Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
04/10/2006
1
A System Theoretical Approach to Situation Awareness and its application
A Holistic View of Purposeful Elements
Charlie LagervikPer Gustavsson
Cambridge, September 27, 2006
Student
04/10/2006
2
Objectives
• Address Situational Awareness
• Concept of Situational Awareness
• Machine human context
• Example of our usage
• Our continued work
• A glimpse of ideas in building Situation Awareness
04/10/2006
3
Outline
• Why want a theoretical framework for Situational Awareness?
• Theory:• Ackoff - Adaptive Management• Bedny & Meister - Orientational Activity
• Lagervik conceptual design for the common part of SA, • Architecture that uses the idea:• An Implementation
• Costal Control Application
• Summarization• Future Work• Questions
04/10/2006
4
Transformation in defense operations Not in paper
04/10/2006
5
Requirements
• Net-Centric
• Adaptive
• Multi User
• Multi Role
• Information Fusion
• Knowledge
• Interoperability
• Decision Support
• Temporal
• Networked
Not in paper
04/10/2006
6
I-Terminal
I-Terminal
I-Terminal
I-Terminal
I-Terminal
I-Terminal
EU
OwnOwn
CivilAuthority
CivilAuthority
Service
Own
I-Terminal
I-Terminal
Net-Centric Technical Networks enables Human Networks
Not in paper
04/10/2006
7
DecideDecide
OrientOrient ActAct
ObserveObserve
DecideDecide
OrientOrient ActAct
ObserveObserve
DecideDecide
OrientOrient
ObserveObserve
ActAct
DecideDecide
OrientOrient
ObserveObserve
ActAct
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient Act
Observe
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient Act
Observe
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient Act
Observe
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
Decide
Orient
Observe
Act
0
1
2
3
4
0
1
2
3
4
0
1
2
3
4Data
KnowledgeInformation
Data
KnowledgeInformation
Data
KnowledgeInformation
Joint / Coalition / Interservice
Asymmetric – Constantly change only stable state
Not in paper
04/10/2006
8
Background
• Our interest comes from • Want to supply commander with Intention Awareness
• Need a system that can execute multi-hypothesis of COA from previous, present and predicted states/situations
• Need models that are adaptive, goal oriented
• Models need internal representation of state, mission, vision, goal
• Models need to be able to exchange information
• Models need to have some internal representation of the environment
• Need an architecture
• Realized that such representation of situation also is needed for operational systems.
• Thought that if we have such representation present in the M&S domain the same representation can be used in the operational domain.
Not in paper
04/10/2006
9
Definitions
Decision-Making with Act• A decision is made to make something occur that should not else
occur or to make something not to occur that should else occur• Expected result in a certain time intervall• Based on assumptions, information, feelings• Conscious vs. unconscious
Situational Awareness• The conscious and unconscious dynamic processes to achieve a
reflection of the systems purpose and form, function, system state and future system states
Intention Awareness• The sub part of Situation Awareness that express the path between
current state and goal state for own and opponents’.
04/10/2006
10
The change of change
• Constant change makes it harder to make decisions• Uncertainty in what to look for• Past information that were valid then is not guaranteed for decision-making
now
Myth: The more information the decision maker has, the better decisionis made .
Information Overload• Filtering • Limited time -> Filtering delegated to other people • Limited time -> computer filtration• But computer has no understanding for the situation of the data or the
situation of the decision-making
04/10/2006
11
JDL-Model (altered)
Signal
Object
Situation
Intention
Control
Human
information fusion, data mining and data warehousing
Database
Queries
Human
CategorizationIdentificationGroupingAggregationSummarizationCorrelation…
04/10/2006
12
Understanding / Human
Patterns / Computer Based
Signal
t+0, position ± offsett+1, position ± offset
…t+n, position ± offset
t+n, Position ± offset
Object
SituationObjA
t+0, pos ± oft+1, pos ± of
…t+n, pos ± of
ObjBt+0, pos ± oft+1, pos ± of
…t+n, pos ± of
ObjCt+0, pos ± oft+1, pos ± of
…t+n, pos ± of
IntentionObjA
t+0, pos ± oft+1, pos ± of
…t+n, pos ± of
ObjBt+0, pos ± oft+1, pos ± of
…t+n, pos ± of
ObjCt+0, pos ± oft+1, pos ± of
…t+n, pos ± of
Human
04/10/2006
13
ManualSemiAuto
Semi
Increased information uncertainty
Increased scenario complexity
Manual
Manual
Orient
Observe
Situation Awareness
Semi
04/10/2006
14
ManualAuto
Increased information uncertainty
Manual
Orient
Observe
Situation Awareness
Semi
Increased scenario complexity
ManualSemiAuto
Manual
Semi
04/10/2006
15
Automated
• Computers will be in the position of deciding what to look for. • Need understanding of the nature of the decision and what will effect the
decision.
• Computer must be able to reason about the situation.
• Reasoning systems: Goal seeking systems. • The properties of the elements are monitored and are used to understand
the goal of the element. The understanding about how the different properties of the environment and other elements in the system affect the goal of the element is the main concern
04/10/2006
16
Social Purposeful System
• Humans are non-deterministic systems
• A system that at a first glace seems to be totally mechanical and deterministic often involves humans as decision makers, for a part of the system or for the whole system.
• So to understand the system the goal of the human has to be determined and evaluated in the process.
• A social system should be described as a social system with models indented for this description.
04/10/2006
17
How to describe Situation Awareness
• System Theory• Adaptive-Learning (Ackoff )
• intended for describing the internal process for an organization to reach SA
• Orientational-Activity (Bedny & Meister)• intended for describing the internal process for a person to reach SA
• Even though the difference in fields there are common aspect according to SA that can be compared.
• The comparison is focusing on these parts that are common for both of the theories (the intersection of the theories).
04/10/2006
18
(D)Decision-making
subsystem
(F)Diagnosis and
prescription
(E)Memory and comparator
(G)Symptom and
presymptom analyzer
(C)Information subsystem
(B) Environment
(A)Organization
(1)Data
(4)Inquire
(2) Information
(3) Requests
(5)Instructions
(6) Decision record
(7) Instructions on information required
(8)Assumption -
relevantinformation
andactual
performance
(9) Deviants
(10a) Change
(10d) Change
(10c) Change
(16) Change
(11) Performanceindicators
(12) Symptoms andpresymptoms
(10b) Change
(13) Threats andopportunities
(14) Diagnostic andprescriptive record
(15)Instructions on
informationrequired
(17) Threats andopportunities
Theory:Adaptive ManagementOrientational ActivityConceptual SA design
Architecture:ImplementationSummarization
Adaptive-learning management system, after Ackoff
04/10/2006
19
Theory:Adaptive ManagementOrientational ActivityConceptual SA design
Architecture:ImplementationSummarization
Past experience Criteria ofevaluation
Making a decsion andperformance
Orienting a explorativeaction
Meaning of inputinformation Evaluative and inducing components of
motivation
Sense(significance) Motivation
Subjectively relevant tasksconditions
Image-goal
Conceptual model
Operativeimage
Situationawareness
1
2
3
4
5
678
functional model of orientational activity, after Bedny & Meister
04/10/2006
20
Orientationalactivity
(Bedny & Meister)
Management information
(Ackoff)
Conceptualsystem
Meaning of input information Information subsystem Data gateway
Evaluative and inducingcomponents of
motivationMemory and comparator Evaluative system
Conceptual model, pastexperience Memory and comparator Memory
Image-goal (Decision-makingsubsystem) Goal
Making a decision and performance
Decision-makingsubsystem Perform
Subjectively relevant tasks conditions
Symptom and presymptomanalyzer, Diagnosis and
prescriptionDiagnostic
Theory:Adaptive ManagementOrientational ActivityConceptual SA design
04/10/2006
21
Systemof
Interest
Perform
GoalDataGateway
EvaluateSystem
Diagnostic
Environment
Memory
Data
Data
Info req.
04/10/2006
22
Perform
GoalDataGateway
EvaluateSystem
Diagnostic
Memory
Data
Info req.Perform
Goal DataGateway
EvaluateSystem
Diagnostic
EnvironmentMemory
Data
Info req.
Connection between two systems
• Interchange of data, information, awareness, knowledge, wisdom
Orient
Decide
Act
Observe
Systemin
Focus
Systemin
Focus
Other
Orient
Decide
Act
Observe
04/10/2006
23
Information SharingInteroperability
Network / Connectivity
Simulation /Implementation
Modeling / Abstraction
Data
Information
Knowledge
Awareness
Wisdom
Level 0No Interoperability
Level 1Technical Interoperability
Level 2Syntactic Interoperability
Level 3Semantic Interoperability
Level 4Pragmatic Interoperability
Level 5Dynamic Interoperability
Level 6Conceptual Interoperability
04/10/2006
24
Multi-Hypothesis Intention Simulation Agent Architecture
Database
*ML
Doctri
ne Protocol
Representation
*MLDoctri
ne Protocol
Representation
n
fe
n
f e
n
f
e
n
f e
e nf
f
f
f f
ff
Queries
ECA-Rules Object Hypothesis
Situation Hypothesis
Signal Hypothesis
Intention HypothesisK-10d
K-10hK
K+10hK+10d
Fusion that are derived from doctrinal patternsUser/Agent state that this doctrinal pattern existsUser/Agent “discover” new patterns
New patternsSimulation
Simulation
refinementrefinement
Hypothesis
Sensor
04/10/2006
25
Onotlogy (other representation of SA)
Agent
hasInput
Information ControlCommandcommands
ObjectPropertyDatatypePropertySubClass/Property
Simulated NewFusion
controls
agentName
hasOutput
Data
name
value
hasLiaison
04/10/2006
26
Agent
Information
Control
Command
Simulated NewFusion
Data
04/10/2006
27
Costal Control
Radar Directionfinder TKM BD
OpenSIS
AIS UW C2 Sim
04/10/2006
28
Future Interest
• Memory: shared memory between SA component and decision-maker. • Representation of information, LCIM , images, C-BML
• Unconscious process, the N component finding new relations, goals, similarities etc
• Requiring new Data/Information. Dynamic and Conceptual Interoperability
• Curiosity. The computer should not stop to execute just because there are no instructions to execute.
• Motivation and sense. Motivation is regulating the energy of achieving the goal, and sense is responsible for determine the priority (significance) of the goal for the system.
• Sensor feedback. the importance and how the data is used and how the sensors should be design to support the given feedback of the gathered data.• also need sensor system with ability to make use of such information
04/10/2006
29
In Focus
• Representation• Conceptual representation of the belief process (OWL-S, UML)
• Semantic, Pragmatic representation C-BML• Web services, grammar, ontology for orders and reports
• Syntactic, Semantic representation JC3IEDM, EXDL
• Implement more fusion algorithms
• Experimentation
• Qualitative VV&A
04/10/2006
30
Summarization
• The Lagervik System Theoretical Approach to Situation Awareness helps us to think
• Developing• Information fusion framework (MuHISA)• Costal application
• Future• Representation• More fusion algorithms• Qualitative VV&A
04/10/2006
31
Questions?Charlie.Lagervik & Per.M.Gustavsson
Enviroment
DataGateway
Evaulation
Memory
Goal
Perform Diagnostic
Systemof
interest
Data
Data
IntepretationReflection
Deviants
Critera of evalation
Pastexperience
Information
Definitions
T &O
Input ongoal
Motivation
Goal
Prefomance indicators/data (external)
InquireInquire
Diagnostic and prescriptive recordDerivants/data(internal)
Instruction on information required
Request of information
InformationRequest information
Information
Criteria of evaluation
Xxxx Data
Function block Association Change Instruction System
Xxxx
04/10/2006
32
www.saabgroup.com
04/10/2006
33
Ontology
Service
Capability
hasCapabilityCapabilityLevel
PersonnelInformation ControlLiaisonEconomyTrainingInfrastructureCommandLogisticImpact Delivery
capabilityName
capabilityLevelName
capabilityLevelDescription
hasCapabilityLevel
hasRecursiveCapability
ObjectPropertyDatatypePropertySubClass/Property
Land Sea Air Info