The solution of the inverse problem of kinematics of robots using neural networks

Abstract

Solving of the inverse kinematics problem due to neural networks applicable to robots is examined. It has shown, that the same neural nerwork can be effectively used for numeric position, velocity and acceleration transformation, during the robot drives sample time. That makes possible a feed-forward joint position control with improved control performance