针对传统的机器人柔性坐标测量方法中，机器人模型不完善及机器人固定参数不断变化导致测量精度难以提高的问题，提出一种基于双目视觉原理的全局实时校准方法，组建由两台相机组成的高精度全局校准单元，通过测量布置在机器人末端视觉传感器上的控制点阵，实时得到机器人末端的空间位姿，实现机器人在全局空间的精确定位。提出基于空间网格控制场的相机校准方法，构建像面坐标系上的残差库，实现相机在全视场空间内的高精度校准。实验表明，采用上述方法可实现±0.1 mm的双相机校准精度，整个系统的测量精度可达±0.15 mm，从根本上摆脱了机器人运动学模型及参数误差带来的影响，有效地保证了柔性坐标测量系统的精度。

To solve the problem of the traditional flexible measuring system due to the imperfect robot module and the changing parameters, a calibration technology of high-precision flexible coordinate measurement system is proposed. Binocular stereo vision method is employed to construct the global measurement unit. In order to ensure the accuracy, a camera calibration method which is based on space grid control field, and a method of the global space precise positioning are proposed. The pose transformation of the robot in space can be accurately measured by the two global cameras in the scene, through measuring the controlling points on the stereovision sensor. The experimental results verify that the error of the global camera is less than ±0.1 mm, and the error of the whole measurement system is smaller than ±0.15 mm. The measurement can satisfy the precision requirement of the system.