Path Curvature Based Variable Weight Multi-Constraint MPC for Smart Vehicle Trajectory Tracking Control

Jianmin Xu,Yunxun Wu,Yongjie Lin

DOI: https://doi.org/10.1109/cac59555.2023.10451293

2023-01-01

Abstract:In order to balance the accuracy and stability of trajectory tracking for intelligent driving vehicles under the com-plex working conditions, this study proposes trajectory tracking control strategy based on path curvature variable weight model predictive control. model predictive control is established with front wheel steering angle and front wheel speed as control variables, using real-time lateral displacement error, longitudinal displacement error and azimuth error as inputs, and setting control volume constraints, control volume increment constraints and friction circle constraints. Based on changes in road curvature, the weight of the model predictive control objective function is optimised online, and the influence of the weight matrix on trajectory tracking accuracy and stability is coordinated. Based on Simulink, a vehicle kinematic model and control strategy were established to validate the effectiveness of the proposed control strategy through simulation experiments. The results show that compared with linear time-varying model predictive control, the proposed variable weight control based on path curvature can reduce the maximum tracking displacement error by 65.70 %; and reduce the trajectory tracking mean square error by 55.06%. The maximum speed which satisfy the control tracking accuracy of not less than 0.3m can be increased by 20.69 %; The maximum speed which satisfy the control tracking accuracy of not less than 0.1m can be increased by 46.67%.