Motion Planning And Collision Avoidance For Robot In Human-robot Collaboration

With the development of industrial robot technology,robots have been widely used in manufacturing.In the flexible deployment of production lines,the human-robot collaboration is an efficient way to improve productivity.Therefore,it is of great significance to research robot dynamic motion planning and collision avoidance in the human-robot coexistent environment.In this thesis,we study dynamic obstacles detection and avoidance problem in human-robot collaborative assembly.The specific content includes the following four aspects:(1)The operators are the main moving obstacles in the human-robot coexistent environment,a depth camera is used to monitor the workspace of the robot.We proposed a dynamic obstacle detection and tracking algorithm base on background subtraction method.the Euclidean Cluster Extraction is used to determine the point cloud of the moving obstacle closet to the robot and the Kalman Filter is used to predict and update the movement state of the obstacle.(2)Based on the position and velocity of moving obstacles,we proposed a generation of costmap method to represent the spatiotemporal information of obstacles with the Artificial Potential Field algorithm.Then,the path points of the robot's initial path are interpolated to increase the number of path points.Finally,the total potential energy of the initial path is calculated with the costmap.The potential energy is used to estimate the collision risk between robots and obstacles.And it is used to determine the part of the path that needs to be re-planned.(3)Based on the costmap of the moving obstacles,we proposed an improved Rapidly-exploring Random Tree(RRT)based on the costmap of the environment.The new-added nodes are biased according to the gradient of costmap near the tree nodes.It is an effective method to avoid moving obstacles.Finally,the path that needs to be re-planned is used to transform the collision avoidance problem into a local path re-planning problem.(4)In order to verify the performance of moving obstacle detection and robot dynamic obstacle avoidance algorithm in real scenes,we design a human-robot collaborative assembly system based on the Robot Operating System(ROS).The modules,such as sensor acquisition and robot control,are integrated with the obstacle detection and robot collision avoidance method proposed in this paper.Experiment shows that the robot can adjust the initial path based on the spatial information of the moving obstacle and complete the original task.