Development Of Weld Grinding Robot System

Weld seam grinding is an important task for industrial robots on the production line.For small and medium-sized enterprises,the characteristic is that the factory buildings are small and it is impossible to establish a complete industrial production line.Its main task is to process some small-batch,multi-type and multi-size artifacts.And the grinding task is manually completed by the workers.Therefore,it is of great significance to develop a weld grinding robot system applied in small and medium-sized enterprises.In this paper,the weld grinding task of metal cabinet is taken as the research object,and the two-dimensional laser profiler and UR5 robot are selected to develop the weld grinding robot system.Firstly,the robot pose data and laser data are read using the moveit module and the message mechanism of the ROS operating system.For the mismatch of two kinds of data in time,a data synchronization algorithm based on queue caching strategy is proposed to complete the time synchronization of two kind of data and improve the accuracy of point cloud model.Secondly,the target plane normal vector and the weld direction vector are extracted based on the plane extraction algorithm and the line extraction algorithm.For the problem that the noise point cloud affects the weld height extraction,a method combining the height layering and histogram statistics is proposed to extract the height information of the weld.The extracted information is used to adjust the pose and motion direction of the robot end so as to ensure the reliability of the automatic scanning.The "segment identification" strategy is used to scan all of the weld information and complete the automatic grinding of the weld.Finally,a simulation experiment platform based on the ROS system's Gazebo simulation environment is built to verify the accuracy of each part of the algorithms.A real experimental platform is built to verify the polishing performance of the system.