STATIC AND VIBRATIONAL KINEMATIC PARAMETER ESTIMATION FOR CALIBRATION OF ROBOTIC MANIPULATORS

Kinematic positioning accuracy has been shown to be an important performance criterion for robotic manipulators. Estimation of the static and vibrational kinematic parameters of robotic manipulators provides means for (1) increasing the absolute static positioning accuracy of the manipulator without the expense of tighter tolerances on manipulator components, and (2) tracking the vibrational performance of the manipulator's kinematic parameters throughout its workspace with significant savings in both experimental and computational time over multiple-posture modal analysis. From experimentally obtained data, the estimation algorithms synthesize sets of "optimal" kinematic parameters using a Marquardt nonlinear least-squares optimization scheme. The optimal parameters are those which minimize error between the end effector pose predicted by a manipulator's internal kinematic model and the actual end effector pose. Both the static and vibrational algorithms have been successfully tested using data collected from a General Electric model A4 industrial robot. The most important considerations in interpreting the experimental results were found to be (1) presence of redundant parameters in the test model, and (2) estimation of the efforts of input forcing direction on structural response. Both parameter estimation techniques have proven to be viable methods in the laboratory, and are being refined for practical industrial use.