Osher Yadgar, Sarit Kraus, and Charles L. Ortiz, Jr.
In this paper, we describe the Distributed Dispatcher Manager (DDM), a system for managing large collections of dynamically changing tasks. We assume that task entities are distributed over a virtual space and that teams consist of very large groups of cooperative mobile agents which each have direct access to only local information about their immediate environment. DDM organizes teams hierarchically and addresses two important issues that are prerequisites for success in such domains: (i) how agents should process local information to provide a partial solution to nearby task entities, and (ii) how partial solutions should be integrated into a global solution. DDM’s contributions include: (1) realtime processes for combining partial results to form an accurate global solution, (2) increased system fault tolerance, and (3) scalability to very large task and agent problem domains. We describe extensive experimentation in a simulated domain.