A novel initial weld point guidance and path planning method for robot welding based on structured light vision

Aiming at the problems of low efficiency and poor adaptability of teaching-and-playback and offline programming. This paper proposed a method for positioning initial weld point (IWP) and planning welding trajectory based on structured light vision. Firstly, a camera is fixed at the end of the robot to capture images of the welding workpiece from two specific positions. An improved Yolov8 algorithm is employed to recognize IWP, achieving approximate positioning of IWP with "monocular vision dual-position shooting." Subsequently, the position of the camera relative to IWP is determined based on the structured light plane. Image processing is then conducted to extract the structured light feature points, and the feature points are fitted to obtain groove planes and weld seam line, thereby planning the overall welding pose. Experimental results demonstrate that this method can effectively complete the recognition of IWP, the extraction of weld seams, and trajectory planning tasks, which is of positive significance in improving the welding efficiency and adaptive ability of robots in various welding environments and workpiece conditions.