Learning-based Sliding Mode Predictive Trajectory Tracking and Stability Control for Autonomous Vehicle in Unstructured Environments
Aiming at the difficulty in balancing trajectory tracking accuracy and yaw stability of high-speed autonomous vehicles in complex unstructured environments,a trajectory tracking and stability coordination control method based on learning sliding mode predictive control(L-SMPC)was proposed.Firstly,a data learning prediction model is constructed based on Gaussian process regression(GPR)to solve the problem of residual model uncertainty and noise interference in the complex unstructured environment.Then,a trajectory tracking and stability coordination control method based on Gaussian-sliding mode predictive control(GP-SMPC)is proposed,the basic model and Gaussian uncertainty prediction are combined as the control model,and the sliding mode control method based on rolling prediction optimization is designed,thus satisfying the real-time performance and robustness of the controller under multiple constraints.In addition,the future vehicle driving risk prediction model is constructed,and the weight coefficients of the multi-objective fusion function are decided in advance by recursive Bayes theorem based on the relative residuals in the prediction horizon,which satisfies the global optimal performance.The simulation results show that,the proposed method effectively improves the trajectory tracking accuracy and dynamics stability of high-speed autonomous vehicles on unstructured roads with road noise interference.
NETL
NSTL
万方数据
