Estimating dynamic positioning errors of coordinate measuring machines

Abstract

Modern industrial production requires fast and automated quality control using state-of-the-art surface metrology sensors embedded in fast high-precision measuring machines. In order to achieve both fast and precise positioning, active control systems with model-based compensation of dynamic positioning errors due to increased acceleration forces can be applied. However, these active control systems require an accurate estimate of the dynamic positioning errors of the tool-center-point (TCP) with respect to the precisely measured position of the drive axes. A novel optical camera sensor system with high subpixel precision based on multi-spot detection enables the direct measurement of the TCP position at a slower rate than the sampling time of the control system and with significant latency or dead-time. Based on a general yet simple deviation model with large model mismatch, three approaches to estimate the TCP position are presented and compared with simulation and test bench results of a modified Mahr MFU 100 measuring machine. First, a multi-rate Kalman filter with delay compensation is designed based on a simple physical modal model generalizing the applicability of the concept for different types of measuring machines and machine tools. Second, additional sensors in form of accelerometers placed at the TCP are used to obtain an indirect measurement of the TCP position at a fast sampling rate to reduce the effect of the model mismatch. Third, instead of additional sensors, an alternative concept consisting of enhanced Gaussian process modeling to improve the model accuracy with a data-based error model is incorporated in the Kalman filter framework in form of fast pseudo-measurements outperforming the other approaches.