Game-based Lane-changing Decision Model for Leading CAVs in Mixed Platoons

The inclusion of connected and automated vehicles as leading vehicles in mixed platoons has the potential to achieve smoother and safer road traffic, but current research seldom focuses on the formation of mixed platoons with multi-lane distribution characteristics. To address this problem, a game-based lane-changing decision model is proposed for leading CAVs in mixed platoons. The model establishes a lane-changing decision mechanism that combines optimized and game-based approaches. It initiates a leading CAVs' target lane initialization process with the objective of minimizing the number of CAVs lane changes. Based on the payoff from CAVs' game-based lane-changing, it updates the target lanes of leading CAVs to achieve multi-lane distribution of mixed platoons. Furthermore, a non-cooperative game matrix between CAVs and HDVs is established based on game theory, considering both lane-changing efficiency and safety. Time and safety payoff functions are designed to quantify the lane-changing risk of CAVs. Microscopic simulations are conducted using the traffic software SUMO. Experimental results demonstrate that compared to the baseline model, the proposed game-based lane-changing strategy maintains a mixed formation completion rate of over 97% under different mixed traffic volumes, with an average reduction of about 40% in the lane-changing time for each group of leading CAVs, while keeping the lane-changing frequency of each group of leading CAVs at a low level.