Research on trajectory control algorithm based on MPC and LQR

This study proposes an integrated control strategy combining LQR and MPC algorithms to improve trajectory tracking in self-driving vehicles. A two-degree-of-freedom vehicle dynamics model is used to design transverse and longitudinal controllers. Key parameters of the longitudinal MPC are optimized using a genetic algorithm. The control algorithms are validated in lane-changing scenarios using a CarSim and Simulink simulation platform. The results show that the strategy achieves a lateral deviation of less than 0.004 m and longitudinal deviation below 0.2 m, with smooth heading and steering angle changes, improving tracking accuracy, stability, and ride comfort.