通用机器人视觉检测系统的全局校准技术

介绍了基于通用机器人的车身在线柔性视觉检测系统的工作原理和在复杂现场环境下使用激光跟踪仪和测量臂进行全局校准的实用方法。基于测量姿态的全局校准方法利用机械约束将传感器坐标系和靶标坐标系统一起来，实现测量姿态时传感器坐标系与系统基准坐标系的坐标转换。基于机器人运动学模型的全局校准方法利用机器人末端关节位姿可通过正向运动学获知的特点，在现场构建机器人基坐标系，并通过坐标变换获得与机器人运动精度无关的手眼关系，实现了任意姿态时的全局校准。实际测量的结果表明，应用两种校准方法后系统的重复性精度都十分理想，测量装置重复测量标准差和系统重复测量标准差分别为0.07mm和0.13mm，能够满足在线监控白车身加工尺寸变化的精度要求；系统进行坐标测量时的误差分别为±0.2mm和±0.8mm。

Global calibration of visual inspection system based on universal robots

An on-line and flexible visual inspection system for car body-in-white (BIW) based on universal robot and two practical global calibration methods were described. A laser tracker and a measuring arm were used in complexity field environment. The method based on measuring posture realized the coordinate transformation between the sensors coordinates to the system reference coordinate by combining the sensors coordinates and the target coordinates through mechanical constraint. The method based on robot kinematics realized global calibration at arbitrary posture in the presence of robot forward kinematics model, base coordinate frame and hand-eye calibration which was independent of kinematics accuracy. Experimental results show that the standard deviations of repeatability for the measurement equipments and the system are 0.07mm and 0.13mm respectively, which indicates that both of the methods can meet the requirement of monitoring the manufacture dimension changes of BIW. At the same time, the coordinate measuring error of the system are ±0.2mm and ±0.8mm for the two methods respectively.