Human-guided robotic manipulation: Theory and experiments

Abstract

Emerging applications of robot systems that involve close physical interaction with human have opened up new challenges in robot control. For these applications, it is important to consider the stability and coordination of human-robot interaction. While various control techniques have been developed for human-robot interaction, existing methods do not take the advantages of human ability in responding and adapting to unknown environment. In this paper, a human-guided manipulation problem which is able to take advantages of both the human knowledge and the robot's ability, is formulated and solved. The workspace is divided into a human region, where human play a more active role in the manipulation task, and a robot region, where the robot is more dominant in the manipulation. The proposed formulation allows the involvement of human control action to deal with unforeseen changes or uncertainty in the real world. We present a theoretical foundation that allows the stability and coordination of the human-guided manipulation problem to be analyzed. Based on the human region and the robot region, an adaptive tracking controller is developed. Experimental results are presented to illustrate the performance of the proposed control method.