A robust neural network (NN) controller is proposed for the simultaneous fo
rce/motion control of constrained rigid robots. The NN weights here are tun
ed on-line, with no off-line learning phase required. Most importantly, we
can guarantee the boundedness of constraint force errors, joint position tr
acking errors, and NN weights. When compared with adaptive controllers, we
do not require linearity in the unknown parameters, and the tedious computa
tion of the regression matrix. Novel passivity properties of the NN control
ler are stated and proven. (C) 1999 John Wiley & Sons, Inc.