| In recent years,with breakthroughs in sensor technology,computer data processing capabilities,and 5G communication technology,autonomous driving technology is entering a period of vigorous development.The emergence of autonomous driving technology provides a new way to solve transportation problems.The quality of the autonomous driving motion control algorithm directly affects the safety and reliability of autonomous driving vehicles,and has always been a core research direction in the field of autonomous driving technology.This thesis focuses on the lateral control process of vehicle path tracking,and based on the established nonlinear dynamic model of the vehicle,designs the lateral controller using mainstream control methods.Through experiments,the thesis compares the advantages and disadvantages of different algorithms and explores practical ways to further improve the performance of the controller.The main contents are as follows:(1)Based on the dynamic characteristics of the controlled object,a nonlinear dynamic model of the tire,a vehicle kinematic model,and a vehicle two-degree-offreedom dynamic model are constructed.The relevant parameters of the linearized tire model are identified based on actual vehicle experimental data.(2)Based on the established vehicle dynamic model,the vehicle lateral control system is designed using both Linear Model Predictive Control(LMPC)and Nonlinear Model Predictive Control(NMPC)algorithms.To address the nonlinear tire problem in the NMPC algorithm,a genetic algorithm is used to identify the key parameters of the nonlinear tire model.A joint simulation platform using Car Sim/Simulink is built,and both controllers are compared and analyzed through simulation experiments on the platform.(3)A vehicle lateral control system is designed using the Active Disturbance Rejection Control(ADRC)algorithm,and compared and analyzed through simulation experiments with the two previously discussed Model Predictive Control(MPC)algorithms.The impact of changes in vehicle model parameters on controller performance in traditional model control methods is investigated,as well as the advantages and disadvantages of model-based control methods versus model-free control methods.(4)A method for generating a virtual safe driving environment based on expected trajectories and boundary constraints is proposed.A virtual potential field is generated using differential homomorphism mapping theory.The traditional NMPC lateral controller is improved by applying constraints based on the virtual potential field.Finally,the effectiveness of the virtual field constraint controller is validated through simulation and real vehicle experiments. |
