Scaling in Domains with Uncertainty: Criticality-Sensitive Coordination

We consider a team of agents that are required to coordinate their actions in order to maximize a global objective. Our domains are characterized by uncertainty, dynamism, and distributed information. Determining appropriate actions becomes quite difficult, especially as the number of agents and the coupling between them increases. This paper discusses five contributions toward the goal of coordinating agents in large-scale settings: (i) an approach based on identifying the criticality of various activities with respect to their effect on the team reward; (ii) an architecture and implemented coordinator agent that can execute this approach in a distributed manner; (iii) a vast suite of visualization tools that considerably aid the challenging task of monitoring and debugging; (iv) metrics for evaluating system performance in such settings, and (v) a proof of concept of our approach on both focused and randomly-generated experimental domains.