3D curve weld seam path and posture planning based on line laser sensors

Abstract

The range of welding scenarios is expanding and becoming more complex in line with the development of manufacturing, which has led to increased demand for automatic and intelligent welding solutions. Line laser sensors are a crucial tool and technology in achieving intelligent welding. However, with the rise in the diversity of weld types and welding parts, the increasingly complex welded components pose challenges to point cloud construction. In contrast, the fixed posture of the welding torch in conventional digital welding makes it difficult to meet the welding needs of 3D curve weld seam. Therefore, this paper proposed a point cloud construction method based on robot pose. Firstly, the multi coordinate system transformation relationship was solved, and precise coordinate transformation between multi frame point clouds was achieved through coordinate transformation matrix, constructing a complex weld seam point cloud model. Furthermore, the point cloud was processed to extract weld seam information with the Ransac algorithm. Based on this, considering the characteristics of robot motion, the welding torch posture is divided into three components: deflection angle, elevation angle, and rotation angle, each of which is calculated separately to achieve welding posture planning. Experimental results have shown that the accuracy of the point cloud construction method proposed in this paper is better than 0.2mm, and the planning errors of the three posture angles are 0.75°, 1.2°, and 0.28°, which aligns well with the requirements of practical welding operations.