Incremental Negotiation and Coalition Formation for Resource-Bounded Agents
Charles L. Ortiz, Régis Vincent, Eric I. Hsu, Bruno Dutertre, Barbara J. Grosz · 2004
This report describes SRI's contributions toward the solution of the problem of real-time distributed resource allocation. For the most part, the distributed multi-sensor challenge problem utilized by the Defense Advanced Research Projects Agency's Autonomous Negotiating Teams (ANTs) program was used to motivate the research. However, the contributions are not restricted to the sensor domain. We model the resource allocation problem as a multiagent problem in which each resource is modeled as an agent which can communicate with other agents to exchange requirements or task information. Agent interactions generally take the form of message exchanges to support auction-style algorithms in which a mediator requests bids on a task or a collection of tasks and then receives bids from agents. Each bid encapsulates local information, important to the allocation decision, in the form of utility or cost estimates.