Computer vision body modeling for gesture based teleoperation

Abstract

Dependable robots and teleoperation, taken in its broadest sense, require natural and friendly human-robot interaction systems. The work presented consists of a methodology for human-robot interaction based on the perception of human intention from vision and force. The vision system interprets human gestures from the integration of a stereovision and a carving system, from which it extracts a model of the human body when a person approaches the robot. The interaction can be performed by contact as well, from the perception of the forces applied to the robot either through a force sensor on the wrist or a sensing skin. The perception of human intention makes possible an intuitive interaction to modify on line the robot trajectory when required. © 2007 Springer-Verlag Berlin Heidelberg.