Research On AGV Path Planning Algorithm For Autonomous Parking Considering Safety Margin In Parking Lots

The application of automatic parking AGV in parking lots can save the time for car owners to park and pick up their cars and improve the operation efficiency of parking lots.In order to improve the safety of parking AGV in engineering projects,this paper uses the safety margin layer and the dynamic obstacle layer to improve the map,and then affects the path planning.The improved method is not limited to specific environment and robot type,but also applicable to other mobile robots.The main content of the paper is as follows:(1)In the global path planning,the concept of safety margin layer was proposed in order to make the self-parking AGV far away from static obstacles and plan a shorter path as far as possible.The margin layer is a functional map in the multi-layer cost map.The multi-layer cost map divides the traditional grid map into multiple functional maps,which are updated and mapped to the total costmap respectively to meet different needs.The heuristic function of the traditional A-star algorithm only considers the distance between the current node and the starting point and the distance between the node and the target,and does not take the distance of obstacles into consideration.In this paper,the cost map around obstacles is set up in the form of safety margin layer.The greater the cost value is,the higher the collision risk is.Then,the cost value is included in the heuristic function of A-star algorithm.The cost of this layer satisfies the normal distribution,and the more closer the robot is from the obstacle,the greater the cost will be,and the robot will be guided away from the obstacle without generating too far path.(2)In order to improve the security of the system in a complex environment,a dynamic obstacle layer is added to guide the AGV to avoid moving obstacles.Firstly,the Kalman filter is used to predict the next dynamic obstacle position.Then take the midpoint of the line between the current and predicted position of obstacles as the center and use two-dimensional Gaussian distribution as the cost value function to set the map.Finally,the map is used to guide the path planning algorithm.In this paper,the cost value of the predicted position of the obstacle is set to 254,that is,impassability,and then the coefficient of the cost function is backward deduced.The faster the speed of the obstacle,the higher the cost function coefficient,the greater the grid cost value around the obstacle,and the greater the collision risk.Therefore,the dynamic obstacle layer can well represent the increase of collision risk brought by the increase of obstacle speed on the map,thus affecting the path planning algorithm.(3)Finally,in order to show the improvement of the above algorithm,simulation was carried out on the ROS platform under Ubuntu system,and verification was carried out on the mobile robot platform in the laboratory.The experimental results show that the proposed margin layer can effectively guide the robot to create a path away from obstacles,while avoiding the problem of planning a path too far.The dynamic obstacle layer can guide the robot to plan in advance to avoid moving obstacles.In the case of slower updating of global planning algorithm or faster moving of obstacles,the dynamic obstacle layer can guide the robot to pause in advance and wait for re-planning.It shows that the system is very secure.