无标定手眼视觉装配系统及其应用

Study and application of uncalibrated hand-eye vision assembly system

目的:针对工业机器人自动化生产线的产品装配问题,搭建一套基于手眼(eye-in-hand)系统的工业机器视觉装配实验平台。方法:引入无标定算法完成视觉定位,以获得产品质心的世界坐标,用于完成产品装配的任务。其中以机器人定位到的目标点坐标为理想位置坐标,对视觉定位与机器人定位得到的目标点位置进行比较分析。结果:实验数据显示视觉定位最大误差为0.9 mm。最后将该方法用于产品装配并对系统进行不确定度分析。结论:实验表明产品装配成功率为100%,能满足工业现场产品装配需求。

Aims: Aiming at the product assembly problem of the industrial robot automated production line, an industrial machine vision assembly experiment platform based on the eye-in-hand system was built. Methods: A non-calibration algorithm was introduced to complete the visual positioning to obtain the world coordinates of the product centroid, which was used to complete the task of product assembly. Among them, the target point coordinates located by the robot were taken as the ideal position coordinates, and the visual positioning and target point positions obtained by the robot were compared and analyzed. Results: The experimental data show that the maximum error of visual positioning is 0.9mm. Finally, the method was applied to product assembly and the system was analyzed for uncertainty. Conclusions: The experiment shows that the product assembly success rate is 100%, which can meet the industrial field product assembly requirements.