Interactive Resource Management in the Comirem Planner - Robotics Institute Carnegie Mellon University

Interactive Resource Management in the Comirem Planner

S. F. Smith, D. W. Hildum, and D. A. Crimm
Workshop Paper, IJCAI '03 Workshop on Mixed-Initiative Intelligent Systems, August, 2003

Abstract

In this paper, we describe Comirem, a lightweight, interactive tool for resource management in continuous planning domains. Comirem is designed for domains where complex, heterogeneous sets of re-sources are required to execute planned activities and usage must be synchronized to satisfy complex temporal and spatial constraints. Comirem promotes an opportunistic planning paradigm, where resource allocation decisions are made and revised incrementally as plans and availability constraints become known and refined during the planning process, and as execution deviates from planned behavior. To this end, resources are assigned (and re-assigned) via a least-commitment, constraint-posting scheduling procedure. The Comirem system design follows three basic mixed-initiative principles: (1) that users will want to make planning and resource allocation decisions at different levels of detail in different circumstances and correspondingly delegate more or less decision-making responsibility to system processes in different contexts, (2) that abstract domain models, coupled with graphical visualization can provide an effective basis for communicating decision impact and proposing decision options, and (3) that incremental, adaptive problem-solving capabilities, which attempt to localize change whenever possible and appropriate, are central to maintaining continuity in the planning and resource management process. These principles are used to provide a variety of tools for mixed-initiative resource allocation, feasibility checking, resource tracking and conflict resolution.

BibTeX

@workshop{Smith-2003-120529,
author = {S. F. Smith and D. W. Hildum and D. A. Crimm},
title = {Interactive Resource Management in the Comirem Planner},
booktitle = {Proceedings of IJCAI '03 Workshop on Mixed-Initiative Intelligent Systems},
year = {2003},
month = {August},
}