Research On Autonomous Obstacle Avoidance Navigation Of Multi-AGV Based On Vision

With the rapid development of China’s warehousing and logistics industry,the reduction of manual efficiency and the increase of labor costs,the use of mobile robots to complete the warehousing and logistics work has become an inevitable trend.Mobile robots need to meet the requirements of safe driving while completing logistics,so how to keep autonomous and safe obstacle avoidance in the complex storage environment is the premise and guarantee of completing logistics.How to autonomously avoid obstacles and navigate by robot is a hot research issue in warehousing logistics.However,related researches mainly focus on the autonomous obstacle avoidance of static obstacles,and relatively few researches on the dynamic obstacle avoidance of multi-mobile robots.Therefore,based on Xtion Pro Live vision sensor,this paper explores the obstacle avoidance problem of AGV(Automated Guided Vehicle)autonomous mobile robot in dynamic environment,and studies the detection of static and dynamic obstacles and obstacle avoidance problems of multiple AGVs respectively.The main work of this paper is as follows:For the complex logistics storage environment,firstly,the point cloud image of the storage environment is obtained by depth camera,and the ground plane is separated by RANSAC algorithm to obtain the obstacles on the ground.Secondly,the matching feature points are extracted by orb algorithm.According to the motion parameter model,the background motion is separated by PROSAC algorithm to detect the static obstacles.Finally,in the dynamic environment,due to the motion of the camera,the dynamic obstacle superimposes the motion of the background.The dynamic obstacle detection method based on background motion compensation is studied,and then combined with the traditional frame difference method to extract the dynamic obstacle.Aiming at the static and dynamic obstacle avoidance problem of multiple AGVs working simultaneously,the obstacle avoidance behaviors of multiple AGVs with different conflict types were designed.Aiming at the obstacle avoidance failure caused by the balance of gravitational repulsion force between target point and obstacle,an improved artificial potential field method with distance factor was designed.Based on the dynamic variation of the velocity and direction of the dynamic obstacles,the obstacle avoidance strategy of mutual velocity obstacle method was proposed to realize the autonomous obstacle avoidance of multiple AGVs in the dynamic environment.Finally,the experimental verification of obstacle detection and obstacle avoidance is carried out.The simulation experiment verifies that the dynamic obstacle detection algorithm designed in this paper can accurately detect dynamic obstacles in dynamic scenes.Simulation experiments verify the rationality and feasibility of the multi-AGV station-dynamic obstacle avoidance algorithm,which can accurately avoid obstacles to reach the destination.The research in this paper is of great significance to the cooperative work of multiple AGVs.