Adaptive Optimization of MPC Algorithm for Autonomous Vehicle Trajectory Tracking Control

To ensure the motion performance of intelligent vehicles under various driving conditions, the vehicle trajectory tracking controller must have strong robustness and adaptability. However, it’s difficult for the control algorithm with fixed parameters to meet both the requirements of tracking accuracy and stability when operation parameters and driving conditions such as speed and road adhesion coefficient change. This paper proposed an adaptive adjustment method for path tracking control based on linear model predictive control (LMPC), where the predictive model and control parameters are optimized automatically. First, modelling errors caused by simplifications of the prediction model of LMPC are compensated to represent the nonlinear characteristics of tire dynamics. Second, constraints about the front wheel side slip angle are imposed to further enhance the robustness of trajectory tracking control in the nonlinear region of vehicle dynamics. Third, the predictive horizon and weight coefficient are optimized by fuzzy control according to the speed and road adhesion coefficient. Finally, the offline simulation and real time simulation are carried out. The results show that the proposed adaptive optimization method for MPC enhances path tracking accuracy and vehicle stability when driving conditions change, which improves the robustness of trajectory tracking controller significantly.