The Cooperative Real-time Control Architecture (CIRCA) was designed for mission-critical applications autonomous planning and control. As space becomes mature and budgets shrink, NASA is towards this type of autonomy to support future missions including the New Millennium Space One probe and the Mars Rover projects. We are actively investigating the application of CIRCA to a variety of domains including spacecraft planning and control and autonomous aircraft control. In this paper, we discuss one particularly challenging type of planning problem that arises in mission-critical applications, drawing on an example from the Cassini Saturn mission. Prepositioning problems arise when certain actions must be taken to preposition assets or otherwise pre. pare for contingencies, before those contingencies could possibly occur. In the Cassini example, a backup inertial reference unit (IRU) must be preheated long before an engine burn is planned, so that if the primary IRU fails during the burn, the backup will be immediately available. The IRU preheating operation is a "prepositioning action." To build plans that include this type of prepositioning, a planner must "look ahead," recognize the contingency, and identify appropriate prepositioning actions. CIRCA's new Dynamic Abstraction Planner (DAP) efficiently builds plans that include prepositioning. This paper is not meant to be an introduction to CIRCA; instead, our goal is to describe how CIRCA can address the Cassini prepositioning problem, and discuss the various strengths and weaknesses of the approach. Accordingly, we refer readers to other publications (Musliner, Durfee, & Shin 1993; 1995; Goldman et al. 1997) for CIRCA overviews and dept;a.mamtg algorithms. This paper begins elements of prep<OSltlo,nll1l! P·roiblems and then the details of the »n.u!JlLUl<~u Cassini problem. We then present the CIRCA solution to this problem. We review how other attempt to solve this type of problem, comparing their features with CIRCA, and conclude with a discussion of the more general directions for future work on CIRCA.
[1]
Oren Etzioni,et al.
An Approach to Planning with Incomplete Information
,
1992,
KR.
[2]
David J. Musliner,et al.
CIRCA: a cooperative intelligent real-time control architecture
,
1993,
IEEE Trans. Syst. Man Cybern..
[3]
Robert P. Goldman,et al.
Dynamic Abstraction Planning
,
1997,
AAAI/IAAI.
[4]
Erann Gat.
News from the Trenches: An Overview of Unmanned Spacecraft for AI Researchers
,
1996
.
[5]
Edmund H. Durfee,et al.
Plan Development using Local Probabilistic Models
,
1996,
UAI.
[6]
Michel Barbeau,et al.
Planning Control Rules for Reactive Agents
,
1997,
Artif. Intell..
[7]
David H. D. Warren,et al.
Generating Conditional Plans and Programs
,
1976,
AISB.
[8]
Nicholas Kushmerick,et al.
An Algorithm for Probabilistic Planning
,
1995,
Artif. Intell..
[9]
Craig Boutilier,et al.
Abstraction and Approximate Decision-Theoretic Planning
,
1997,
Artif. Intell..
[10]
Robert P. Goldman,et al.
Conditional Linear Planning
,
1994,
AIPS.
[11]
John L. Bresina,et al.
Anytime Synthetic Projection: Maximizing the Probability of Goal Satisfaction
,
1990,
AAAI.
[12]
Mark A. Peot,et al.
Conditional nonlinear planning
,
1992
.
[13]
Daniel S. Weld,et al.
Probabilistic Planning with Information Gathering and Contingent Execution
,
1994,
AIPS.
[14]
David J. Musliner,et al.
World Modeling for the Dynamic Construction of Real-Time Control Plans
,
1995,
Artif. Intell..
[15]
Gregg Collins,et al.
Planning for Contingencies: A Decision-based Approach
,
1996,
J. Artif. Intell. Res..