A direct method in discrete decentralized time-varying control of robot manipulators

Abstract

A simple decentralized time-varying digital trajectory controller for robot manipulators, based on manipulator dynamics and a parameter identification scheme, is developed and tested using multiprocessor architecture and a PUMA 560 robot arm. A discretized equivalent model of the continuous manipulator system is used to design the necessary model-based digital feedback and compensation filters for this purpose. Time schedules of parameters of these filters are generated offline when the desired trajectory is planned. An identification scheme based on output measurements is developed and used to compute the actual values of the model parameters. The nominal parameter values computed offline are used to compensate for timing delays. The performance of the controller is very encouraging and compares very favorably with experimental performance of other controllers, considering that the manipulator was driven very close to its maximum speed.<>