A Game-Theoretic Approach on Conflict Resolution of Autonomous Vehicles at Unsignalized Intersections

In order to resolve the driving conflict and improve the safety and efficiency of autonomous vehicles at unsignalized intersections, a new decision-making method based on game theory with personalized driving preferences considered is proposed in this paper. In the decision-making method, the alterable game mode is constructed by combing the designed game entrying mechanism and replanning of game sequential order under different intersection conditions to ensure the adaptiveness and effectiveness of the decision-making algorithm. Besides, four payoff indicators and personalized payoff function are designed with the consideration of driving efficiency, safety and comfort requirement. The advantage of proposed method is that it not only reduces the complexity of game mode and improves the effectiveness at the same time, but also fully considers the personalized driving preferences to realize human-like driving and personalized decision while the existing methods usually neglect the difference among the drivers and its influence. Five different typical scenarios at the unsignalized intersection are given under co-simulation environment of Matlab and Prescan. The results show that the proposed decision-making method is able to resolve the driving conflict and ensure the safe passage of autonomous vehicles at unsignalized intersections, which verifies the effectiveness of the proposed method.

Automated summary

In this paper, a new decision-making method based on game theory is proposed to resolve the driving conflict and improve the safety and efficiency of autonomous vehicles at unsignalized intersections. The proposed method constructs an alterable game mode by combining a designed game entrying mechanism and replanning of game sequential order under different intersection conditions to ensure adaptiveness and effectiveness. Four payoff indicators and personalized payoff function are designed considering driving efficiency, safety, and comfort requirements. The advantage of the proposed method is its ability to reduce complexity, improve effectiveness, and consider personalized driving preferences. Five different typical scenarios at unsignalized intersections are simulated, and the results demonstrate the effectiveness of the proposed method.