Multi-agent planning with quantitative capability

Abstract

In the recent years, there has been a lot of research in multi-agent planning (MAP). Multi-agent planning applications include search and exploration, robotics, and logistics etc. Existing works show that agents have different capabilities if the agents have different action sets. However, agents may have different capabilities even if the action sets are same. In this paper, we present an approach for multi-agent planning problem, where capabilities of the agents are represented in a quantitative manner. The proposed approach translates such MAP problem into classical planning problem. The resultant classical planning problem is solved by an existing state-of-the-art classical planner. Experiments were performed on some benchmark planning domains and the results are quite promising.