Abstract: To achieve the structured light vision guided welding robot system calibration, and solve the problem of complex, and difficult to make target in the existing calibration methods, a new self-calibration technique based on active vision is proposed. This self-calibration technique calibrates the camera inner parameter, rotating section of hand-eye matrix by controlling the robot 5 pure translational motions. And it calibrates the translation section of hand-eye matrix and the equation of the light plane by controlling the robot 2 motions with revolving with three feature points in the scene without a specific target. The actual test results in the structured light vision guided welding robot system based on Denso robot show that the calibration approach is stable, and can reach the precision of ±0.93 mm. This self-calibration method to the structured light vision guided welding robot system with easy feature selection is simple, and has significance to the actual use in the field of industry.

Key words: welding robot, laser vision sensor, eye-in-hand system, structured light, active vision, light plane calibration

摘要： 为实现结构光视觉引导的焊接机器人系统的标定，解决现有标定方法复杂，标定靶标制作要求高等缺点，提出一种基于主动视觉的自标定方法。该标定方法对场景中3个特征点取像，通过精确控制焊接机器人进行5次平移运动，标定摄像机内参数和手眼矩阵旋转部分；通过进行2次带旋转运动，结合激光条在特征点平面的参数方程，标定手眼矩阵平移部分和结构光平面在摄像机坐标系下的平面方程；并针对不同焊枪长度进行修正。在以Denso机器人为主体构建的结构光视觉引导的焊接机器人系统上的测试结果稳定，定位精度可达到±0.93 mm。该标定方法简单，特征选取容易，对焊接机器人系统在实际工业现场的使用有重要意义。

关键词: 焊接机器人, 激光视觉传感器, 手眼系统, 结构光, 主动视觉, 光平面标定