Sensor-Based Real-Time Error Compensation

Abstract

Machining accuracy can be greatly increased through the use of highresolution position sensors, calibration of machine tool or robot components, and other measures which serve to improve the machine’s positional response to part programs. Unfortunately, much of the inaccuracy in finished parts is due to dynamically varying quantities, such as tool wear, chatter, or thermal expansion; random errors such as part misplacement or dimensional tolerancing; errors in the geometric models of the machine; and errors such as backlash, gear eccentricity, and slide nonlinearity. Methods have been developed which rely on sensors to measure these quantities in real time as the part is being machined, and modify the position of the machine tool or robot accordingly. Improvements in absolute accuracy by a factor of twenty have been demonstrated on machine tools. A limit to implementing these methods is in the ability of machine tool and robot controllers to accept real-time sensor feedback.