Assistance Robotics: the role of human-like modes of control

Daniel R Mestre¹, Yves Rybarczyk¹, Philippe Hoppenot², Etienne Colle²

¹Mouvement & Perception Lab (LMP), CNRS & University of the Mediterranean, Marseille, France
²Complex Systems Lab (LSC), CNRS & University of Evry, Evry, France
E-mail: mestre@laps.univ-mrs.fr

The overall framework of this project is to restore autonomy to disabled people by increasing their field of intervention. This involves a teleoperated system, composed of a mobile teleoperated robot-mounted arm and a remote control station. The required objective for the design of a "non-totally autonomous" machine, controlled and supervised by a human operator, is its adaptation to the user's capabilities. According to this logic, the ideal system should perfectly fit into the human sensori-motor capabilities. The system would disappear from the field of consciousness and the operator would use it as a "natural" extension to his/her own body (hence the concept of natural interface).

Concretely, we designed an experimental task, in which the operator had to control remotely the displacements of the robot, using a monitor linked to a camera on-board the mobile base and a keyboard interface. First, we implemented a functional, human-like, speed-curvature trade-off on the mobile base. Secondly, we tested the role of a mobile camera on the robot, controlled by the operator. Experimental data show that, both on the teleoperated device side and on the operator's interface side, the implementation of human-like modes of control result in easier and smoother control of the robot trajectory.