Research On Path Planning And Tracking Control Algorithms Of Autonomous Driving Vehicles

In the current rapid development of artificial intelligence,autonomous driving vehicles came into being.Studying autonomous driving vehicles and promoting the formation of an intelligent network of "vehicle-road-central transportation system" will greatly alleviate road traffic pressure and reduce traffic accidents caused by driver errors.In the architecture of autonomous driving vehicles,decision-making and control occupy two important parts.Studying the real-time obstacle-avoiding path planning algorithm and high-reliability path tracking control algorithm is the key to realize autonomous driving.This thesis firstly studies the autonomous driving vehicles path planning problem,it uses OpenStreetMap to generate a global planning path,and applies the Frenet coordinate system to the local path planning research.The main research contents of this part are as follows:(1)Automatically export road network information using OpenStreetMap,then generate a global planning path;(2)Derive the mapping relationship between Global coordinate system and Frenet coordinate system,and transform the autonomous driving vehicle path planning problem into Frenet coordinate system,simplifying the solution process and improving real-time performance;Establish the horizontal and vertical path planning model of the autonomous driving vehicle,and obtain the autonomous driving vehicle motion state through sampling,solve the horizontal and vertical path planning equations,obtain the horizontal and vertical feasible path sets,and design the path evaluation function according to different performance indicators,generating an optimal path in different situations;Join the plan to avoid the obstacle path function,design the safe driving distance,select the optimal path that meets the safe driving conditions in the feasible path set,and plan the path that makes the autonomous driving vehicle truly safe;(3)Carry out the loop experiment in the Northeastern University,verify the performance of the path tracking algorithm designed in this paper in the actual environment,and continuously adjust the optimization algorithm according to the experimental results.Then this thesis designs different tracking controllers for autonomous driving vehicles under different working conditions.The main research contents are as follows:(1)Design horizontal and vertical controllers for low-speed autonomous driving vehicles:Using the vehicle kinematics model,design a horizontal pure tracking controller and a longitudinal speed PI controller.After verifying the reliability of the controller at low speed in the simulation environment,it is applied to the autonomous driving experimental vehicle of Northeastern University for real vehicle verification:after completing the initialization of CAN initialization,serial communication initialization and STM3 2 port configuration,the campus is circled.Experiments,tracking based on path planning results,and verifying the performance of the reliable tracking path of the controller at low speeds;(2)Design a lateral controller for high-speed autonomous driving vehicles:Design a lateral model predictive controller using the vehicle steering dynamics model.The predictive model of model predictive control is constructed by vehicle state space model,and the corresponding objective function is designed.Vehicle dynamics constraints are added to the traditional model predictive controller in the controller design process to improve the stability of vehicle path tracking.Finally,the simulation platform is built in Matlab&Simulink to simulate the highspeed driving of autonomous driving vehicles under fixed road conditions,and the longitudinal speed of the vehicle is changed to verify the universality of the controller at different speeds.The results show that the controller has better path tracking.The effectiveness of the model predictive control on the path tracking control of high-speed autonomous driving vehicles is verified.