command and control modeling for synthetic battlespaces: flexible group behavior randall w. hill,...
TRANSCRIPT
Command and Control Modeling for Synthetic
Battlespaces:Flexible Group Behavior
Randall W. Hill, Jr.Jonathan Gratch
USC Information Sciences InstituteASTT Interim Progress Review
May 24, 1999
Agenda
Synthetic Forces ProblemProgram HypothesesTechnologies and R&DSignificant Results & Expected ResultsTechnology Transition Products & EffortsProblem AreasProgrammatic Issues
Synthetic Forces Problem
Problem
Need cost-effective C2 modeling Replace / augment human controllers with automated C2 Represent a wide range of organizations and situations
Need realistic C2 behavior C2 models must make believable decisions The outcomes of C2 operations need to be credible
Project Goals
Develop autonomous command forces Act autonomously for days at a time
Reduce load on human operators
Behave in human-like mannerProduce realistic training environment
Perform C3I functionsReduce the number of human operatorsCreate realistic organizational interactions
Program Hypotheses
Hypotheses
Flexible behavior requires the ability to handle situation interrupts
Flexible group behavior requires: Understanding behavior of groups of entities Planning a mission for groups against groups Executing a mission in a coordinated manner
Hypotheses
Flexible group behavior interleaves the processes of situation assessment, planning, execution, and plan repair
Coordinated group behavior requires a theory of multi-agent interaction
Technologies and R&D
Technologies
Continuous Planning Depends on understanding evolving situations Implements planning as a dynamic process Achieve goals despite unplanned events
Collaborative Planning Coordinate group behavior Requires understanding behavior of other groups Reason about organizational constraints
Technologies
Situation Awareness Current situation
Need a consolidated pictureRequires situation assessment at multiple echelons
Future situationIntegrate planning with future sensing requirements Formulate Priority Intelligence Requirements (PIR)
Mission Capabilities
Army Aviation Deep Attack Battalion command agent Company command agents CSS command agent AH64 Apache Rotary Wing Aircraft Suppression of Enemy Air Defense
(SEAD) by indirect fire (partially implemented)
Intelligence assets (partially implemented)
BP
FARPCSS
HA
HA
FLOT
SEAD
SLARMLRS
Battalion Deep Attack
….Operations
Order(plan)
C2 Architecture
Battalion Commander
Company ACommander
Company XCommander
Company A
PilotHelicopter
PilotHelicopter
PilotHelicopter
ModSAF
Company X
PilotHelicopter
PilotHelicopter
PilotHelicopter
….
Operations Order(plan)
Operations Order(plan)
Situation Report(understanding)
Situation Report(understanding)
Situation Report(understanding)
PerceptsActions PerceptsActions
Architecture
Planner Implements continuous planning capabilities
Plan manager Augments collaborative planning with organizational
reasoning and Military Decision Making Process
Time Manager Manages temporal constraints
Domain Theory Maintains plan management and tactical knowledge
Situation Assessment Fuses sensors, reports, and expectations Generates and updates current world view
C2 Entity Architecture
Planner(General Purpose
Reasoner)
Plan Manager Management
Plans
Management Plans
Tactical PlansTactical Plans
ManagementTheory
(domain independent)
ManagementTheory
(domain independent)
Tactical Domain Theory
Tactical Domain Theory
World ModelWorld Model
Situation Assessment
Synthetic Battlespace
Situation Reports, Sensing
Facts, inferencesExpectations
OPORDER Other Communications
Technologies and R&D:Continuous Planning
Continuous Planning
Plan generation Sketch basic structure via decomposition Fill in details with causal-link planning
Plan execution Explicitly initiate and terminate tasks Initiate tasks whose preconditions unify with the current world Terminate tasks whose effects unify with the current world
Plan Repair Recognize situation interrupt Repair plan by adding, retracting tasks
What are Plans?
Hierarchically ordered sequences of tasksPlans capture assumptions
Column movement assumes enemy contact unlikely
Plans capture task dependencies Move_to_Holding_Area results in unit being at the HA,
(precondition to moving to the Battle_Position) OPFOR and Co must be at the Engage_area
simultaneously
Plan Generation Example
Destroyed(Enemy)
Attack(A, Enemy)
Move(A,BP) Engage(A,Enemy)
at(A,BP)at(A,FARP)
at(Enemy,EA)
at(A,BP) Destroyed(Enemy)
Destroyed(Enemy)
at(A,FARP)
at(Enemy,EA)
World Model
. . .
init
Company A plan
Company B plan CSSplan
Move Move Move
Move Move Move Move
Engage EngageReturn ReturnMove
OPFOR Plan
Move Move
Move
Battalion Tactical Plans
CoDeep Attack
CoDeep Attack
FARPOperations
Situation Interrupts Happen!
destroyed(Enemy)
Attack(A, Enemy)
Move(A,BP) Engage(A,Enemy)
at(A,BP)at(A,FARP) at(A,BP) destroyed(Enemy)
destroyed(Enemy)
at(A,FARP)
at(Enemy,EA)
Current World
active(A)
Star
t of
OP
ADA
Attack
active(A)active(A)
Reacting to Situation Interrupt
Situations evolve unexpectedly Goals change, actions fail, intelligence incorrect
Determine whether plan affected Invalidate assumptions? Violate dependency constraints?
Repair plan as needed Retract tasks invalidated by change Add new tasks Re-compute dependencies
Technologies and R&D:Collaborative Planning
Collaborative Planning
Represent plans of others Extend plan network to include others’ plans
Detect interactions among plans Same as with “normal” plan monitoring
Apply planning modulators: Organizational roles What others need to know Phase of the planning Stance of the planner wrt phase and role
Plan Interaction Example
Move(A,BP) Engage(A,Y)
Dead(Y)
Move(CSS,HQ)
at(CSS,HQ)at(CSS,FAA)
at(gas,FAA) at(gas,HQ)
at(A,BP)at(A,FAA) at(A,BP)
at(gas,FAA)
Op e
rati
on B
egin
s
Combat Service Support Plan
Attack Helicopter Company Plan
resupplied(HQ)
Planning Stances
Authoritative Order subordinate to alter his plans
Deferential Change my plans to de-conflict with superior
Helpful Help peer to resolve conflicts in plan
Self-servingAdversarial
Try to introduce conflict in other agent’s plan
Elaboration: Being Helpful
Planning issues Propose doing activities that facilitate others’ plans Avoid introducing threats into others’ plans
Communication Issues Collaboration protocols: propose, accept, counter Relevance reasoning
Which of my tasks would others want to know• e.g. “Honey, I’m going to the market”
Elaboration: Self-serving
Planning issues Notice things that others might do for me Ignore threats I introduce into other’s plans
Unless that keeps them from doing things for me
Communication Issues Deception
e.g. Someone might not help me if the knew what I was really planning
Plan Management
Must model when to use different stances Involves organizational issues
Where do I fit in the organization
Stances may need to change over timeDuring COA Analysis, adopt an adversarial stance towards ones own plans
Must model how stances influence planning How do we alter COA generation
C2 Entity Architecture
Planner(General Purpose
Reasoner)
Plan Manager Management
Plans
Management Plans
Tactical PlansTactical Plans
ManagementTheory
(domain independent)
ManagementTheory
(domain independent)
Tactical Domain Theory
Tactical Domain Theory
World ModelWorld Model
Situation Assessment
Synthetic Battlespace
Situation Reports, Sensing
Facts, inferencesExpectations
OPORDER Other Communications
When to Use a Stance
Model the collaborative planning process Includes management tasks that modulate the
generation of tactical plansTasks refer to specific tactical plansSpecify preconditions on changing stance
Includes knowledge of one’s organizational role
Planner constructs management plans Use same mechanisms as tactical planning
Management Plan Example Explicitly model the Military Decision Making
Process
COADevelopment
Authoritative towards subordinatesDeferential towards superiorsAdversarial towards OPFOR
COAAnalysis
Authoritative towards OPFORAdversarial towards self (war gaming)
Tasks Stances
Implementing Stances
Implemented as search control on planner Plan manager
Takes executing management tasksGenerates search control recommendations
Example: Deferential Stance When giving orders to subordinates
Indicate subset of plan is fixed (defer to this)Indicate rest of plan is flexible
Plan manager enforces these restrictions
Interaction Example
Move(A,BP)
Move(CSS,HQ)
at(CSS,HQ)at(CSS,FAA)
at(gas,FAA) at(gas,HQ)
at(A,BP)at(A,FAA)
at(gas,FAA)
Init
ial S
tate
PlannerRetract
Retract
Deferential towards
Combat Service Support Plan
Make CSS Planner defer to Company A’s Plan
Manager
C2 Entity Architecture
Planner(General Purpose
Reasoner)
Plan Manager Management
Plans
Management Plans
Tactical PlansTactical Plans
ManagementTheory
(domain independent)
ManagementTheory
(domain independent)
Tactical Domain Theory
Tactical Domain Theory
World ModelWorld Model
Situation Assessment
Synthetic Battlespace
Situation Reports, Sensing
Facts, inferencesExpectations
OPORDER Other Communications
Technologies and R&D:Situation Awareness
Situation Awareness
Planner needs a consolidated picture of the current situation in the battlespace Determines which goals and tasks are achievable Influences the choice of strategies and actions Allows the detection of imminent plan failure Enables re-planning
Situation assessment produces a current World Model Monitor plans with respect to world model Situation awareness = world model + plans/tasks
Situation Assessment
Performed at multiple echelons Scouts performing reconnaissance of battlespace C2 staff assimilates scouting and sensor reports
General process: Identify entities Classify groups of entities as units Determine units’ functionality, capabilities, plans, intent
Technical Issues Pilot awareness and information overload Situation assessment techniques
Pilot Situation Awareness
Synthetic worlds are information rich 100’s of other entities Vehicle instruments Terrain, weather, buildings, etc. Communications (messages) Amount of information will continue to increase ….
Perceive, understand, decide and act Comprehend dynamic, complex situations Decide what to do next Do it!
Information Overload
Roots of the Problem
Naïve vision model Entity-level resolution only
Unrealistic field of view (360o, 7 km radius)
Perceptual-Cognitive imbalance Too much perceptual processing
Cognitive system needs inputs, but …
It also needs time to respond to world events
Approach
Create a focus of attention Apply attention mechanisms to entity perception initially Incorporate filters Implement a zoom lens model (covert attention)
Stages of perceptual processing Attention in different stages: preattentive & attentive
Control the focus of attention Goal-driven Stimulus-driven
Zoom Lens Model of Attention(Eriksen & Yeh, 1985)
Attention limited in scope Multi-resolution focus Magnification inversely proportional to field of view
Low resolution Large region, encompassing more objects, fewer details Perceive groups of entities as a coherent whole
High resolution Small region, fewer objects, more details Perceive individual entities (e.g., tank, truck, soldier)
Low Resolution
Perceptual Grouping
PreattentiveGestalt grouping
Involuntary Proximity-based Other features
DynamicVoluntary grouping
K
K
K
Group Features
Quantity and compositionActivity
Moving Shooting
Location Center-of-mass Bounding-box
Geometric relationships wrt pilot Slant-range, azimuth, etc.
High Resolution
Entity Features
Location (GCS) Speed Velocity Orientation Slant Range Force Object, Object Type Vehicle Class Function Sense Name
Altitude Angle Off Target Aspect Magnetic bearing Heading Status Lateral Range Lateral Separation Closing Velocity Vertical Separation
Control of Attention
Goal-driven control Agent controls the focus / resolution of attention
Low resolution: Scouting groups of enemy; escorting group High resolution: Search for air-defense entities; engage target
Sets filters that select entities for WM
Stimulus-driven control Attention can be captured involuntarily by a visual event
Muzzle flash (luminance contrast, abrupt onset) Sudden motion (abrupt onset)
Sea
LandOverwatchPosition
OverwatchPosition Transport Carrier
Escort Carrier
Rendezvous Point
Escort task• Orient on group
• Voluntary grouping
Goal-driven Attention
Low Resolution High Resolution
Stimulus-driven Attention
Situation Awareness at Higher Echelons
Command Entity
Command Entity
Command Entity
Situation Reports Situation Reports
Situation Reports
Situation Assessment
Identify entities Fuse scouting reports
Classify groups of entities as units Cluster entities into unit-sized groups Classify units into functional types
Determine capabilities, plans, intent
Clustering and Classification
Bottom-up and top-down approachBottom-up clustering based on
proximity Identify a group of entities close to each other Other useful features: color, orientation, speed
Top-down classification based on doctrine Threat templates Issues: which template, partial matching
Bottom-up Clustering
Hierarchical Clustering Partitioning starting at the top until a satisfactory level
(e.g. individual units)
Robust Clustering Nearest-neighbor using center of mass
Works well for hierarchical clustering Requires a parameter of minimal distance
Density-based clustering Works well on different shapes of patterns No parameter is required (or can be learned)
Top-Down Classification
Classification and prediction Classification based on threat templates
Doctrine of situations, actions, formation and capacities Matching clustered units with templates for classification
Partial matching to predict the location of missing units
Encoding threat templates Encoding spatial information for symbolic processing
kD-tree to encode spatial relationships
Adding possible actions to nodes (units)
Future Situation Awareness
Model how tactical intelligence influences planning
Future situation: knowledge goals What will I need to know for this plan to work? Establish Priority Intelligence Requirements (PIR)
What commander needs to know about opposing force Drives the placement of sensors and observation posts Constrains the pace of plan execution
Rarely addressed in current C2 models
Intelligence Critical for Realistic C2
Close interplay between intelligence and COA Development
Intelligence guides COA development
COA development drives intelligence needs
Intelligence availability constrains actions• Some COA must be abandoned if one can’t gather
adequate intelligence
Intelligence Critical for Realistic C2
Intelligence imposes temporal constraints
When can a satellite observe?
How long to insert surveillance (LRSU)?
How long before I must commit to COA?
Intelligence critical for realistic C2
Intelligence collection must be focused Commanders must:
Prioritize their intelligence needsUnderstand higher-level intelligence prioritiesProvide intelligence guidance to subordinates
e.g. Simulation Information Filtering Tool [Stone et. al]
Brigade Planning (simplified)
Identify Engagement Area (EA Pad)Should canalize OPFOR and restrict movement
Identify launch time Require 2-hour notice EA Pad
AALincoln
Attack 2nd echelon tank division (TD)
PL ECHO
Brigade PIR
When will TD leave AA Lincoln?Verifies enemy intent
When will TD reach PL Echo? Satisfies the need for 2-hour noticeFurther verifies enemy intentLocation of PL Echo driven by PIR
EA Pad
AALincoln
2hrs
EA Pad
PL ECHO
Intelligence Plan
SLAR Monitor movement from assembly area
LRSU Trigger attack: TD 2hrs from EA Pad
Assembly Area
Final Brigade Plan
Execute Mission
Arrive at EA
Break Contact
DecisionPoint
H H+2 H+3H-8H-10
Insert LRSU LRSU monitor PL Echo
Deep Attack
SLAR monitor AA
Automating PIR
Identify PIR in my own plans Find preconditions, assumptions, and triggering
conditions that are dependent on OPFOR behavior
Extract PIR from higher echelon orders Specialize as appropriate for my areas of operation
Derive tasks for satisfying PIR Sensor placement
Ensure consistency of augmented plans
Identifying PIR
Examine COA dependencies on OPFOR e.g. Precondition of engaging:
OPFOR will-be-at EA Pad at time H+2
Look for dependencies that: Are not under my direct control Are uncertain
Implemented with PIR recognition schema: Abstract rules that scan plans and assert PIR
Some domain-independent, some domain-specific
Interpreting Higher Level Guidance
Need to convert into PIR at my echelon e.g. Brigade’s PIR:
When will lead regiment reach forward defense
becomes Battalion PIRWhen will lead battalion of lead regiment reach fwd
def
Implemented by specialization rules Encode doctrinal and terrain relationships
Deriving Sensor Plans
Implemented via tactical planning mechanism PIR represented as “knowledge goals” Domain theory augmented with sensing tasks
Sensing tasks achieve knowledge goalsTasks encode maneuver / temporal dependencies
Planning process fills in detailsSensing tasks added to achieve knowledge goals
• e.g. Observe TD activity near PL_ECHOOther tasks added to satisfy maneuver dependencies
• e.g. Use UH-60 to insert LRSU near PL_ECHO
Ensuring Consistency
Implemented via tactical planning mechanism
If PIR goals cannot be satisfied, COA is invalidor
Use unsatisfied PIR to request external assets
Sensing plans constrain timing of events If temporal constraints inconsistent, COA is invalid
Significant Results
Significant Results
Continuous planning paradigm works well for modeling C2 behavior in the joint synthetic battlespaces Dynamic planning, monitoring, and execution Handles situation interrupts in test cases
Collaborative planning is made possible by adding a few extensions to a general purpose planner
A model of perceptual attention and situation awareness implemented in RWA-Soar pilot
Developed a technique for deriving Priority Intelligence Requirements with planner
Significant Results (2)
Publications Continuous Planning and Collaboration for Command
and Control in Joint Synthetic Battlespaces, CGF&BR ‘99
Deriving Priority Intelligence Requirements for Synthetic Command Entities, CGF&BR ‘99
Modeling Perceptual Attention in Virtual Humans, CGF&BR ‘99
Perceptual Grouping and Visual Attention in a Multi-agent World, Agents ‘99
Scope of Task Coverage
ATKHB Attack MissionAchieve Tactical
Disposition
Reduce Enemy Posture
Achieve Culminating
Task
Consolidate
1-4-1101: Personnel (S1) planning (C2)1-4-1201: Intelligence (S2) planning (C2)1-4-1301: Operations (S3) planning (C2)1-4-1401: Logistics (S4) planning (C2)1-4-1302: Establish and maintain
tactical operations center (C2)1-4-1305: Coordinate maneuver with
CSS and rear ops (C2)---------------------------------------------------1-2-0320: Provide supply support (CSS)1-2-7723: Perform maintenance (CSS)1-2-7728: Process ammo and fuel (CSS)1-4-1103: Replacement operations (CSS)1-4-1402: Coordinate supply/equip. (CSS)1-4-1405: Plan and coordinate transport
assets (CSS)
Achieve Readiness
1-3:0001: Plan and organize move (Mnv)1-2-0101: Move to and occupy assembly
area (Mnv)1-4-1306: Establish and maintain tactical
command post (C2)1-2-7726: Conduct FARP operations (CSS)
Achieve Physical Posture
1-4-1305 (Section 6.1.2): Integrate fire support
Attack (METL task) 1-4-1206:
1-2-xxxx: Establish satellite comm. (C2)1-2-xxx0: Establish ground comm (C2)1-2-7509: Establish voice comm (C2)11-5-0104: Establish FM radio (C2)1-4-1001: Perform C2 operations (C2)1-4-1303: Control tactical operations (C2)------------------------------------------------------------1-4-1202: Implement security measures (Int)1-4-1203: Process intelligence information (Int)1-4-1311: Liaison operations (Int)------------------------------------------------------------1-4-1105: Provide admin services (CSS)1-2-7708: Provide food support (CSS)1-2-7710: Operate field mess (CSS)1-2-7720: Establish med support (CSS)1-2-7721: Conduct med activities (CSS)1-4-1102: Perform strength management (CSS)1-4-1104: Conduct casualty reporting (CSS)1-4-1308: Direct army airspace C2 (CSS)1-4-1310: Civil-military operations (CSS)1-4-1403: Monitor equipment readiness (CSS)1-4-1406: Provide logistic services (CSS)
Continuous Tasks
Legend
Implemented
Partially implemented
Desire to implement
Less relevant
Expected Results
Detailed evaluation of planner Empirical Analytical
Extended model of situation awareness at entity and C2 levels Attention, hierarchical clustering, classification, fusion
Extended model of collaboration Abstract technical description of planner Journal articles and conference papers
Measures of Success
Collective Measure Ability of a group of entities (RWA Battalion) to achieve
mission objectives in scenarios containing a wide range of situation interrupts
Individual Measures Scalability: size of groups that can act autonomously Flexibility: classes of situation interrupts handled by group
behavior Types of multi-agent reasoning integrated into framework
i.e., collaborative, adversarial, temporal, ...
Breadth and depth of domain knowledge e.g., # of tasks, echelon levels, functional categories (battlefield operating
systems)
Evaluation
Empirical Developed scenario generator, logging function Will collect data from scenarios run in batch mode Encode additional domain knowledge (WARSIM?) Evaluate scalability
Analytical Develop abstract description of planner Complexity measures for scalability Analyze properties of collaborative planner -- can it be
de-coupled from Soar-CFOR implementation?
Technology Transition
Efforts
Formulated concept for C2 in NASMDemo at JPMR in February ‘99Presented 3 papers at CGF&BR, May ‘99
Perceptual attention, C2 Modeling, PIR
JSIMS/ASTT workshop, May ‘99 WARSIM commonality (POC’s: Milks & Karr) ONESAF?
Problem Areas
Focused Efforts Required
Not yet addressing role of learningNeed good evaluation
Scalability, robustness, efficiency, …
Programmatic Issues
Schedule
Milestone 4: 12/98 Design Review 2
Approach to learning improved group modelsApproach to temporal planning
Schedule (2)
Milestone 5: 9/99 (revise to 12/99?) Technology POP Demonstration 3
RWA Attack Battalion Demonstrate advanced group understanding Demonstrate more advanced group planning
• Temporal planning• Group understanding: plan recognition
Demonstrate advanced group execution• Commander utilizes teamwork model (scaled down)
Demonstrate group learning• Improve group models through experience
Deliver software and domain independent descriptions of new capabilities
Demonstration
Demonstration Scenario
Attack Helicopter Battalion (AH-64) Battalion Commander 3 Helicopter Companies
Company Commanders Apache Pilots
1 Combat Service Support Commander
Deep Attack Mission Scenario Companies move from Assembly Area to Holding Area Situation interrupt: unexpected enemy forces in Holding
Area Dynamically re-plan and execute mission