Application of O-Plan to Logistics Planning
Introduction
The aim of this article is to provide a description of the application
of the O-Plan system to the task of logistics planning. Distribution
logistics for commercial organisations and retailing chains is an area
where systems such as O-Plan are ideally suited. AIAI has
previously worked with ICL on distributions logistic and AIAI is currently working with a petroleum products company in
Scotland in this area (an article on the Schedule-it system is
included in this issue of Airing).
The research of the O-Plan project is motivated by the transportation
planning domain that is the focus of the DARPA/Rome Laboratory Planning
Initiative (ARPI) [4]. This domain involves the
movement of materials and forces with a mixture of aircraft and ships.
The task is to have the materials in place by a designated starting
date usually referred to as D-DAY. This is simple to state but often
difficult to achieve in practice. The main problem is that materials
move through a number of ports and airfields which have finite
capacities in terms of warehouse and parking space. In addition a
number of support personnel are required to monitor and operate these
facilities. The forces and materials to be moved are identified and a
fixed number of transport assets are provided by the US Transportation
Command. The number and make up of the forces can vary and as such a
number of alternative COAs can be generated. These COAs
are plans that specify at a high level the sequences of actions for
movement and employment of forces. The commander of the operation is
presented with several alternative COAs and an evaluation of the
tradeoffs among them. These options are explored and different
aspects/variables altered to identify potential new COAs. A
decision is finally made on the scale of the mission and the chosen
COA needed to support it. This COA is refined to a more
detailed level with improved plan feasibility estimators.
The function of the O-Plan system is to generate a number of
alternative COAs for a given crisis and to provide them in a
form in which they can be evaluated by the user and other support
tools. The reason for considering using systems such as O-Plan is that
current military planning faces a number of major problems:
- lack of support for considering the impact on a plan of
execution time changes,
- lack of support for explaining aspects of the plan to the
execution agent,
- inability to consider all scheduling aspects while COAs
are being analysed
The problem of logistics planning using the O-Plan system has been
investigated in two main areas:
PRECiS/Pacifica Transportation Planning Domain
The PRECiS domain was developed by AIAI for use on the ARPI
as a non-confidential domain and is based around the fictitious but
realistics island of Pacifica (Full details are described in
[2][3]. The focus of the PRECiS domain is
Non-Combatant Evacuation Operations (NEOs) which involve the
evacuation of a number of foreign nationals from the island. A number
of military cargo planes and fuel tankers are available in Honolulu
and can be used to transport helicopters, trucks and fuel to the
island. The helicopters and trucks are used transport evacuees from a
number of outlying cities to the island's capital Delta for evacuation
via a passenger plane to Honolulu. The military cargo planes are then
used to transport the trucks and helicopters back to Honolulu. A
number of different scenarios were explored involving different
numbers of evacuees, transport assets, time constraints and fuel
levels.
This domain was used to test a number of different employment and
deployment plans involving a number of naval, air and sea forces from
the continental US to an country in the Middle East. The plans
generated moved the materials of these forces by strategic and
tactical air and sealift and by ground transports. The plans
generated needed to identify a number of major points:
Military Employment Planning
- Which seaports and airports were to be used in the destination
country and which could be used as staging points along the way.
- Which air and sea assets were to be used and to identify a
number of keys pieces of information: the number of sorties flown,
number of berths and runways needed, the amount of parking and storage
needed. etc. This information was extracted from the plans generated
by O-Plan through the EXPECT system being developed by USC/ISI
[1] as part of the ARPI.
- The lines of communication needed to support the sea and airlift
operations.
- Which was the most appropriate method of transportation, i.e.
send it all the forces material by sea or by air, or by a mixture of
sea and air. In the latter case the plan needed to identify a staging
point where the materials would be re-assembled.
A typical crisis action OPLAN for this problem contained a few
hundred actions and the related Time Phased Tactical Deployment Data
book (generated from the crisis plans) contained a few thousand
entries which describe which units (or parts of larger units) are
moved from where to where, when and by which method of transportation.
References
- 1
- Gil, Y. Knowledge Refinement in a Reflective
Architecture Proceedings of the Twelfth National Conference on
Artificial Intelligence, Seattle, WA, USA. August 1994. Published by
AAAI Press/ The MIT Press Menlo Park, CA, USA.
- 2
- Gil, Y., Hoffman, M., and Tate, A.
Domain-Specific Criteria to Direct and Evaluate Planning
Systems, presented at the 1994 DARPA/Rome Laboratory Planning
Initiative Workshop, Tucson, Arizona, USA. Published by Morgan
Kaufman, San Francisco, CA, USA. Also available as Information
Sciences Institute Technical Report ISI-RR-93-365, University of
Southern California.
- 3
- Reece, G., Tate, A., Brown, D., and Hoffman,
M. The PRECiS Environment, papers of the DARPA/Rome Laboratory
Planning Initiative Workshop at the National Conference on Artificial
Intelligence (AAAI-93), Washington D.C., USA. ARPI Report
DARPA-RL/CPE/Version 1, August 1993. Also available as Artificial
Intelligence and Applications Institute Technical Report AIAI-TR-140,
University of Edinburgh.
- 4
- Stillman, J. and Bonissone, P., Developing New
Technology for the DARPA-Rome Laboratory Planning Initiative, IEEE Expert Intelligent Systems and their Applications, pp10-17,
February, 1995.
Brian Drabble: b.drabble@ed.ac.uk
Planning and Scheduling Group
or
AIAI information
