Compensation of Industrial Manipulator Dynamics in the Presence of Variable Payloads

Abstract

The objective of this study was to experimentally investigate the payload sensitivity of model-based control algorithms for the class of vertically articulated industrial manipulators with high torque amplification drive systems. Once the pay load sensitivity was established, the minimum subset of payload parameters required to permit payload invariant tracking performance was determined. Analysis of open-loop feedforward compensation torques correctly predicted the experimental PUMA results. High amplification drive sys tems reduce, but do not eliminate, the payload sensitivity of industrial manipulator dynamics. Model-based controllers must adapt to payload variation in order to maintain high speed tracking accuracy. For a wide range of end-effector payload, the payload estimation/adaptation algorithm for the large positioning links need only compensate for variation in payload mass. To accurately adapt to changes in small link dynamics, the algorithm must estimate both the payload mass and centroid. Knowledge of payload sensitivity can re duce the complexity of all model-based adaptation schemes and bring them a step closer to routine implementation.