Lane-Changing Decision-Making and Planning Based on Multi-Vehicle Game Interaction

Decision-making and planning technology, as a critical component of autonomous driving systems, has been a focal point of research in structured road environments in recent years. Previous studies often neglected the interaction between the ego vehicle and surrounding vehicles or failed to accomplish decision-making and planning tasks in real time. The paper mainly studies the decision-making and trajectory planning tasks of autonomous driving in a scenario with three lanes and seven human-driving vehicles, and proposes a lane-changing decision and planning method based on multi-vehicle game interaction. The optimal lane-changing duration and longitudinal distance are selected considering safety and efficiency, while identifying and predicting surrounding vehicles based on the intelligent driver model and analyzing styles. Decide the optimal lane and whether to initiate lane-changing considering vehicles in the adjacent lanes through game interaction based on the optimal lane-changing duration and distance. The lane-changing path planning is conducted using a clothoid curve, while the speed is planned using dynamic programming (DP) and quadratic programming (QP). Finally, by comparing with the rule-based method and human-driving through simulations and experiments, the superiority of the proposed lane-changing decision and planning method based on multi-vehicle game interaction is verified.