A Novel Means of Software Compensation for Robots and Machine Tools 2006-01-3167

Current methods of machine calibration and software compensation focus on either the joint motion errors (classic machine tool software compensation) or the geometric errors between the joints (robot calibration). However, both types of errors have a significant impact on the volumetric accuracy of a machine tool or robot. We have developed a calibration method that simultaneously identifies joint motion errors and geometric errors in a machine or robot with an arbitrary number and arrangement of links using a laser tracker. The simultaneous identification of all error sources decreases measurement time, with a typical calibration for a moderate sized machine taking about four hours and 200-500 measurements.

The model presented is based on a mathematically minimal parametric model of the machine. Parameter identification is done in a statistically significant way, resulting in both the “best-fit” values for the parameters and the statistical confidence in those values. The results provide a very accurate assessment of the volumetric accuracy of the machine. Compensation is done through a simple, rapidly converging, iterative algorithm that is being implemented real time in the Siemens 840D controller. This paper covers the development of the model, identification of the parameters, and implementation of the software compensation scheme. Examples are given of implementation including: a high-speed five-axis machine tool and a low cost five-axis drilling system. Both machines showed significant improvements to accuracy (approximately 80%) after software compensation, with volumetric accuracy approaching machine repeatability.