An MDP Model of Vehicle-Pedestrian Interaction at an Unsignalized Intersection Conference Paper uri icon

abstract

  • 2018 IEEE. Though autonomous vehicles are currently operating in several places, many important questions within the field of autonomous vehicle research remain to be addressed satisfactorily. In this paper, we examine the role of communication between pedestrians and autonomous vehicles at unsignalized intersections. The nature of interaction between pedestrians and autonomous vehicles remains mostly in the realm of speculation currently. Of course, pedestrian's reactions towards autonomous vehicles will gradually change over time owing to habituation, but it is clear that this topic requires urgent and ongoing study, not least of all because engineers require some working model for pedestrian-autonomous-vehicle communication. Our paper proposes a decision-theoretic model that expresses the interaction between a pedestrian and a vehicle. The model considers the interaction between a pedestrian and a vehicle as expressed an MDP, based on prior work conducted by psychologists examining similar experimental conditions. We describe this model and our simulation study of behavior it exhibits. The preliminary results on evaluating the behavior of the autonomous vehicle are promising and we believe it can help reduce the data needed to develop fuller models.

name of conference

  • 2018 IEEE 88th Vehicular Technology Conference (VTC-Fall)

published proceedings

  • 2018 IEEE 88TH VEHICULAR TECHNOLOGY CONFERENCE (VTC-FALL)

author list (cited authors)

  • Hsu, Y., Gopalswamy, S., Saripalli, S., & Shell, D. A.

citation count

  • 6

complete list of authors

  • Hsu, Ya-Chuan||Gopalswamy, Swaminathan||Saripalli, Srikanth||Shell, Dylan A

publication date

  • August 2018