A vision-based method for narrow weld trajectory recognition of arc welding robots

With the rapid development of the manufacturing industry, the demand for autonomy of welding robots is also improving. To improve the autonomy of welding robots, the first problem to be solved is the identification and positioning of the weld seam. In general, it is a challenge to extract narrow weld seams in workpieces with characteristics such as no texture, smoothness, and strong reflection using passive vision sensors. In this paper, we propose a vision-based method for the 2-dimensional (2D) and 3-dimensional (3D) detection and localization of narrow weld seams to improve the sensing capability and automation of welding robot systems. The method enhances narrow weld seam features by adjusting the image grayscale expectation at the time of shooting to achieve weld seam recognition within the field of view. Then, the point cloud of the weld area is obtained using the triangulation technique of stereo vision to realize weld seam localization. Finally, the calculated weld trajectory is matched with the trajectory extracted from the workpiece model to realize recognition of welding tasks. Experiments were conducted on ferrous and galvanized workpieces, and the final experimental results demonstrate the effectiveness of the proposed method.