Closed-Loop Kinematic Calibration of Robots Using a Six-Point Measuring Device

The low absolute positioning accuracy of robots has limited their applications for many precision tasks. Although the robot accuracy can be improved through kinematic calibration, traditional calibration approaches need either expensive external measurement devices or tedious manual intervention. To overcome these issues, this work proposes a closed-loop kinematic calibration method for robots using a six-point measuring device. This device consists of six displacement sensors to determine the pose of a gauge block as a reference workpiece mounted on the robot end-effector. A comprehensive error model is derived to calibrate the kinematic parameters, including the Denavit-Hartenberg (D-H) parameters of the robot geometry and the pose of the robot base. Then, the Levenberg-Marquard (LM) algorithm is adopted to identify these parameters. An experimental study has been conducted on a UR10 robot, combining a coordinate measuring machine (CMM) to verify the calibration results. It is found that the maximum and average distance errors of the robot movement could be significantly decreased from over 5.60 and 2.50 mm to approximately 0.50 and 0.30 mm, respectively. Benefited from the fact that only six displacement sensors are used, the proposed method is cost-effective. Also, the proposed calibration process can be completed without tedious manual teaching and, thus, easy to achieve autonomous calibration.