Robust force/motion control of constrained robots using neural net network

Presents a neural net (NN) robust controller for the simultaneous force/motion control of a constrained robot. The method does not require the robot dynamics to be exactly known. Compared with adaptive control, no linearity in the unknown parameters is needed and no persistent excitation condition is required. Compared with other NN approaches, the authors´ method does not require off-line “training phase”. All errors including force, position and weight are all guaranteed to be bounded. The force error and position tracking errors can be reduced to arbitrarily small values by choosing certain large enough gains. Connections of NN control with passivity notions are stated and proved