Multi-Agent Management of Joint Schedules

Distributed Plan and Schedule Management

Volume

Papers from the 2006 AAAI Spring Symposium

Contents

In this paper, we describe an incremental scheduling framework designed to support joint management of inter-dependent schedules by multiple executing agents. We assume an uncertain execution environment and a distributed representation of the overall problem and schedule such that no single agent has a complete view. Hence as unexpected execution events force changes to individual agent schedules, agents must recognize inter-dependencies and coordinate to revise schedules in a way that continues to maximize the quality of their joint activities. Our approach combines an underlying flexible-times representation of the schedule, which exploits temporal flexibility to absorb executional uncertainty and insulate dependencies across agent schedules, with an incremental approach to schedule revision, which promotes solution stability and tends to minimize the ripple effect of change across agent schedules. We summarize basic mechanisms for constructing and revising schedules, for integrating scheduling with execution and for generating options to focus coordination with other agents. Our framework is being developed and applied within the DARPA Coordinators program.