A net-based approach to the synthesis of nondeterministic robot plans

A technique is discussed for both representing and automatically synthesizing high-level multirobot plans through the use of generalized condition/event (GC/E) nets. An example is given of the functionality of the approach for representing and solving problems of resource allocation in environments containing more than one agent. An important property of nets called persistence, which is essential to multiagent plans, is identified. Persistence, of allowed suborderings of events must be preserved by any net-growing algorithm that produces a plan involving nondeterministic mutual exclusion.<
>