Trajectory tracking for low-speed autonomous vehicles based on MPC and ADRC

This paper proposes a model predictive control (MPC)-based trajectory tracking controller with active disturbance rejection control (ADRC) feedback to reduce the lateral tracking errors of autonomous vehicles caused by simplified kinematic models. The controller employs MPC for feedforward optimization and ADRC for negative feedback regulation. In the designed ADRC controller, the reference steering angle output from MPC replaces the smoothing signal generated by the tracking differentiator (TD), eliminating the TD while retaining the first-order extended state observer (ESO). In addition, a linear state error feedback (LSEF) control law is adopted to replace the nonlinear state error feedback (NLSEF) control law, which simplifies the ADRC structure and parameter tuning while preserving its disturbance compensation capability. Simulink/CarSim co-simulations on circular and sinusoidal trajectories at 3–10 m/s show that, compared with conventional MPC, the proposed MPC-ADRC controller reduces the mean lateral error by 10.6%–13.4% on circular paths and 8.7%–23.5% on sinusoidal paths; compared with an MPC-PID composite controller, the reductions are 6.1%–9.3% and 3.1%–14.8%, respectively, without compromising steering smoothness or ride comfort. Monte Carlo tests with heading-angle noise and speed fluctuations confirm that MPC-ADRC achieves the smallest mean lateral error under random disturbances, demonstrating robustness and real-time applicability.