Adaptive position-velocity-force control of two manipulators

Abstract

The motions of two manipulators operating in leader-follower mode are controlled by means of adaptive controllers of self-tuning type. The controllers are designed on the basis of the stochastic multi-variable discrete time model (ARX) with the external inputs in which the parameters are estimated on-line recursively, and a quadratic criterion. The resulting controllers are adaptive with time-varying gains. They also depend on predicted errors of both manipulators. A higher level controller implementing the control laws needs complete information about the two co-operating manipulators: their complete state (position, velocity, force), and the current parameter estimates in the ARX-models. The structure of the adaptive controllers is illustrated with an example.