Automatic action planning as a key to virtual reality based man-machine-interfaces

Experiences in space robotics showed, that in a field where the need for autonomy of automated systems is traditionally very high, there is also a great concern about operability and supervision efforts. In order to give a user trust in autonomous operation of a system, he must be provided with convincing information concerning the conduct of the tasks to be done and the status of the-potentially-different agents working. In this paper, it is shown, which planning and task deduction capabilities a virtual reality system must have to be well suited for the control and supervision of an autonomous multi-robot-system for space-laboratory servicing. Furthermore, it is emphasized, that the usefulness and operability of a virtual reality based man-machine-interface can be maximized by employing automatic action planning techniques to automatically "project" actions carried out by the user in the virtual environment into the physical environment with the help of robots or other means of automation ("projective virtual reality").