Control of flexible joint robots using neural networks

Abstract

The objective of this paper is to present a new controller design for robot manipulators with elastic joints. The model, which is used to represent the dynamics of elastic joint manipulators, is derived under two assumptions regarding the dynamic coupling between the actuators and the links. This model is useful for cases where the elasticity in the joints represents the dominant dynamic characteristic, and especially when it is of greater significance than gyroscopic interactions between the motors and the links. The analysis of the problem is based on results in nonlinear control theory and particularly on the feedback linearization technique, and the controller design is achieved using dynamic neural networks.