Inverse Kinematics Neural Approximation and Neural Control of Two-link Planar Robot

Abstract

In this paper is proposed neural control of a two-link planar robot. The control system contains two neural networks. The first of them is off-line training and is used as approximator of the robot kinematics. The second neural network is on-line training and is used as both - neural model and neural controller. Accomplished simulation research confirms sufficient system performance at load variation and links length variation.