为解决螺栓自动化装配过程中存在的零件位姿快速获取和在弱纹理、反光等复杂场景中实现精确定位等难题,基于双目结构光视觉原理,研制了一套螺栓自动装配系统.首先,通过双目结构光视觉系统获取工件场景的三维信息,涉及光栅条纹图案的获取、双目标定、多频外差法求解绝对相位、相位校正、极线校正、立体匹配,接着采用欧式聚类对重建的点云数据进行分割,获取单个螺栓的三维点云聚类,然后利用迭代最近点(ICP)算法估计模板点云到目标点云的位姿变换矩阵,识别连接螺栓在空间中的位姿,最后结合机械臂末端进行自动装配拧紧.实验结果表明,整个装配过程可自动完成,单个螺栓拧紧装配时间约为10 s,系统测试平均轴孔对中误差不超过1.20 mm,误差标准差不超过0.11 mm.
Design of bolt automatic assembly system guided by binocular structured light vision
To solve the problems of rapid acquisition of parts positions in automatic bolt assembly process and accu-rate positioning in complex scenes such as weak texture and reflection,an automatic bolt assembly system was de-veloped based on the binocular structured light vision principle.The 3D information of the workpiece scene was ob-tained by binocular structured light vision system,which involved the grating fringe pattern acquisition,binocular calibration,multi-frequency heterodyne method to solve the absolute phase,phase correction,polar line correction and stereo matching.The reconstructed point cloud data was segmented by European clustering,and the 3D point cloud clustering of single bolt was obtained.Then,the pose transformation matrix from template point cloud to tar-get point cloud was estimated by ICP algorithm,and the pose of the connecting bolt in space was identified.Finally,the end of the manipulator was automatically assembled and tightened.The experimental results showed that the whole assembly process could be completed automatically,and the assembly time of a single bolt was about 10 s.The average axle hole alignment error of the system test was less than 1.20 mm,and the standard deviation of the error was less than 0.11 mm.
NETL
NSTL
万方数据
