Research on path tracking control of human-machine co-driving vehicles based on game theory

To address the control challenges of human-machine co-driving systems under complex conditions and reduce driver workload, this paper proposes a game theory-based integrated framework combining state estimation, path tracking control and intelligent decision-making. First, the extended Kalman filter and the cubature Kalman filter are designed to estimate the key parameters in real time. Through joint simulation verification, the extended Kalman filter with better estimation performance is selected. Then, an adaptive model predictive controller and a single-point preview driver model are constructed for path tracking at different speeds. The controller time-domain parameters are selected using the approach to ideal value sorting method, and the adaptive MPC and driver model are compared with nonlinear model predictive control and a linear quadratic regulator, respectively. Finally, for the dynamic allocation of human-machine lateral driving authority, a driving authority allocation model based on game theory is constructed. The human-machine participation index, comfort index, vehicle anti-collision index, and human-machine conflict confrontation index are comprehensively considered to optimize the control weight allocation from multiple dimensions. The Nash product is used to solve the equilibrium strategy. Simulation results verify the effectiveness of the game-based authority allocation strategy, while real-vehicle tests demonstrate the feasibility of the path-tracking control framework.