An efficient calibration method for serial industrial robots based on kinematics decomposition and equivalent systems

Calibration of the serial industrial robot with large workspace usually requires a time-consuming measurement task due to the exponential growth of measurement configurations with respect to degrees of freedom (DOFs). To improve efficiency, this paper presents a novel calibration method based on kinematics decomposition and equivalent systems. The trick is to use three lower-mobility sub-robot systems to replace the original robot, where these sub-robots possess quite the same base and end effector to avoid detection of any intermediate frame. For calibration, each sub-robot is treated as a kinematically equivalent system that only contains configuration-dependent joint motion errors, and least-squares support vector regression (LS-SVR) is utilized for joint motion error function approximation. Calibration experiments are conducted on a 6-DOF serial robot ABB IRB 2600. Compared with other methods, the proposed method can significantly save the required measurement configurations and the controller's memory space without losing high calibration accuracy. Experimental results show that the maximum position/orientation errors can be reduced to 0.393 mm/0.038 deg. After calibration, the robot can be applied to assemble parts with small clearance successfully, further demonstrating the effectiveness of the proposed method.