Distributed Continual Planning for Unmanned Ground Vehicle Teams

Some application domains highlight the importance of distributed continual planning concepts; coordinating teams of unmanned ground vehicles in dynamic environments is an example of such a domain. In this article, I illustrate the ideas in, and promises of, distributed continual planning by showing how acquiring and distributing operator intent among multiple semiautonomous vehicles supports ongoing, cooperative mission elaboration and revision. It is this longer-term view that motivates the use of planning such that an agent should decide between alternative anticipated sequences of activities; otherwise, the application might be better served with simpler reactive agents that only decide on their very next actions. Second, what the agent knows about the application domain, or what the agent's objectives are, or both, can change over time. Information about the domain could be revealed incrementally or could dynamically change in ways outside the agent's control, and thus, the agent should continually reevaluate its ongoing plans and revise or elaborate them to accommodate the changes.