Acceleration feedback implemented through the utilization of an observer as applied to a robotic manipulator

Abstract

It is difficult to measure all of the coupling forces and torques acting on a robot manipulator. Real-time calculations of the coupling and nonlinear parameters utilized to formulate an estimate of the forces and torques is also difficult. If these coupling terms are treated as disturbance inputs to each joint, acceleration feedback can be utilized to increase the systems stiffness and reject these disturbances. Most industrial robots do not have acceleration feedback transducers. Many robots do not have velocity feedback as well. This paper presents the results of the investigation into the use of observers to provide an estimate for both velocity and acceleration feedback signals. Following a review of previous work, this paper presents the theory and simulations of various observers. The observers were then implemented and tested on two experimental stands: a single axis motor/drive with a programmable load and a six axis Puma robot. The observers were implemented in both software and hardware. The results indicate that it is possible to improve the disturbance rejection capabilities of a motion controlled system with an acceleration signal generated by an observer.