A trajectory tracking and motion control framework for automated vehicles using adaptive robust control and machine learning

To address the degradation in tracking accuracy and stability of autonomous vehicles (AVs) under extreme conditions, this paper proposes an integrated trajectory tracking and motion control (TTMC) framework utilizing the Adaptive Robust Control and Machine Learning methodology (ARC-ML). Within the control framework, a modular architecture is adopted that implements a Linear Quadratic Regulator (LQR) alongside three adaptive robust controllers to enhance tracking precision and stability while reducing computational complexity within each controller. The three adaptive robust controllers explicitly account for parameter uncertainties, unmodeled subsystem mismatches, and external disturbances, guaranteeing robust performance across diverse operating scenarios. Furthermore, to achieve an optimal balance between conservatism and robustness within the control framework, the nominal system model with time-varying uncertain parameters is precisely identified via the Recursive Least Squares (RLS) algorithm, while the robust boundaries of the error model are determined and adaptively scaled using machine learning methods integrating Gaussian Process Regression (GPR) with Bayesian optimization. Sufficient conditions for closed-loop stability under diverse robust factors are provided by the Lyapunov method analytically. The simulation results on MATLAB/Simulink and Carsim joint platform, along with the actual vehicle experiments, demonstrate that the proposed methodology considerably improves tracking accuracy, driving stability, and robust performance, guaranteeing the feasibility and capability of driving in extreme scenarios.