Market-based Coordination Strategies for Large-scale Multi-Agent Systems

This paper studies market-based mechanisms for dynamic coordinated task assignment in large scale agent systems carrying out search and rescue missions. Specifically, the effect of different auction mechanisms and swapping are studied. The paper describes results from a large number of simulations of homogeneous agents, where by homogeneous we mean that agents in a given simulation use the same strategy. The information available to agents and their bidding strategies are used as simulation parameters. The simulations provide insight about the interaction between the strategy used by individual agents and the market mechanism. Performance is evaluated using several metrics: mission time, distance traveled, communication and computation costs, and workload distribution. Some of the results obtained include: limiting information may improve performance, different utility functions may affect the performance in non-uniform ways, and swapping may help improve the efficiency of assignments in dynamic environments.