Increase of Industrial Robot Accuracy Based on Kinematic Errors Compensation

Abstract

Industrial robots are used in many technical applications from simple pick-and-place tasks to complex machining, welding and assembly applications. The repeatability of the robots is usually high. However, the accuracy is much lower, and it decreases with the robot size selected. Robot calibration represents a possible way to increase accuracy, and two main approaches have been distinguished. Kinematic calibration deals with geometric errors only, and the robot is considered as a rigid body. By contrast, non-kinematic calibration takes into account further sources of errors. This paper deals with kinematic calibration, where an artifact is attached to a robot flange and its position is measured using a laser tracker. The novelty of the method is based on the consecutive rotation of only a single joint, where the artifact trajectory is circular. Real robot geometry is calculated based on identified circles. Numerical simulations seem promising, as well as verification with Staubli TX2-90, where the accuracy was increased more than 43%.