Abstract: To meet the needs of welded seam milling automation,a robotic welded seam milling guidance method based on surface structured light vision detection was proposed.The point cloud data and grayscale image were processed accordingly to extract the 3D and 2D contour volumes in the field of the vision,and completes the segmentation and recognition of the weld butt joint and the welded seam body by combining 2D/3D geometric constraints and 2D image HOG+SVM methods.Further,the welded seam path was calculated based on the welded seam body point cloud data,the final robot milling path was generated through the hand-eye calibration relationship calculated based on the correspondence of point sets and the process requirements of welded seam milling.The experiments showed that the proposed method could adapt to welded seam detection under various adverse conditions such as joint mismatch and base metal deformation,and could better complete robot automatic milling of actual welds.It had the characteristics of fast extraction speed,high detection accuracy and strong applicability of path planning.

Key words: surface structured light visual inspection, weld seam inspection, milling path planning, automatic milling

摘要： 为满足焊缝铣削自动化的需要,提出了一种基于面结构光视觉检测的机器人焊缝铣削导引方法。该方法首先对面结构光视觉采集的点云数据和灰度图进行处理以提取视野中的3D和2D轮廓体,并通过结合2D/3D几何约束以及2D图像HOG+SVM方法完成焊缝对接接头与焊缝本体的分割识别,进一步基于焊缝本体点云数据计算焊缝路径,并通过基于点集对应性计算的手眼标定关系和焊缝铣削的工艺要求生成最终机器人铣削路径。实验表明,该方法能够适应接头错位和母材变形等多种不良工况下的焊缝检测,并能够较好的完成实际焊缝的机器人自动铣削,具有提取速度快,检测精度高,路径规划适用性强等特点。

关键词: 面结构光视觉检测, 焊缝检测, 铣削路径规划, 自动铣削