Using Simulation to Study Communication between Autonomous Vehicles and Vulnerable Road Users
Autonomous vehicle (AV) technology is advancing at a rapid pace. One concern that has come to the forefront is how AVs will communicate and interact with vulnerable road users such as pedestrians and bicyclists. The goal of this project is to use a pedestrian simulator to examine how pedestrians learn to respond to visual cues about whether AVs acknowledge them at a crosswalk and intend to yield to them, in both daytime and nighttime conditions. We will use a cue-learning approach in which participants learn through experience about AV cues signaling awareness and intention. The task for participants is to stand at the edge of a crosswalk at a 4-way stop and cross without colliding with a vehicle. We will use a between-participants design to examine how pedestrians respond to cues from autonomous vehicles. In the gaze-directing condition, the vehicle “eyes” will direct their gaze toward the participant to signal acknowledgement. In the attention-getting condition, the vehicle “eyes” will flash on and off to signal acknowledgement. We expect that it will be easier for pedestrians to learn about the meaning of cues in the gaze-directing condition because the gaze-directing cues mimic eye contact between real drivers and pedestrians and the attention-getting cues do not clearly communicate “go” to the pedestrian. This research will fill a critical gap in our understanding of communication between autonomous vehicles and vulnerable road users that can be used to increase pedestrian safety and inform vehicle design.