Efficient weld seam detection for robotic welding based on local image processing

Abstract

Purpose The weld seam detection is required for a welding robot to preplan the weld seam track before the actual welding. The purpose of this paper is to investigate this subject in natural lighting conditions. Design/methodology/approach This paper presents an efficient algorithm of weld seam detection for butt joint welding from a single image. The basic idea of the approach is to find a pair of weld seam edges in local area first. Then, starting from the two endpoints of each edge, search for the remnant edge by iterative edge detection and edge linking. Findings The proposed method is insensitive to the variance of the background image and can apply to most shapes of weld seams in butt joint welding. Research limitations/implications The proposed method is designed only for butt joint welding, and it is performed before actual welding. Practical implications The system is applicable to preplan the weld trajectory for most shapes of weld seams in butt joint welding. In addition, the proposed technique may have some potential applications in the field of tailor‐welded blanks. Originality/value The proposed algorithm is based on local image processing and detects the whole weld seam from a single image without giving any initial seam, which is insensitive to the variance of the background image and has low‐computation cost.