message. Pedestrian body movements were derived from the video to determine frequency for looking at the approaching vehicle while crossing and stops after initiating crossing. Average durations for initiating crossing after signal were calculated. For no feature condition, the waiting time was calculated when participants observed the car. Data were compared with pedestrians’ self-reported ratings for receptivity to investigate body movements’ sensitivity to participants’ receptivity level. Results suggest body movements are sensitive to individual differences in reported receptivity. Future work should further examine the utility of this behavioral metric by further examining situational differences.