To Cross or Not-to-Cross: A Robotic Object for Mediating Interactions Between Autonomous Vehicles and Pedestrians

Abstract

A main challenge in incorporating autonomous vehicles into urban environments concerns communication with pedestrians crossing in front of them. We introduce a robotic object for facilitating natural crossing interactions by leveraging pedestrians’ existing crossing habits. The robot was designed to fit on the vehicle’s dashboard and simulate head gestures to indicate if it is safe to cross. We evaluated the perception of the robot as mediating the AV’s communication and the comprehension of its gestures, in an in-person experiment. Participants were asked to cross the road in front of a vehicle that was presented as having autonomous driving capabilities. The robot was placed on the vehicle’s dashboard in a location where pedestrians habitually look. The results indicated that when asked to cross, participants immediately looked toward the driver’s seat and easily noticed the robot. They consistently categorized the robot’s gestures into ”cross” or ”do-not-cross” categories and reported a strong sense of safety. Our findings suggest that robotic objects are a promising technology for mediating natural pedestrian-AV communication.