利用3D打印标定球的机械臂与RGB-D相机手眼标定方法

因为彩色镜头和深度镜头不在同一位置，并且深度图像测量精度差、分辨率低、没有颜色纹理信息，传统的手眼标定方法并不适用于RGB-D相机．本文提出一种利用简单低成本的3D打印球作为标定件对机械臂与RGB-D相机进行手眼标定的方法．本方法只需要测量标定件的3D位置信息，避免使用测量不便、精度稍差的姿态信息．文中给出了该方法的封闭解和迭代优化解．100组仿真结果表明，标定精度与RGB-D相机自身测量精度一致；封闭解不需要机械臂与相机时间同步；迭代优化解的标定精度略有提升，误差最大值和误差方差都很稳定．最后，在7自由度的KUKA ⅡWA机械臂和Kinect相机上做了手眼标定实验，结果与仿真实验一致．总之，本文方法简单可靠，可实现机械臂与RGB-D相机之间的快速部署手眼标定．

Hand-Eye Calibration Method for Manipulator and RGB-D Camera Using 3D-Printed Ball

Traditional hand-eye calibration methods are practically inappropriate for RGB-D cameras, because that the RGB optical center and the depth optical center are not in the same location, and the depth-image is of poor measuring accuracy, low resolution, and no RGB or texture information. A hand-eye calibration method for robotic manipulator and RGB-D cameras using a simple low-cost 3D-printed ball as the calibration object is proposed. Only 3D position measurements of the calibration object are used, which avoids the inconvenience and inaccuracy of orientation measurements. A closed-form solution and an iterative optimal solution are formulated. Results of 100 simulations suggest that the calibration accuracy depends on the measuring accuracy of the RGB-D camera itself, the closed-form solution does not need time alignment between the manipulator and the RGB-D camera, and the calibration accuracy of the iterative optimal solution improves a little bit, while the maximum and variance errors are very stable. Finally, 7-DOF KUKA ⅡWA robot and the Kinect sensor are used to perform hand-eye calibration experiments, and the results are consistent with the simulations. In conclusion, the proposed method is simple and reliable, and can realize the quickly-deployable hand-eye calibration for robotic manipulator and RGB-D cameras.