In the production process, customers need to sort and grind various types of steel plate materials. The steel plate has burrs after cutting and has many different specifications, so if the robot is used to teach the workload manually, the customer would like to eliminate the teaching process.The introduction of the 3D vision system, on the one hand, solves the problem of recognising the disordered incoming materials, and is able to quickly locate the workpieces that have a large difference in position, and guide the robot to accurately grasp the workpieces and place them on the grinding table. On the other hand, the 3D vision system automatically extracts the grinding path of the steel plate, which eliminates the workload of teaching and programming the robot, and effectively improves the working efficiency and intelligence of the workstation.