Simulation of a Kinematic Calibration Procedure that Employs the Relative Measurement Concept

Abstract

Presented in this paper is a project in which an autonomous camera-based calibration system is being developed.As with all other calibration methods, the desired goal is to improve the accuracy of a robot to the same degree asits repeatability. The distinct feature of this system is that it will employ the novel Relative Measurement Concept(RMC) to identify the discrepancies between nominal robot parameters, defined by its specified geometry, and theactual parameters defined by its manufacture. The geometry of a KUKA KR 15/2 was chosen for the simulation asa preliminary experiment was performed with this particular serial robot, however, substitution of other serial robotgeometries is possible. Derivation of the error model will be presented along with discussion on the components ofthe simulation. Program components include Pieper’s solution method to the inverse kinematic problem and SingularValue Decomposition (SVD). Results from the absolute measurement case and the relative measurement case, in itscurrent form, will be presented. The RMC method allows for the identification of 20 of the 24 robot parameters, in itspresent state, and will be experimentally validated with a Thermo CRS A465 six-axis serial robot.