Discovering and Exploiting Synergy Between Hierarchical Planning Agents

Contents

Agents interacting in a multiagent environment not only have to be wary of interfering with each other when carrying out their tasks, but also should capitalize on opportunities for synergy. Finding overlapping effects between agents plans can allow some agents to drop tasks made unnecessary by others’ actions, to reduce the cost of execution and improve the overall efficiency of the multiagent system, thus creating synergy. In this paper we define criteria for finding synergy and develop algorithms for discovering synergy between planning agents that exploit hierarchical plan representations. Our results show that our approach not only can dramatically reduce the costs of finding synergies compared to non-hierarchical strategies, but can also find synergies that might otherwise be missed.