Robot Systems Completely Based on Sensory Feedback

This paper tries to outline a unified concept for the outer sensory feedback loop in a robot system. The problem of teaching a robot a certain task including continuous sensory information is simplified by introducing several generalizing definitions relating positions/orientations and (pseudo-) forces/torques. Paths are generated via sensor patterns in the teach phase. Motion commands and sensor data are stored together. The latter ones are then available as reference values for the repetition mode in a possibly changing environment. It is shown that the proposed techniques are equally applicable to different kinds of sensors such as force-torque-sensors, range finders, or inductive sensors. The ``sensor-ball'' technique as developed at DFVLR is discussed as one physical realization.