Research On Path Planning Of Unmanned Vehicle With Variable Speed Obstacle Avoidance Based On Improved A-Star Algorithm

With the rapid development of Internet technology,the automobile industry is also undergoing profound changes,and the research on unmanned driving technology has become the current hot spot.Path planning and obstacle avoidance is a key technology in unmanned driving.The quality of the path directly affects the quality of autonomous driving.A reasonable path can save time and space cost to a great extent.The complex obstacle environment has higher requirements for the driving of unmanned vehicles.The global path planning needs to be combined with the local path planning in a good way to plan a reasonable passable path and avoid obstacles safely in the complex dynamic environment.This paper analyzes and summarizes the existing path planning algorithms,compares the characteristics of different algorithms in the complex obstacle environment,and puts forward the idea of combining improved A* global path planning algorithm with variable speed obstacle avoidance algorithm on the basis of DWA local path planning algorithm.Simulation experiments and real vehicle experiments verify that the improved algorithm can plan a reasonable passable path.Firstly,through the literature analysis of the current global path planning algorithm and local path planning algorithm,the traditional A* algorithm,artificial potential field method and DWA algorithm are analyzed and the traditional A* algorithm global path planning and DWA algorithm local path planning simulation in MATLAB.Secondly,the global path planning of unmanned vehicle is studied.On the basis of A*algorithm for global path planning,the artificial potential field method and obstacle location information are introduced,and the cost function of repulsive field and the cost function of deviation obstacle are constructed,and the particle swarm optimization algorithm is used to select the parameters of the improved A* algorithm.In this way,the planned path can avoid the dense obstacle area,and avoids the problem that the traditional A* algorithm preferentially moves to the target point instead of avoiding the obstacle,and meanwhile reduces the calculation amount of A* algorithm.The path is smoothed by Floyd method,and the redundant nodes on the path are removed,and the smoothness of the path is improved effectively.The correctness of improved A* algorithm by particle swarm optimization is verified in MATLAB.Then,the local path planning and variable speed obstacle avoidance of unmanned vehicle are studied.Using Kalman Filter algorithm to predict the dynamic obstacles movement and planning the trajectory,in DWA algorithm based on dynamic obstacle course evaluation function is established with dynamic obstacles safe distance evaluation function,then design the unmanned vehicle deceleration speed obstacle avoidance strategy,achieve the purpose of the dynamic obstacle avoidance,improved the security of the unmanned vehicle obstacle avoidance and passage efficiency.The feasibility of improved DWA algorithm and variable speed obstacle avoidance strategy is verified in MATLAB.Finally,the ROS software system of unmanned vehicles is built,the function package of improved A* algorithm and improved DWA algorithm is created,and the feasibility of the function package is verified in Gazebo physical simulation environment.The ROS software system is transplanted into the ROS unmanned vehicle platform to realize the function of global path planning to avoid dense obstacles and local path planning to safely avoid dynamic obstacles.