Research On Adaptive Preview Distance Intelligent Vehicle Path Tracking Control

Path-tracking control is the core issue of unmanned driving technology for intelligent vehicles.Since the intelligent vehicle is highly coupled nonlinear systems with time-varying parameters and uncertain external disturbances,this paper developed a layered path tracking control system based on intelligent vehicle driverless test platform,and the research is as follows:Firstly,a 14-degree-of-freedom vehicle dynamic model reflecting the vehicle dynamic characteristics and tire nonlinearity is established on the basis of the actual vehicle structure based on the verification requirements of the control system by adopting the modular modeling concept.At the same time,the estimation models of vehicle real-time position,lateral acceleration and centroid sideslip angle are established based on the state quantities,which are measured easily.The model built is compared with the physical model provided by the vehicle dynamics simulation software IPG/CarMaker with the same conditions and the parameters.The results show that the model built can accurately describe the basic vehicle dynamic characteristics and represent the corresponding vehicle state to provide the model basis for the research on intelligent vehicle path tracking controller.Secondly,the preview error calculation model based on the road curvature is established by researching on the visual navigation system.Aiming at the characteristics of highly nonlinear,uncertain parameters and unpredictable external disturbances of intelligent vehicle systems,a path-tracking controller is designed based on the sliding-mode variable structure theory and the RBF neural network is used to approach the switching part of the sliding mode controller to eliminate the output chatter.The simulation results of double-lane change test show that the controller output is stable with good tracking accuracy and robustness.Thirdly,a path tracking controller based on a fixed preview distance makes it difficult to ensure the vehicle stability when the road curvature varies greatly.By analyzing the effect of different preview distance on the tracking precision and vehicle stability,this paper drew a conclusion that if the preview distance is long,the vehicle stability is superior,if the preview distance is short,and the tracking precision is superior.On account of this,a preview distance adaptive control system composed of fuzzy controller and iterative learning controller is designed and the genetic algorithm is used to optimize the fuzzy rules and membership functions.The simulation results show that the vehicle stability is significantly improved after optimization.Finally,hardware-in-the-loop simulations are performed on the ADAS test bench to verify the real-time and effectiveness of the control system designed in this paper.The control system is validated with the virtual traffic conditions,vehicle model,and real electronic control unit.The results show that both the real-time and effectiveness of the control system can meet the requirements.Then verify the control effect of the control algorithm on the real vehicle based on the intelligent vehicle test platform,under the single lane and ordinary road conditions.The experimental results show that the preview distance adaptive layered path tracking controller designed in this paper can effectively improve the vehicle stability while ensuring that the vehicle is tracking the desired path smoothly and accurately at low speed.The path tracking controller with adaptive preview distance according to the vehicle state effectively improves the vehicle stability while ensuring the accuracy of the path tracking to meet the requirements.The lateral acceleration and centroid sideslip angle are optimized by 42.71% and 42.75% respectively under double-lane change line test at highspeed.Hardware-in-the-loop simulation and real-vehicle tests have proven that the control system is effective.