Research On Autonomous Driving Path Planning And Tracking Control Method Based On Improved RRT

With the development of cities in full swing in recent years,transportation,which is the pulse of the city,is facing great pressure,and the innovation of driverless technology is imminent.Path planning and tracking control as a key technology has important research significance,there are currently some global path planning and trajectory tracking algorithms,but there are some defects so can not be applied to practice,such as the A* algorithm commonly used in path planning,in the planning path is there are a large number of sharp points,large amount of calculation,long planning time,seriously affecting driving efficiency,ant colony algorithm based on bionic ideas has slow iteration speed,initial exploration disorder problems,in recent years more popular artificial potential energy algorithm,It has the inevitable defect of falling into the minimum value,and it has always remained in the theoretical stage in the field of unmanned driving.Therefore,this thesis proposes an improved RRT algorithm for global path planning,which effectively reduces the planning time,the number of sampling points and the number of sharp points,improves the ability to pass through complex areas,and makes the planned path always meet the kinematics and dynamics constraints of vehicles.Aiming at the trajectory tracking algorithm,this thesis proposes an improvement for the MPC controller,and integrates the fuzzy adaptive control algorithm to solve the problem of abandoning the smoothness of large curvature sections to improve the tracking accuracy during trajectory tracking.This thesis mainly focuses on global path planning,horizontal and longitudinal control and heading angle control.Global path planning is to give the starting point and the target point,consider the surrounding environment to find the optimal driving path between the two points,provide a basis for the subsequent local path planning and tracking control,for the planned path,the improved control algorithm is used to control the reference trajectory to achieve the control effect that meets the actual driving requirements.The specific research content is as follows:(1)Vehicle kinematics model research.Taking the unmanned vehicle itself as the research object,considering the vehicle kinematics and obstacle collision detection mechanism,simplifying the vehicle model and introducing Ackerman corner,and designing a fast collision detection model based on the improved outer circle,it provides prerequisites for global path planning and tracking control to meet the driving requirements of real-world unmanned vehicles.(2)Research on global path planning for unmanned driving.Aiming at the problems of repeated exploration of the area by the traditional RRT algorithm,inability to quickly bypass obstacles,too many sharp points of the final planned route,the corner does not meet the requirements of vehicle dynamics,oscillation at the end point at the final convergence,and inability to pass quickly in the narrow environment,the target paranoid strategy,adaptive step size strategy,node drop strategy,minimum corner restriction strategy,improved algorithm convergence conditions,and greedy strategy are added to improved the above problems one by one,through four typical scenarios in RRT,RRT*,The effect of the three Bi-RRT algorithms is compared and verified to improve the path planning of the algorithm.(3)Research on tracking controller based on reference trajectory.Under the condition that the trajectory required for global path planning for unmanned driving has been obtained,a tracking controller based on MPC and fuzzy adaptive control is designed,and the secondary programming of the MPC objective function is carried out using the discrete linearization method,and the design constraints are considered,considering that the problem of abandoning smoothness to achieve tracking accuracy will occur when MPC tracking the trajectory,and the fuzzy adaptive control algorithm is combined with MPC to exert control over the weight of the controller input.Finally,the Carsim-Simulink joint simulation platform is used to build the vehicle model and road model,and the rationality of the path planning and the accuracy of tracking control are verified by taking the reference trajectory and the actual trajectory,very longitudinal error and heading angle error as the evaluation criteria.