Control of flexible joint robots using neural networks

The objective of this paper is to present a new controller designfor robot manipulators with elastic joints. The model, whichis used to represent the dynamics of elastic joint manipulators,is derived under two assumptions regarding the dynamic couplingbetween the actuators and the links. This model is useful forcases where the elasticity in the joints represents the dominantdynamic characteristic, and especially when it is of greatersignificance than gyroscopic interactions between the motorsand the links. The analysis of the problem is based on resultsin nonlinear control theory and particularly on the feedbacklinearization technique, and the controller design is achievedusing dynamic neural networks.