A discrete decentralized hybrid computed torque-model referenced adaptive controller for robotic manipulators

Abstract

A decentralized controller for robotic manipulators was designed and tested using a distributed multiprocessor architecture and a PUMA 560 robot arm. The controller is made up of a nominal component based on a computed torque approach and a variational component based on a model referenced adaptive control approach. The controller is applied to each manipulator joint separately. The motivation for decentralized manipulator control is that the derived algorithm can be implemented using a distributed architecture where each joint is assigned a microprocessor and, thus, the control algorithms for the joints can be executed in parallel using relatively inexpensive hardware. Nonlinear interaction and coupling between joints are treated as a disturbance torque that is compensated for. The function of the developed controller is to drive the manipulator along a desired trajectory.<>