A Game Theory-Based Approach for Modeling Freeway On-Ramp Merging and Yielding Behavior in an Autonomous Environment

This paper uses video from UAV to optimize the Connected-Automated Vehicles (CAVs) operation system, to ensure the merging and yielding behavior safety and efficiency at freeway on-ramp section in an autonomous environment. The vehicle trajectories are extracted by Yolo and Tracker. An algorithm is supposed to identify the key point of merging behavior. Next, this paper applies the framework of game theory to the autonomous environment, uses bi-level programming to minimize deviation and find the pure strategy Nash equilibrium solution. Using TTC to test the safety of the model, the results indicate that this framework can effectively choose the best action of freeway on-ramp merging and yielding situations while achieving more effective merging operation.