REPRESENTING AND ANALYZING ACTION PLANS AS NETWORKS OF CONCURRENT PROCESSES

Constructing action plans for a robot operating in an environment containing uncertain and dynamic events is a difficult task. Indeed, the inadequacy of the standard approaches for representing and producing plans for such environments has led some researchers to abandon explicit plan representation and to directly program behavior instead. Ultimately, no matter which approach is taken, producing appropriate behavior in such environments requires writing robot programs or action plans that include conditionals, loops, requests for sensory data, concurrency, etc. Existing approachs are simply not adequate to the task of modeling and analyzing such plans. Nonetheless, it is necessary that such plans be open to formal analysis: the complexity of control necessary to operate in uncertain and dynamic environments demands that more than human intuition be used to verify, or preferably autogenerate, such plans. The problem of constructing a plan representation that can deal with the complexity of representing and analyzing robot behavior in uncertain and dynamic environments is addressed. The key contributions are as follows. A concurrent-process based representation is developed which represents both the plan (or controller) and the uncertain and dynamic environment in which the plan operates. A methodology is outlined for analyzing the behavior of this interacting system of plan and world. This methodology is illustrated with a mixed-batch example from the domain of robotic kitting. To balance the theoretical work, a description of the implemented robot kitting cell is presented.