Expert Systems in Defense

1
Expert Systems in Defense

• Ben Allegretti - Juan Carlos Nogueira

2
Agenda

• Introduction
• Background
• Expert Systems in Defense
• Examples
• Conclusion

3
Background
4
Artificial Intelligence

• Definitions
• AI is behavior by a machine that, if performed
  by a human being, would called intelligent.
  (Turban Aronson)
• Is the study of how to make computers do things
  that at the moment, people are better. (Rich
  Knight)
• AI is a theory of how the human mind works.
  (Fox)
• Navy Center for Applied Research in Artificial
  Intelligence

5
AI vs NI

• Permanent
• Duplicable disseminable
• Less expensive
• Consistent
• Documentable
• Faster in certain tasks
• Better for certain tasks

• Creative
• Use sensory experience
• Use of wide context of experience

6
Expert Systems

• Definition
• Computarized advisory programs that attempt to
  imitate the reasoning process and knowledge of
  experts in solving specific problems.
• Lets consult an expert system

MIT AI Lab, Start System
7
Intelligent DSS

• Active (Symbiotic) DSS
• Capable of take the initiative
• Understand the domain
• Help to formulate the problem
• Relate the problem to a solver
• Interpret the results
• Explain the results
• Self-evolving DSS
• Capable to modify their behavior according to the
  user

8
History of ES

• 1950 Turing. Can a machine think?
• 1950-60 General-purpose Problem Solver (GPS),
  Newell and Simon
• 1965-70 DENDRAL Feigenbaum, Stanford U.
• 1970-75
• Reddy Carnegie-Mellon U.
• MYCIN
• 1980-90 Enterprise disasters
• 1990s Mature ES
• (Trilogy, Red Pepper)
• Feigenbaums students (Christy Jones, Zweben,)
• 1997 IBMs Deep Blue defeats Kasparov

9
Knowledge

• Knowledge encompasses the implicit and explicit
  restrictions placed upon objects, operations and
  relationships along with general and specific
  heuristics and inference procedures involved in
  the situation being modeled. (Sowa)

Abstraction
Knowledge
Information
Data
Quantity
10
Definitions

• Expertise
• Extensive task-specific knowledge acquired from
  training and experience.
• Theories
• Rules and procedures
• Global strategies
• Meta-knowledge
• Facts

11
Definitions

• Expert A human capable of
• recognize, formulate, solve, explain and
  determine the rules of a problem,
• learn from experience,
• break the rules,
• determine relevance and
• be aware of limitations.

12
Definitions

• Knowledge Base
• Repository for the knowledge.
• Inference Engine
• Software that provides the methodology for
  reasoning, composed by
• Rule Interpreter
• Scheduler
• Consistency Enforcer

13
(No Transcript)
14
Benefits of ES

• Increase productivity
• Decrease DM time
• Increase process product quality
• Capture of scarce expertise
• Integrate several experts opinions
• Ability to work with uncertain info
• Flexibility
• Improve decision quality

15
Limitations of ES

• Scarcity of knowledge
• Difficulty in extracting knowledge
• Narrow domain of expertise (no GPS)
• Experts limitations
• Engineers limitations
• Users limitations
• Let interact with another ES

US Department of Labor Occupational Safety and
Health Admin.
16
Expert Systems in Defense
17
DoD goal

• By 2010, to know with certainty where enemy and
  friendly forces are within a given battlespace.
• The term for this is Dominant Battlespace
  Awareness (DBA).
• Not only location, but near-perfect, real time
  discrimination between targets and non-targets on
  the battlefield of the future.

18
The Awareness/Knowledge Problem

• Vast amounts of digital data will need to be
  processed, correlated, stored, and displayed.
• The data base of a particular battlespace will
  have to be continuously updated with real-time
  information to make it useful to a warfighter.

19
Some difficulties with achieving
awareness/knowledge are

• What common data standards will be utilized?
• Need of power-projection world-wide with little
  advance notice.
• How will targets/nontargets be detected and
  tracked?
• Where will the data be stored?
• How will the data be presented?

20
The Requirement

• Preparing for the future is the third leg of the
  United States National Military Strategy.
• Joint Vision 2010 is the concept for how to
  dominate the battlefield of the future.
• Enhanced command and control
• Improved intelligence
• Technology
• Transform the traditional functions of maneuver
  and strike into dominant maneuver and precision
  engagement.

21
Keys of success

• An integrated "system of systems," linking
  intelligence collection and assessment, command
  and control, weapons systems and support elements
  to achieve battlespace awareness.
• A standard command, control, comm., computers,
  intelligence, surveillance and recon. (C4ISR)
  architecture.
• Total integration of the vast amounts of data.
• New technology wide area surveillance sensors,
  automated decision making tools, and comm. links.

22
Agencies

• Defense Advanced Research Projects (DARPA)
• Three areas sensors and communications,
  exploitation, and information integration.
• Defense Information Systems Agency (DISA)
• Joint Technical Architecture(JTA) for all new C4I
  systems.
• Program Offices
• Acquisition process and as system components.
• Service Battle Labs
• Experimentation an exercises.

23
Examples of Military ES
24
Advanced Concept Technology Demonstrations (ACTD)

• Opportunities for the Integration of Expert
  Systems

25
ACTD

• Rapid Terrain Visualization
• weather, terrain, and electromagnetic picture
  concerning a particular battlespace.

26
ACTD

• Tier II-plus High Altitude Endurance Unmaned
  Aerial Vehicles (HAE UAV)
• broad-area, all weather,
• day-night identification of
• both fixed and mobile
• targets on land, air
• and sea.

27
ACTD

• Semi-automated Imagery Processing (SAIP)
• Exploitation of national and theater imagery
  intelligence using AI.

28
ACTD

• Joint Combat Identification (CID) ACTD
• Real-time, accurate knowledge of friendly,
  neutral and enemy forces.

29
ACTD

• Integrated Collection Management (ICM)
• Integration of national,
• theater, and tactical
• sensors as a
• surveillance system.

30
ACTD

• Dynamic Database Program (DDP)
• Continuously updated, integrated, multi-echelon
  picture of a dynamic battlespace managed with AI.

31
ACTD

• Battlefield Awareness and Data Dissemination
  Dissemination of large databases utilizing the
  Joint Broadcast Service (JBS) at all echelons.

32
ACTD

• Global Command and Control System (GCCS)
• Client served network
• C2
• Logistics
• Intelligence legacy systems.

33
Developmental and Operational Expert Systems
34
Advanced aircraft Prognostics and Health
Management (PHM)

• Joint Strike Fighter
• (JSF)

35
JSF

• Uses advanced sensors, on-board the aircraft,
  integrated through algorithms and intelligent
  models such as neural nets to monitor, predict,
  and manage aircraft health.
• Goal of PHM is to enable what the JSF program
  calls Autonomic Logistics

36
Autonomic Logistics

• A maintenance and supply system.
• Information on aircraft faults are detected while
  the aircraft is airborne.
• Automatically down-linked to trigger the
  logistics system.

37
Autonomic Logistics

• Meet the returning aircraft with appropriate
  parts, maintenance personnel, and maintenance
  equipment.

38
Autonomic Logistics

• Software has built-in learning capabilities
  contained within
• Individual aircraft PHM systems
• Fleet-wide logistics system
• Result is condition based maintenance
• System will eventually, very accurately predict
  impending failures and provide for replacing
  parts just before they might fail

39
PHM and the Related Autonomic Logistics System

• Payoff
• Reduce maintenance manpower requirements by
  approximately 20-40 .
• Increase combat sortie generation rate by 25 .
• Reduce logistics footprint (numbers of C-141
  cargo aircraft loads) by 50
• All relative to current strike aircraft.

40
JSF

• PHM and the related Autonomic Logistics System is
  being incorporated into both competing designs of
  the JSF.
• The PHM and ALS will become operational with the
  JSF IOC.

41
Special Operations Forces (SOF)Mission
Effectiveness Model (MEM)A Fuzzy Logic
Decision Support System
42
SOF MEM

• Developed to support analysis of theoretical SOF
  employment.
• Could be employed in support of real-world
  tactical decision-making.

43
SOF MEM

• SEAL platoon was subject for system.
• Focus of this project was the launch phase of a
  submarine-based SEAL mission.
• The application determines the impact of 14
  environmental conditions on expected mission
  effectiveness.

44
SOF MEM 14 environmental conditions

• Sea state (wave height),
• Water temperature,
• Current set (direction),
• Current drift (speed),
• Water visibility (feet),
• Air temperature,
• Humidity,

• Wind direction,
• Wind speed,
• Air visibility,
• Illumination,
• Load towing/carrying, and
• Exposure (warm, dry, cold, or wet),
• Distance (Ship to Shore).

45
SOF MEM

• CubiCalc by HyperLogic Corporation (commercially
  available fuzzy logic software toolkit) used in
  model development.
• CubiCalc is a fully integrated fuzzy logic
  development system.
• Runs on a PC under Microsoft Windows.

46
SOF MEM

• Natural language was used to form the expert rule
  base of the model.
• Rules were developed to relate launch method,
  mobility platform, element size, and tactical and
  environmental conditions to mission
  effectiveness.

47
SOF MEM

• Rules were created using an IF-THEN-AND-OR-ELSE
  format.
• Example
• Air Temperature
• IF air temperature is Cold, THEN mission
  effectiveness is Medium
• IF air temperature is Cool, THEN mission
  effectiveness is High
• IF air temperature is Nice, THEN mission
  effectiveness is Very High

48
SOF MEM

• 111 rules were required to support all of the
  combinations of launch method, mobility platform,
  element size, and tactical and environmental
  conditions for the launch phase.
• Rule sets can be saved for future use.

49

• SOF MEM Output

50
SOF MEM

• SEAL launch example limited in scope but,
• Model demonstrates the broad applicability to
  other tactical decision-making problems.
• Reuse of Rule Sets is a key factor.

51
Joint Computer Aided Logistics System

• (JCALS)

52
JCALS

• An infrastructure capable of integrating
  digitized technical data.
• Support a weapons systems acquisition and
  logistics life cycle.
• Provides an automated information system
  architecture independent of application.
• Supports the Services/DLA's goal of automating
  technical manual processes and functions.

53
JCALS

• Software architecture consists of seven
  functional components
• Application Process Functions component
• Expert System used within the analysis tools and
  supportability assessment tool
• JCALS on line and functioning at 273 sites and
  continuing to be expanded.

54
Operator-Agent

• Australian Department of Defence

55
Operator-Agent

• Tool used to model and analyze weapons systems.
• Models the reasoning processes preformed by
  humans
• Autonomous behavior,
• Reactive/proactive behavior,
• Ability to interact.

56
Operator-Agent

• Take the form of synthetic forces operating in a
  virtual world.
• Examples of simulated scenarios (conducted
  against a virtual opposing force)
• Aviation strike mission,
• Maritime engagement,
• Mechanized attack.

57
Operator-Agent

• Using the agents
• Inputs
• Environmental data,
• Order of battle data,
• Mission orders.
• Outputs
• Results of engagements/missions
• Data on Operator performance

58
Operator-Agent

• Benefits
• Allows testing of weapons systems, operation
  plans, and concepts.
• Fully operational models in existence.
• Agent plans are easily created and read.

59
Hierarchical Interactive Case-Based Architecture
for Planning (HICAP)

• Noncombatant Evacuation Operations (NEOs)

60
HICAP

• Commanders required characteristics for decision
  support tools
• Doctrine-driven,
• Interactive,
• Providing case access (historical data).
• HICAP provides these capabilities for NEOs.

61
HICAP

• System architecture
• Integrates a task decomposition editor with a
  case-based planner.
• Allows editing of doctrinal tasks to be added to
  the OPPLAN.
• Provides interactive refinement of the plan based
  on past NEO OPPLANs and lessons learned.

62
HICAP

• The software ensures that OPPLANs are framed
  within doctrine and the commanders planning
  guidance.
• Doctrine is used as a guide for planning.
• The editor interactively allows users to generate
  the plan.
• Knowledge of previous NEOs is represented as
  cases
• Interactive cases used to augment or replace
  doctrinal tasks and SOPs

63
HICAP

• Case Knowledge is framed in the form of question
  and answer pairs for problem specific tasks.
• Cases obtained from unit SOPs and after action
  reports detailing previous problem solving
  episodes.

64
HICAP

• ES compares decomposed elements of a task to
  those in its database.
• If all preconditions are met, that case will be
  used as a task list.
• Otherwise, the most similar case is used as a
  reference.

65
HICAP

• Other system uses
• Allows use of stored plan fragments,
• Editing the assignment of personnel and units to
  tasks,
• Investigate the status of tasks to be completed
  (assisting in managing the operation).

66
Conclusion
67
Conclusion

• ES have a wide range of DoD applications
• Operational,
• Logistical,
• Administrative.
• Failure to apply ES could result in an
  inefficient processes that could not be sustained
  in the resource constrained DoD of the future.

68
Questions?
69
(No Transcript)