Research On Path Planning And Tracking Control Of Autonomous Vehicle

With the rapid development of science and technology,autonomous vehicles have gradually entered the public vision.Autonomous vehicles bring convenience to people’s travel,but its safety problem can not be ignored.In order to ensure road safety,autonomous vehicles are usually equipped with active collision avoidance control system,and the current collision avoidance method is mainly achieved by longitudinal braking.Some studies have shown that the lateral collision avoidance control system based on active steering has better collision avoidance effect under high-speed,low adhesion road and other limit conditions.In this paper,the autonomous vehicle is taken as the research object,and the dynamic path planning and tracking control under steering operation to avoid collision are mainly studied.Planning and tracking an optimal path to avoid collision in real time is the key of emergency collision avoidance control.In the planning of collision avoidance path,a continuous reference path for collision avoidance is designed based on Sigmoid function in combination with road environment information and vehicle dynamics constraints.When the vehicle travels along the planned collision avoidance path,new obstacles may suddenly appear,and the obstacles sometimes appear in the form of movement.Therefore,this paper constructs the motion trend of obstacles in the predictive horizon of model predictive control in accordance with the movement law of obstacles,and designs the risk index between vehicles and obstacles.The original reference path is dynamically planned based on model predictive control.On the basis of dynamic path planning,path tracking controllers are designed based on linear and nonlinear vehicle models respectively,and the front steering angle is optimized and input to the CarSim vehicle model to realize the tracking control of the reference path planned by the upper controller.Simulation results show that NMPC controller has better stability and path tracking performance than LMPC controller.However,when the vehicle speed increases,the path planning model adopted by the upper controller is too simple,ignoring the nonlinear factors of the vehicle,so the planned collision avoidance path under emergence condition is not ideal.In this case,the lower path tracking controller will not be able to track the planned reference path better.In order to solve the problem of path tracking failure caused by hierarchical controller under emergency conditions,dynamic path planning and tracking control are integrated into a constrained optimization problem,and they share the same high-precision vehicle model.An integrated collision avoidance control strategy is designed based on MPC,and the positions of moving obstacles are predicted in the predictive horizon of model predictive control.In the predictive horizon,the strategy of varying discrete steps is designed to realize the remote prediction and high precision control of the collision avoidance controller.The simulation results show that the designed integrated collision avoidance controller can guarantee good collision avoidance effect and vehicle stability.