Autonomous vehicles are the most advanced intelligent vehicles and will play an important role in reducing traffic accidents, saving energy and reducing emission. Motion control for trajectory tracking is one of the core issues in the field of autonomous vehicle research. According to the characteristics of strong nonlinearity, uncertainty and chang-ing longitudinal velocity for autonomous vehicles at high speed steering condition, the robust trajectory tracking control is studied. Firstly, the vehicle system models are established and the novel target longitudinal velocity planning is carried out. This velocity planning method can not only ensure that the autonomous vehicle operates in a strong nonlinear coupling state in bend, but also easy to be constructed. Then, taking the lateral location deviation minimiz-ing to zero as the lateral control objective, a robust active disturbance rejection control path tracking controller is designed along with an extended state observer which can deal with the varying velocity and uncertain lateral dis-turbance effectively. Additionally, the feedforward-feedback control method is adopted to control the total tire torque, which is distributed according to the steering characteristics of the vehicle for additional yaw moment to enhance vehicle handing stability. Finally, the robustness of the proposed controller is evaluated under velocity-varying condi-tion and sudden lateral disturbance. The single-lane change maneuver and double-lane change maneuver under vary longitudinal velocity and different road adhesions are both simulated. The simulation results based on Matlab/Simulink show that the proposed controller can accurately observe the external disturbances and have good performance in trajectory tracking and handing stability. The maximum lateral error reduces by 0.18 meters compared with a vehicle that controlled by a feedback-feedforward path tracking controller in the single-lane change maneuver. The lateral deviation is still very small even in the double lane change case of abrupt curvature. It should be noted that our proposed control algorithm is simple and robust, thus provide great potential for engineering application.
Meta Reinforcement Learning-Based Lane Change Strategy for Autonomous Vehicles
