An Action Decision Mechanism Using Fuzzy-Neural Network in Voice Commanded Fuzzy Coach-Player System for Robots

Abstract

In this paper, research on achieving a cooperation between human and robot is described in the framework of a fuzzy coach-player system, where a human is regarded as a coach, a robot is viewed as a player, and the human speech languages including several fuzzy implications are used as one communication way. In particular, a reasoning mechanism based on fuzzy neural network is proposed to determine the control input to the robot, in order to obtain the adequate amount of the robot action desired by the human in response to the environmental situation. For the training of the network, design parameters in the consequent part are adjusted so as to realize the holding and moving of an object by the robot commanded with the human voice instructions. The effectiveness of the present system is illustrated through experiments for a piling work of colored objects in 3D space