Automated Vehicle (AV) technologies promise to enhance traffic safety through precise, sensor-based control. However, their effectiveness depends on accurately predicting and responding to the behaviors of other road users, including other AVs and Vulnerable Road Users (VRUs). This study addresses the decision-making processes for AVs by proposing a discrete-time Markov sequential framework grounded in game theory. The interactions between AVs and VRUs are modeled to maximize utility while accounting for the dynamic behaviors of VRUs. The study develops efficient heuristic algorithms integrating customized dynamic programming and adaptive methods to enable real-time application and potential deployment in real-world AVs. The proposed framework is validated in a two-lane traffic scenario with a crosswalk. The results highlight the framework’s potential to enhance AVs’ operational safety in mixed-traffic environments.
Simulation in high density of vehicles
Simulation in high density of pedestrians
