论文部分内容阅读
With the rapid development of intelligent driving vehicles in recent years,the decision-making planning layer and tracking control layer are important components of core technology.The decision-making planning layer uses the on-board computer to combine the output results of the environment perception layer and the navigation layer for action planning and path planning.The task of the tracking control layer is to ensure that the vehicle follows the desired driving path output from the planning layer,and directly control the vehicle’s speed and steering angle.This article will conduct research from two aspects of path planning and tracking control.The specific contents are as follows:(1)First,the necessity of the combined inertial navigation system for intelligent vehicles was verified.We converted the geodetic coordinates in the system output data into local coordinates,and converted the speed to the body coordinate system.Integrated inertial navigation system is used for high-precision map acquisition.The Dijkstra algorithm is used for global path planning,and the global path is generated using a method of fifth-order B-spline curve fitting.Finally,the vehicle speed at each point is determined according to the curvature of the global path.We optimize the sudden change in speed to obtain a global path for the vehicle to travel.(2)First,the path planning problem is transformed into the Lidar coordinate system for processing.The hybrid algorithm is a combination of two methods.The ARA*algorithm is used to plan the dynamic target nodes with sudden changes in heading angle,and the dynamic target points are further determined through screening.Then we improved the traditional artificial potential field method,added the gravity of the preset path,road boundary repulsion,and vehicle steering constraints,and finally segmented the dynamic target points.The comparison between the simulation environment and the real-vehicle lidar map shows the effectiveness of the hybrid algorithm in vehicle local path planning.(3)With the vehicle’s lateral and longitudinal state comprehensive control dynamics model,the relationship between the system state and the control input is obtained,and then the control input is solved in conjunction with the vehicle dynamic constraints to obtain the vehicle’s