A robust method for vision-based seam tracking in robotic arc welding

This paper presents a robotic seam tracking system which is aimed at achieving robustness against some welding noises such as arc glares, welding spatters, fumes etc. In particularly, a syntactic analysis is used to improve the extraction reliability of the joint features. The joint features thus obtained are used to extract the 3-dimensional information of the weld joint and then achieve the robot path correction. To show the performance of the developed system, a series of experiments on joint feature detection and robotic seam tracking are conducted for different types of weld joints. The results exhibit that the system is very robust to various welding noises as well as variations in appearance of weld joint and workpiece.