Multi-Dimensional Taylor Network Optimal Trajectory Tracking Control For Intelligent Vehicle

In recent years,intelligent vehicle is a common motion carrier,due to development of artificial intelligence technology and computer technology,intelligent vehicle has attached more and more attention from researchers.At present,the main research interests of intelligent vehicle are localization,route planning,motion control and so on,and motion control is the basis of the research in the related fields of intelligent vehicle.Common control methods still have many disadvantages on solving the problem of trajectory tracking control for intelligent vehicle,such as slow response and poor anti-disturbance ability.In this thesis,a controller is designed to eliminate the unstable factors in the trajectory tracking of the intelligent vehicle system.Firstly,a multi-dimensional Taylor network optimal control is designed for the problem of trajectory tracking control for intelligent vehicle.Since intelligent vehicle is affected by interference like torque interference,sensor disturb,etc.For this problem,multi-dimensional Taylor network optimal control can improve the dynamic performance,anti-interference ability and robustness of the intelligent of the intelligent vehicle system by its high term.Based on adaptive control conception,an adaptive law is designed to make the Lyapunov function satisfy the Lyapunov stability condition,and the optimal parameters of the controller are obtained through the training of the adaptive law.Secondly,in order to simulate the performance of multi-dimensional Taylor network optimal control applied to the intelligent vehicle in the actual environment,a navigation system is constructed for the intelligent vehicle,which includes map module,localization module and route planning module.And environment map is the basic element of location module and route planning model.Considering unknown of the pose and environment map of intelligent vehicle,a Fast SLAM algorithm based on scan matching and particle filtering is designed in this thesis,which can determine the position and pose of the intelligent vehicle and construct the environment map at the same time.Based on the traditional Fast SLAM algorithm,a scanning matching segment is designed to correct particle's pose,which can improve the accuracy of the Fast SLAM algorithm.Thirdly,a localization module is designed for in navigation system.Since the performance of the navigation system depends on computational efficiency,this thesis improves the adptive Monte Carlo algorithm commonly used in localization,and simplifies the updating process of importance coefficient by evaluating the matching of local obstacle information generated by sensors and map information,which improve the efficiency of localization algorithm.Then,a decision module is designed for navigation system,which is designed as two parts: global route planning module and local route planning module.Global route planning module is aim to calculate the best trajectory from the starting point to the target point,which also avoid all obstacles.In this thesis,an A* algorithm is designed for global route planning,and a route re planning system is designed to avoid the failure of the original global route when the intelligent vehicle deviates too far from the global route.Local route planning module is designed to avoid obstacles.It is mainly used to give the controller an optimal control command,which can ensure that the car can move towards the global planning route and keep a safe distance from the obstacles.A dynamic window algorithm is designed to plan the local route in this thesis,which determines a speed range based on the motor performance,and solves the best speed command in the speed range.Finally,it is verified that the intelligent vehicle can track the global route stably and reach the target point under the guidance of the navigation system and multi-dimensional Taylor network optimal control.Finally,the conclusion of this thesis is that the multi-dimensional Taylor network optimal controller can be well applied to the trajectory tracking of intelligent vehicle,and effectively improve the speed,robustness and anti-interference of intelligent vehicle car.