Towards a predictive model for temperature-induced deformation of an industrial robot

摘要

The warm-up and cool-down behaviour of a KUKA KR-15/2 six-axis industrial robot for a particular task is measured and its effect on the robot repeatability determined. The robot is taught two poses. In each pose a tool-flange mounted laser points directly onto the CCD chip of a digital camera. The motion sequence is repeated for 15 hours (900 minutes) three different times at 30%, 75% and 10% of maximum robot speed. The temperature distribution history of the robot during the motion sequence is recorded with a thermal imaging camera. Temperatureinduced robot dimension changes are estimated from the geometry of the camera positions and how the laser spot migrates across the CCD chip over time from start to end. The results from the 30% run are used to estimate linear coefficients of thermal expansion. These empirical estimates are in the same order of magnitude as the coefficient corresponding to the aluminium alloy that is the dominant material comprising the robot links. The empirical coefficient of thermal expansion from the 30% run is used to predict the measured deformations for the 75% and 10% runs.