Kinematic Robot Calibration Using a Double Ball-Bar

Abstract: Calibration of robots is an essential part of industry today when there is a high accuracy requirement. Several calibration methods have proven to be highly accurate but also expensive and time consuming to use. In this thesis a new calibration method has been developed. The method does not depend on any external measurement systems but instead rely entirely on the robot’s own sensors. The method is based on a closed loop system where a robot is connected to a double ball-bar attached to a rigid point. The ball-bar creates a constraint restraining the robot to a spherical surface. All measured joint positions are fitted to the sphere by altering the geometric representation of the robot. The issues with the method proved to be the process effects as well as dynamic effects influencing the joints and links resulting in incorrectly measured joint positions. Some of these unwanted effects were filtered by applying the clamping method to estimate the joint characteristics. When the new robot geometry was identified, a verification was performed. The verification resulted in an improved positioning of almost 80 % and an improved orientation of 23 % compared to the initial robot geometry.