Decision-Making for Vehicle Overtaking on Urban Roads: A Level-k Game Theory Approach

The increasingly complex interaction between ve-hicles on urban roads has made the overtaking problem more challenging. This paper proposes a vehicle overtaking decision- making method based on the Level-k game theory. Firstly, this method considers the uncertainty of actual vehicle behavior and establishes a dynamic model and a prediction model that incorporate the uncertainty of vehicle positions. Secondly, the action set and reward function of the vehicle are designed to describe the overtaking scenarios on urban roads. Then the overtaking problem is constructed based on the Level-k game theory between vehicles. Finally, the optimal overtaking action for the vehicle is obtained from the action set using the Q- learning algorithm in reinforcement learning. The effectiveness of this method is verified through vehicle overtaking simulations on urban roads in the MatlablPython/SCANeR platform. The simulation results show that this method has an appropriate conservatism.