Lateral feedback control of intelligent vehicles with dynamic parameter identification

With the rapid advancement of intelligent driving technologies, accurate acquisition and real-time updates of vehicle state parameters and dynamic characteristics have become critical. Tire cornering stiffness, as a key vehicle dynamics parameter, quantifies the relationship between tire cornering angles and lateral forces under current operational conditions. To address the variation of cornering stiffness during vehicle cornering, a dynamic parameter identification-based model predictive controller (DPI-MPC) is proposed. First, under the small-angle hypothesis, the original three-degree-of-freedom (3-DOF) nonlinear vehicle dynamics are linearized to obtain a tractable control-oriented model. Then, the memory fading recursive least squares method (MDRLS) is used to estimate the cornering stiffness of the tire. Particularly, considering that the prediction step and control step are mostly obtained through empirical settings, a chaotic particle swarm optimization algorithm (CPSO) is designed to search for the optimal solution, and then the dynamic cornering stiffness identification model predictive controller (DPI-MPC) is used for lateral control. The effectiveness and stability of DPI-MPC are verified through path tracking on the CarSim and MATLAB/Simulink co-simulation platform. The results show that the proposed control strategy can achieve precise path tracking while maintaining vehicle stability under different working scenarios and vehicle speeds.