Skip to Main Content
The control of rehabilitation robotic systems on a high level of abstraction requires the integration of task planning capabilities. Due to the close man-machine coupling within this kind of robotic systems the required "task planner" is equivalent to a discrete event controller. This paper describes the design of such a controller as well as its integration into an underlying software control architecture. The design is based on the enhanced assembly planning methods that use AND/OR-nets for high abstraction level task knowledge representation. The task plans generated on this abstraction level are decomposed with the help of generic Petri-nets so that a sequence of system executable operators emerges. The direct integration of man-machine interactions as well as error correction steps within these generic Petri-nets makes the task plan execution robust and predictable even in uncertain environments.