基于机器视觉的机器人装配位姿在线校正算法

Online correction algorithm for posture by robot assembly based on machine vision

为了解决柔性夹具夹取异形零件过程中工件与夹具相对位置具有不确定性的难题,提出基于机器视觉的机器人装配位姿在线校正算法.通过图像预处理及零件表面特征提取,建立工件位姿向量.通过系统建模、误差分析及函数拟合,将工件位姿校正量分解为原始位姿差、旋转引入位姿差及残余位姿差三部分,将三部分位姿差进行线性组合作为零件位姿误差补偿量反馈给机器人,以引导机器人完成装配.为了验证算法的有效性,以涡旋式汽车空调压缩机动盘装配为例,设计机器人手眼装配系统进行实验.实验结果表明,该系统能够保证校正后的零件位姿与目标位姿角度偏差和x及y方向位置偏差分别小于0.6°和0.6 mm,平均装配时间小于20 s,实验过程中装配成功率达到99.67%.

A robot assembly posture online correction algorithm based on machine vision was proposed in order to solve the problem that the relative position between the workpiece and fixture was uncertain in the process of clamping special-shaped parts with flexible fixture.A workpiece pose vector was established by image preprocessing and surface feature extracting.The correction of posture was decomposed into the original posture error,the posture error introduced by rotation and the residual posture error by system modeling,error analysis and function fitting.Linear combination of these three corrections was fed back to the robot as an error compensation for the posture error of the workpiece in order to guide the robot to complete the assembly task.Experiment was conducted by taking the assembly process for the rotating scroll of automotive aircon scroll compressor,and a robot hand-eye assembly system was established in order to verify the effectiveness of the method.The experimental results showed that the angular deviation and the displacement deviation in the x and y directions between the corrected posture and target posture were less than 0.6 degrees and 0.6 mm,respectively.The average assembly time was less than 20 seconds,and the assembly success rate in the experiment was 99.67%.