Research On Path Following Control Method For Vehicle Automatic Emergency Avoidance

With the continuous development of science and technology,automobiles bring convenience to people,but also bring more and more traffic accidents.Advanced assistant driving system can improve the safety and comfort of vehicles and reduce traffic accidents.In the past 40 years,the advanced assistant driving system has become the focus of the major automobile manufacturers and scholars at home and abroad.Emergency avoidance of vehicles is mainly divided into braking and steering two ways,for low-speed vehicles,the use of braking can be very good to avoid obstacles,and for high-speed vehicles,only rely on braking to avoid obstacles is not appropriate.The main research direction of the subject is: how to control the front wheel active steering of the vehicle when the vehicle is running at different speeds and is disturbed by the lateral wind,so as to realize the accurate and stable track of the route in the process of high-speed automatic emergency avoidance.In this paper,it is known that model predictive control not only has the advantage of adding many kinds of constraints in the control process,but also can carry out rolling optimization online by studying the principle of model predictive control theory.For high-speed vehicles,in order to avoid obstacles safely and steadily,a variety of constraints must be set up.Therefore,based on the two-degree-of-freedom vehicle dynamics model,vehicle kinematics model and “Magic Formula” tire model,a model predictive controller is designed,which includes three parts: predictive model,constraints and optimization objective function.In the aspect of planning avoidance path,considering the shortcomings of the common path planning methods,such as curvature discontinuity and excessive lateral acceleration.Using the characteristics of the Sigmoid function which curvature is continuous and can satisfy the lateral acceleration constraints of vehicles to do path re-planning.Considering the uncertainties of external disturbances and other factors,the active disturbance rejection controller with fixed parameters has the problems of poor control accuracy and unsatisfactory effect.To solve this problem,an active disturbance rejection control method based on neural network is proposed.A second-order active disturbance rejection controller is designed,and using the neural network to tune the three-order extended state observer parameters online.Using the joint simulation of Carsim and Simulink to verify the control effect of model predictive controller,conventional active disturbance rejection controller and neural network active disturbance rejection controller under different vehicle longitudinal velocities and different lateral wind speed disturbances.The path tracking effects of different controllers are compared and analyzed.The results show that the comprehensive performance of active disturbance rejection controller based on neural network is better than that of active disturbance rejection controller and model predictive controller.Neural network active disturbance rejection controller can be independent of the accurate model of the system,and can observe the changes of system parameters and compensate for them.So,it has its unique advantages in path tracking and has strong robustness and anti-interference ability for lateral wind disturbance of different wind speeds.