Multi-agent Planning with Collaborative Actions

Abstract

When multiple agents perform different actions simultaneously such that the combined effect of the actions achieve a given goal, then such actions are referred to as collaborative actions. Temporal planners can handle such actions by reasoning explicitly about time. However, algorithms for partial order planning can be used effectively for multi-agent domains with minor modifications without the requirement of explicit time. In this paper, we consider multi-agent planning (MAP) problems with such actions without reasoning explicitly about time. We give a new specification technique for handling collaborative actions in a PDDL-like language. We propose a new approach to solve such MAP problems. We have implemented and evaluated our approach on some benchmark planning problems. The results show the effectiveness of our approach.