In this paper, a control algorithm based on neural networks is presented. T
his control algorithm has been applied to a robot arm which has a highly no
nlinear structure. The model based approaches for robot control (such as th
e computed torque technique) require high computational time and can result
in a poor control performance, if the specific model-structure selected do
es not properly reflect all the dynamics. The control technique proposed he
re has provided satisfactory results. A decentralised model has been assume
d here where a controller is associated with each joint and a separate neur
al network is used to adjust the parameters of each controller. Neural netw
orks have been used to adjust the parameters of the controllers, being the
outputs of the neural networks, the control parameters. (C) 1999 Elsevier S
cience Ltd. All rights reserved.