Burr inspection of the inside of high-pressure pipe joints by motion imitation of skilled worker

Abstract

Burrs occur with high probability in the high pressure pipe joints used in construction machinery such as power shovels and bulldozers. The burr is checked by skilled workers on all joints so far. The automated system of burr detection is urgently needed because the checking all burrs manually is very burden for the skilled workers. In this system, the lighting, posture, and position of the camera have a significant effect on the detection accuracy. Based on the hint that skilled workers inspect the burr effectively, this study imitates the way of skilled workers to lighten the joints during burr inspection using a robotic arm, and searches the position and the posture of the light to detect the burrs with high accuracy.