A Hybrid Neural Network/Newton-Raphson Algorithm Solving the Inverse Kinematics Problem (original) (raw)

Abstract

This paper describes the implementation of a hybrid neural network / Newton-Raphson algorithm to produce the inverse kinematics solutions. Such an algorithm is required in order to administer a variety of robot tasks on different robot platforms. The system is based on the combination of neural networks and the parallel processing capabilities of the Inmos transputer. The exploitation of both architectures results in more efficient and fast numerical solutions. Positioning accuracy experimental results from two robots are presented. Keywords: Neural networks, Newton-Raphson, flexible control, inverse kinematics, parallel processing, transputers INTRODUCTION

Kostas Tarchanidis hasn't uploaded this paper.

Let Kostas know you want this paper to be uploaded.

Ask for this paper to be uploaded.