A recurrent neural network for manipulator inverse kinematics computation

Abstract

A recurrent neural network is presented for the computation of inverse kinematics for redundant robot manipulators. The proposed recurrent neural network is based on a reflexive generalized inverse problem that simplifies the computation of pseudoinverses by reducing the number of matrix equations needed to be solved and the complexity of the physical implementation. The proposed recurrent neural network is shown to be asymptotically stable and is used to solve the inverse kinematics problem for a three degree-of-freedom planar redundant manipulator.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>