suo-pda-sep99 - AIAI - University of Edinburgh

advertisement
Small Unit Operations SUO/PDA
ARPI
Austin Tate & David Wilkins
AIAI & SRI International
E-mail: a.tate@ed.ac.uk, wilkins@ai.sri.com
www: http://www.aiai.ed.ac.uk/~arpi/SUO/
SUO/PDA
ARPI Fall 1999 Workshop 1
Small Unit Operations SUO/PDA
ARPI
SUO/PDA Objective
Demonstrate capability relevant to a SUO/Situation
Assessment System (SAS) environment to generate,
refine, select, communicate, execute and repair plans
across multiple echelons.
Limited interfaces available on soldier-borne device.
Two examples of COA plan generation and use:
- Defensive - Halt an Attack in Restrictive Terrain;
- Offensive - Military Operations in Urban Terrain.
Advanced Planning Technology
•
•
•
•
•
•
•
•
Use results of DARPA/AFRL Planning Initiative (ARPI)
and Planning & Decision Aids (PDA) work
Multi-Agent Planning Architecture (MPA) and O-Plan
Systems Integration Architecture
SIPE-2/CPEF and O-Plan AI Planners
Mixed Initiative Planning Aids
Planning Domain Knowledge Acquisition Tools/Editors
Continuous Planning, Re-planning and Plan Repair
Planning Process Panels
Rich Shared Plan Representations
Schedule
2QCY99 - SUO Scenario and storyboard/demo script
development. Walk-through of technical requirements,
mockup of non-working parts. Produce demo script and
scenarios.
3QCY99 - Technology and demonstration development
first pass, interim demonstration.
4QCY99 & Q1CY00 - Full demonstration development.
1Q & Q2CY00 - Transition and communication of results
to SUO contractor. Refinement of demonstration and
technology.
SUO/PDA
ARPI Fall 1999 Workshop 2
Objective
ARPI
• Demonstrate capability relevant to a SUO/Situation
Assessment System (SAS) environment to generate,
refine, communicate, execute and repair plans across
multiple echelons.
• Limited interfaces available on soldier-borne device.
• Two examples of COA plan generation and use:
• Defensive - Halt an Attack in Restrictive Terrain;
• Offensive - Military Operations in Urban Terrain.
SUO/PDA
ARPI Fall 1999 Workshop 3
Approach
ARPI
• Use results of DARPA/AFRL Planning Initiative
(ARPI) and Planning & Decision Aids (PDA) work
• Multi-Agent Planning Architecture (MPA) and O-Plan
Systems Integration Architecture
• SIPE-2/CPEF and O-Plan AI Planners
• Mixed Initiative Planning Aids
• Planning Domain Knowledge Acquisition Tools/Editors
• Continuous Planning, Re-planning and Plan Repair
• Planning Process Panels
• Rich Shared Plan Representations
SUO/PDA
ARPI Fall 1999 Workshop 4
Long-term Contributions to the Soldier
ARPI
• Fast generation of multiple distinct COAs, including ones the
commander may not have considered.
• Commander can explore more options in detail.
• Avoid mistakes: uniformly high plan quality, even during
high-stress crises.
• Monitor plan execution and respond quickly to events,
helping the commander modify the plan appropriately.
• Provide relevant information to other echelons, allowing fast
communication while preserving bandwidth.
SUO/PDA
ARPI Fall 1999 Workshop 5
Practical Issues/Challenges
ARPI
• Effort to acquire the knowledge base (KB).
•
•
•
•
•
KB will cover a small subset of an officer’s knowledge and will add value.
Effort commensurate with capability.
Reduce effort by limiting scenario, relying on human knowledge, etc.
Human can override PDA -- blind spots not fatal
Other DARPA programs address this problem in the larger scale
• Assumptions about the world information available from sensors.
•
•
Rely only on information which the SUO/SAS can provide
Baseline: SALUTE reports and GPS data.
• Input and output burden on soldiers when using SUO/PDA.
•
•
PDA would augment officer at time he would consult map or talk on radio
Voice input and limited interface modalities being considered by SUO contractor.
• Sensor planning
•
•
•
PDA will plan for awareness requirements
PDA will respond to reports from SAS
Will not model low-level details of sensor operations, or plan their exact deployment
SUO/PDA
ARPI Fall 1999 Workshop 6
Integration/TIEs
ARPI
• Validate MPA by integrating several systems in DARPA Planning
Initiative (TIE 97-1):
INSPECT (ISI)
OPIS (CMU)
Advisable Planner (SRI)
SIPE-2 (SRI)
ACS (UMass)
Process Panel (AIAI -UEdin)
APAT (ISX)
VISAGE (MAYA)
• Domain is Air Campaign Planning
• thousands of objects, several thousand nodes in each plan
• plan down to support mission level (must allocate supporting
resources)
• air superiority objective only
• targets grouped into networks which depend on other networks
• network effectiveness is modeled quantitatively
SUO/PDA
ARPI Fall 1999 Workshop 7
TIE 97-1 Demonstrations
DARPA
ARPI
• Sept 98 - EFX 98, Ft. Walton Beach FL
• May 98 - ARPI Workshop, Monterey CA
• Feb 98 - DARPA, Arlington VA
• Dec 97 - JFACC PMR, San Pedro CA
• Nov 97 - ARPI Workshop, San Francisco CA
SUO/PDA
ARPI Fall 1999 Workshop 8
Technology Transition
ARPI
DARPA SUO program:
-prototype planning and decision aid
DARPA JFACC program:
-building on Cypress, MPA, TIE 97-1 MAPVis
SUO/PDA
ARPI Fall 1999 Workshop 9
SUO Scenarios
ARPI
• Two SUO-SAS scenarios have been chosen:
• Military Operations in Urban Terrain (MOUT):
Operation San Roberto
• Halt an Attack in Restrictive Terrain:
Operation Golden Manacle
• KA has been performed for both scenarios
• Interim demo is in MOUT scenario
• Main demo is in Halt an Attack scenario
SUO/PDA
ARPI Fall 1999 Workshop 10
Differences between the Two Scenarios
ARPI
MOUT
Halt an Attack









Defensive
Restrictive wooded terrain
Battalion-sized operation
Wide area move planning
Mechanized threat
Wide area sensors
BN, CO, PLT
Workstation-based aids
Graphics+typing modalities









Offensive
Close confined urban terrain
Company-sized operation
Close combat
Foot soldier threat
Localized sensors
CO, PLT, SQ
Soldier-borne aids
Glance+select modalities
SUO/PDA
ARPI Fall 1999 Workshop 11
Opportunities for PDA Support in MOUT
ARPI
Overall Process from Receipt of Mission to Success
In Advance
Deliberative
Planning &
Rehearsal
En-route
Rehearsal &
Replanning
During Mission
After Action
Low Tempo
Adaptation &
Plan Repair
High Tempo
Monitoring &
Plan Selection
After Action
Planning &
Support
SUO/PDA
ARPI Fall 1999 Workshop 12
MOUT Concept of Operations
ARPI
2nd Plt
1st Plt
OP “M” 3rd Plt
4th Plt
SUO/PDA
ARPI Fall 1999 Workshop 13
Halt an Attack - Defensive Scenario
ARPI
• Halt a mechanized enemy
regiment in restrictive
terrain with a SUO battalion
• Place obstacles to force
enemy into engagement area
• Use forward infantry to
observe, channel, and delay
• Rely on fire support to attrit
• Channel enemy to
Southwest
• Use a CO(+) to defeat enemy
in engagement area
SUO/PDA
ARPI Fall 1999 Workshop 14
Defensive Scenario - Plan Generation
ARPI
• BN plan/order is input to PDA
• PDA produces plans for each CO and each PLT
• KB covers following:
• Observing avenues of approach (AAs)
• Using sensors to provide and supplement observation and to provide
security
• Covering AAs with obstacles
• Positioning units
• Selecting a channelizing path for OPFOR
• Nominating positions for fire support (FS) units
• Selecting an engagement area (EA) in which to defeat channelized units
• Preparing the EA for the battle
SUO/PDA
ARPI Fall 1999 Workshop 15
Main Demo - Halt an Attack
ARPI
Battalion
Planning &
Execution
System
Domain Model
Activity
Templates &
Constraints
Reports
Company
Planning &
Execution
System
OPORD
FRAGO
Control Panel
User Interfaces
& C2 Process
Management
Reports
Platoon
Execution
Support
System
SUO/SAS System
OPORD
FRAGO
Initial Battalion
OPORD Creation
and Editing
World Simulation
and Scenario
Event Generation
AIAI Responsibility
SRI Responsibility
WorldSUO/PDA
Simulator
ARPI Fall 1999 Workshop 16
SUO/PDA Knowledge Acquisition Stage
ARPI
Knowledge Acquisition for
Halt an Attack
(SRI with AIAI)
SIPE Acts
Army CALL Bulletins
Doctrine, SOP, TTP
Subject Matter Experts
Process Modelling
Methods
& Tools
Knowledge Acquisition for
MOUT
(AIAI with SRI)
Domain Model
Activity
Templates &
Constraints
O-Plan TF
SUO 3 Repn.
AIAI Responsibility
SRI Responsibility
SUO/PDA
ARPI Fall 1999 Workshop 17
AIAI Contributing Technology
ARPI
Austin Tate
with Jeff Dalton, John Levine, Peter Jarvis
AIAI, University of Edinburgh
80 South Bridge, Edinburgh, UK
E-mail: a.tate@ed.ac.uk
Tel: +44 (131) 650 2732
www:http://www.aiai.ed.ac.uk/project/oplan
http://www.aiai.ed.ac.uk/project/ix
SUO/PDA
ARPI Fall 1999 Workshop 18
AIAI Contributing Technology
ARPI
• Generation of multiple qualitatively distinct alternative
COAs dependent upon alternative assumptions and
advice about the situation.
• Support for mixed-initiative incremental plan
development, manipulation and use.
• Situation-dependant plan repair as situation changes.
• Systems integration framework for modular planning
and plan analysis systems.
• Management of planning and execution process promotion of intelligent process management and
workflow concepts.
SUO/PDA
ARPI Fall 1999 Workshop 19
AIAI Contributing Technology
ARPI
•
•
•
•
•
•
•
•
Shared Models of Tasks, Processes and Plans
Issue-based Problem Solving
Constraint and Domain Management
Planning Process Panels
Web Delivery of Planning Facilities
Process Editor
Previous O-Plan Technology
New I-X Technology
SUO/PDA
ARPI Fall 1999 Workshop 20
ARPI
Shared Plan Model - a rich plan representation using
a common constraint model of activity (<I-N-CA>).
Shared Task Model - Mixed initiative model of
“mutually constraining the space of behaviour”.
Shared Space of Options - explicit option management.
Shared Model of Agent Capabilities - handlers for
issues, functional capabilities and constraint managers.
Shared Understanding of Authority - management
of the authority to plan (to handle issues) and
act which may take into account options,
phases and levels.
SUO/PDA
ARPI Fall 1999 Workshop 21
O-Plan -> I-X
ARPI
Requirements
Requirements
Reports
Reports
Interface
Manager
Controller
Processing
Platform(s)
Issue Handlers
Data Base Manager
PlanWorld
Viewers
Constraint Managers
Plan State
Issues
Nodes
Constraints
Constraint
Associator
Information Sources
Requirements
Requirements
Reports
Reports
Processing Capabilities
Technical &
World Viewers
Task & Option
Management
Model
Management
Mediators/Mapping
Constraint Managers
Information Sources
SUO/PDA
ARPI Fall 1999 Workshop 22
I-Plan and <I-N-CA>
ARPI
Plan State
I
Issues or Implied
Constraints
Plan Agenda
N
Node
Constraints
Plan Entities
CA
Detailed
Constraints
Plan Constraints
Choose (IH)
Do (IH)
Propagate
Constraints
Space of Legitimate Plan Elaborations
C=Critical Constraints
A=Auxiliary Constraints
IH=Issue Handler
(Agent Functional Capability)
SUO/PDA
ARPI Fall 1999 Workshop 23
SRI Contributing Technology
ARPI
David Wilkins, Tom Lee
SRI International
Artificial Intelligence Center
Menlo Park, CA
E-mail: wilkins@ai.sri.com
Tel: 650-859-2057
www: http://www.ai.sri.com/~sipe
http://www.ai.sri.com/~cpef
SUO/PDA
ARPI Fall 1999 Workshop 24
Intelligent Operations Management
ARPI
Domain Characteristics
• Tasks are complex and open-ended
• Operating environments are dynamic and possibly hostile
• Complete and accurate knowledge of the world can never be
attained
• Full automation is neither possible nor desirable
 Successful operation requires a mix of
• user involvement and control
• continuous planning
• rapid response to unexpected events
• dynamic adaptation of activities
SUO/PDA
ARPI Fall 1999 Workshop 25
CPEF Architecture
ARPI
MPA Messages
SUO/PDA
ARPI Fall 1999 Workshop 26
Download