Research On Model-free Control For Path Tracking Of Autonomous Vehicle

Automatic driving vehicles can improve the safety of ITS,and as one of the methods to reduce traffic congestion and traffic accidents,the development of automatic driving vehicles has attracted more attention.Perception positioning,decision planning and execution control are three main components of vehicle software configuration layer.In the executive control layer,one of the most important problems of automatic driving vehicles is path tracking.This thesis studies a path tracking Model-free Controller based on adaptive dynamic programming method(ADP)without the knowledge of vehicle model parameters.The controller can minimize the path tracking error and centroid side deflection angle.Considering that the driving of the automatic driving vehicle will be affected by the external environment interference in the complex environment,this paper uses RBF neural network to approach the unknown external disturbance of the vehicle,and generates the compensated front wheel angle.The combination of the compensation front wheel angle and the path tracking controller based on ADP constitutes an ADP-RBF controller with strong anti-interference capability.In order to solve the problem of poor tracking effect of ADP-RBF controller in the high curvature path,this paper improves the ADRC based on the classical theory of ADRC and vehicle dynamics.The decoupling control of multi-input and multi output ADRC is studied.The direct yaw moment is used as another control input in the vehicle except the front wheel angle to improve the vehicle path following the path in the high curvature path Stability of trace.Based on the selfadjusting method of parameters of BP neural network,the parameters which are difficult to be determined in the ADRC are adjusted adaptively,and an adaptive ADRC is formed for path tracking.The phase plane method is used to judge the vehicle lateral stability.When the curvature is small and the vehicle lateral stability is good,ADP-RBF controller is used for path tracking.When the curvature is large and the automatic vehicle becomes unstable,the adaptive ADRC is used to track the path.Considering that the switching between different control algorithms can easily produce large jump of the front wheel angle input,which causes the disturbance and adverse transient response of the vehicle system,this paper designs a fuzzy switching controller based on fuzzy control theory.The simulation results show that the model paste switch controller has excellent control performance in the high curvature Road and the changing road condition.The fuzzy switching controller designed in this paper is verified by hardware in loop experiment.In this paper,through the existing hardware platform and related experimental equipment,the corresponding Car Maker/Simulink control program is established,and the effectiveness of the fuzzy switching controller designed is verified through the HIL experiment,...