Path Tracking Method for Intelligent Vehicles Based on Adaptive Model Predictive Control With Tire Stiffness Estimation

Path tracking is one of the core tasks for intelligent vehicles. Under continuously varying driving conditions, tire forces exhibit strong nonlinear characteristics. Model Predictive Control (MPC) based on a linear tire model may suffer from significant prediction errors because it does not account for the nonlinear variations of longitudinal stiffness and cornering stiffness, thereby making it difficult to ensure high-accuracy trajectory tracking. To address this issue, this paper proposes an Adaptive Model Predictive Control (AMPC) approach that explicitly considers tire-force calculation errors. First, an Adaptive Sliding Mode Observer (ASMO) is employed to estimate the longitudinal tire force. Second, by integrating the High-order Cubature Kalman Filter (HCKF) with the Strong Tracking Filter (STF) concept, a Singular Value Decomposition-based High-order Strong Tracking Cubature Kalman Filter (SVD-HSTCKF) is developed to estimate the lateral tire force. Then, correction factors for longitudinal and lateral stiffness are calculated based on the deviation between the estimated tire forces and those predicted by a linear tire model, and these factors are used to dynamically update the prediction model. Finally, the proposed AMPC strategy is validated through simulation. Results show that, compared with conventional MPC and LQR methods, the AMPC approach reduces lateral displacement errors by 47% and 69%, and yaw angle errors by 45% and 72%, respectively, significantly enhancing path tracking accuracy and control robustness.