Dexterous Robotics Hands: ANN Based Artificial Muscles Control

Abstract

This manuscript describes a methodology to design a controller for an artificial muscles type four fingers dexterous robotics hand. Electro-Active Polymers (EAP)materials have been used to construct the muscles type four fingers hand. The main dilemma in the control of artificial muscles type fingers, is to provide an accurate fingertips tracking within the operational space. Yet, if bending angles are controlled successfully, fingertips might deviate from a desired position. This is because of finger deflections and vibrations. In this sense, an Artificial Neural Network (ANN)based trajectory planning approach is proposed to calculate modifications in the command values of joint angles for tracking control problem, the position error of the tip of each finger in the operational space is minimized. Study has shown, although artificial muscles robotics hand increases a hand dexterity and allows higher hand manipulability index as compared to moving actuation system, however, accurate control of finger jointing and stuffiness remain vital issues.