Model reference control for human-operated robots and application to a robotic manipulator

Abstract

As fully automated robotic systems are difficult to realize from the viewpoint of cost efficiency and currently available technologies, human-robot cooperative systems have been widely studied. This paper presents a model reference control approach for a human-operated robotic system. This approach provides functions to incorporate commands given by a human operator and compensate for mistakes committed by the operator in real time, and is applied to a manipulator control whose objective is to transport an object on a table at the end-effector. The proposed method provides the following three functions: collision avoidance, tumble avoidance of the transported object, and singular point avoidance, even when the human operator makes mistakes during operation. There are many applications of this type of robotic manipulator in industrial and household tasks. The effectiveness of the proposed method is confirmed by the experiments in which several unskilled operators operate the manipulator to transport an object in a narrow space.