Automated Grinding and Polishing System for Carbide Spots

This study presents a robotic automated polishing and grinding system for metals based on a depth camera. The system captures images of the metal component using a depth camera and utilizes visual recognition to identify carbide spot defects. Simultaneously, the coordinates of each carbide spot in the 3D point cloud are transformed into the robotic arm's coordinate system using perspective transformation, ensuring stable conversion of carbide spots into arm coordinates. This guides the robotic arm to perform precise grinding operations, ensuring accurate defect repair or removal and achieving automated robotic polishing. We detected 903 carbide spots across 100 metal components, with an average maximum error of 0.95 millimeters. The system boasts high precision and efficiency, with experimental results demonstrating a welding spots recognition accuracy of up to 99%, and a margin of less than 3 millimeters. This system not only enhances production efficiency and quality but also realizes the goal of intelligent manufacturing.