PIOVRA - Liophant Simulation

PIOVRAGoals• The development of operative Prototypes to be Run in aFederation for completing the Dynamic VV&A andExecution Phase of new generation intelligent CGF.• To develop a new generation of Computer GeneratedForces (CGF) PIOVRA (Polyfunctional Intelligent OperationalVirtual Reality Agents), to be used both for exercise scope,both for defence planning and support operations in anHLA Federation.• New CGF should be in some extent “Intelligent”,meaning that they should demonstrate co-operative andcompetitive behaviours (co-ordinating units both duringoperative actions and situation evaluation) based on thecurrent boundary conditions and situation.2

PIOVRAResearch Motivations• To develop a new Generation of CGF able to simulate“Intelligent” behaviour, filling up the gap between userrequirements and current available CGF performances• To create PIOVRA intelligent CGF as effective modelsto be integrated in HLA Simulation for:– Training– Operation Planning– Operation Support• To guarantee the possibility to Define/ConfigurePIOVRA CGF using Libraries and Effective Paradigmsin order to guarantee Accreditation, Effectiveness andUsability of PIOVRA developments3

PIOVRAPIOVRA Military Relevance• Development of a New Generation of EffectiveIntelligent Units to be used on:– Exercise– Defence Planning– Support to Operations,.• The expected benefits of using PIOVRA CGF are:– reduction of human personnel operating directly the simulation system– increasing objectivity degree in the actions and the reactions of any ofthe simulated entities present in the battlefield (i.e., friend, foe orneutral)– Availability of “intelligent” PIOVRA CGF able to describe the reasonsbehind a particular operational behaviour (allowing to verify in anindirect manner, the doctrine, tactics and ROE).4

PIOVRANew CGF Application AreasPIOVRA developments will be useful in :• Exercise functional area:– PIOVRA CGF constitute the ideal “sparring partner” due to the constancy of theirreactions and their possibility to reproduce realistic opponent actions and reactions.• Defence Planning area:– PIOVRA CGF consents to verify operational plans in less time due to the absence,complete or partial, of human experts employed in the different roles foreseen by theOperational Plan under exam; also in this case the objectivity, the reaction constancy, theCGF decision motivation traceability represent significant advantages.• Support to Operations area:– PIOVRA CGF are an essential element in performing realistic verification and ongoingreal tactical situations possible progress evaluation by inserting them in the subjectoperation’s simulation. Due to elevated degree of realism and extremely rapid simulationfeedback these use become realisable only through the use of intelligent CGF.5

New CGF CharacteristicsThe PIOVRA CGF includes the following characteristics:PIOVRA• User-Defined Initialising Parameters• Analyse Surrounding Environment and React Respectively Capability• Cooperative Capacity• Force Aggregating/Desegregating Capability and relevant military hierarchy• Resultant Aggregation Levels different from aggregating/desegregating elementssum/subtraction• Limit Proper Autonomy to Achieve Common Objective Capability• Stress Level Indicator applicable for the entities behaviour definition• Implementation of Typical Human Behaviour (survival instinct and moral/ethicalmotivations)• Distinct Friend, Enemy, Suspect and Neutral Units• Explicit ROE justifying Proper Behaviour• Military Reports to Higher Commanders Capability• Decision Process Traceability• Feedback capability• CGF entities simulating various force levels6

PIOVRANew CGF & Models• New CGF need to be designed for incorporating anhierarchical scalable structure in order to be able toreach high level of details (i.e. single persons) withoutloosing the possibility to model large entities and withautonomous reporting capabilities for justifying theirchoices to external user.• G-DEVS/HLA developmentsof PIOVRA guarantee the possibilityto plug such models in PIOVRA CGFrespecting consistency and speedingup VV&A processes.7

PIOVRAPIOVRA Units• PIOVRA pays great attention into modelling NeutralUnits representing civilians and their specificbehaviours and logic• Modelling special units such terrorist:a significant part of PIOVRA is devotedto create models for reproducingdynamically the human behaviour in complex scenario.• PIOVRA entities include psychological parameters andmodels (i.e. “stress level” as aggregation level and ofexternal situation function).• The capability to reproduce such behaviour is based onthe use of Artificial Intelligence directly integrated inthe project9

PIOVRAPIOVRA Modeling• PIOVRA Conceptual models for simulating acooperative behaviour of PIOVRA CGF allowing theiraggregation or separation depending on the situation(clearly keeping in mind the command hierarchies for military units andmanaging their dynamic evolution during actions).• Modelling all the aspects and parameters for properlydefine the different units in terms of Rules OfEngagement (ROE); the use of Fuzzy Logic andArtificial Neural Networks allows to extend the validityrange of PIOVRA Models10

PIOVRAPIOVRA Components• PIOVRA Library allows to attribute specific intelligentbehaviours to the operative units.• PIOVRA CGF are integrated using High LevelArchitecture (HLA) in a Federation• PIOVRA Federation as Executable Scenario withextended Capabilities for Testing and Experimentingthe Models• PIOVRA Reports and Feed-backsfor guaranteeing the dynamic changeof Rules of Engagementfor PIOVRA CGFCustomersWarehouseShippingSupplierSiteItemHistoricalOrderSistemComponentSystem TypeUserBill of MaterialSystem TypeSpare PartsResourceBill of MaterialComponentComponent TypeCalendar11

PIOVRAExpertise for New CGF• Development of New Generation CGF is a quitebreakthrough research and need to include real“intelligent” behaviours (in operative term) and able tobe used in future simulation.• This Challenge is enabled by very advancedtechnologies and strong experiences in their area:– Artificial Intelligence– G-DEVS– HLAAddestramentoArtificialNeuralNetworks6 weeks 6 weeks 6 weeks 6 weeksForecastsPrevisione tramite ANNFusionFusion Q%Add ∆QAdd ∆TAdd Acc12

PIOVRAAI & PIOVRA• The use of AI (Artificial Intelligence) in hybrid hierarchicalmodels request to combine different techniques in order tobe successful.• The PIOVRA project expect to experiment differentintegration architecture combining:– Fuzzy Logic (balancing Boundary Conditions &FRCScenario)– Artificial Neural Networks (self-learning)– KBS (behaviour justification)– Data Fusion (Situation Assessment, Threat Assessment)– Swarm Intelligence (Cooperative Intelligent Behaviour)FRC 12SN List52SNFRCPRM* ID C.Fusion ∆Q- ∆Q+ ∆Τ- ∆Q+ Orz ε* 130-2818/03 50 10 20 1 2 52 0.05.. . .ANN_DDM ANN.FRC_FFR ANN13

PIOVRAIntegration & HLA• The interoperability of the PIOVRA components isstrictly based on HLA in order to maximise theirintegration capabilities and improve the efficiency oftheir hierarchical structure.• The Development of PIOVRA Units asHLA federates support the accreditationand testing of their performances• PIOVRA CGF communicating/cooperating/competingthrough HLA allows to guarantee potential extensiveuse of PIOVRA units on complex scenario just byadding federates.14

PIOVRAModeling & G-DEVS HLA• The design and development of PIOVRA Models isexpected to be realised in a G-DEVS/HLA framework,tailored for PIOVRA, able to consider both thecontinuous components (i.e. movement) and discrete(i.e. events and actions) integrate in High LevelArchitecture.• The PIOVRA model repository will be developed as aneffective Library based on G-DEVS paradigm in orderto guarantee the parameters and behavioural firmwarepre-determination very precisely and carefully byexpert teams ofautomated components15

PIOVRAPIOVRA PartnersThe use of Excellence Centers with long tradition inresearch applied to industrial projects and mutualcooperation is a critical issue for guarantee the success ofPIOVRA initiative:• MISS DIPTEM, Genoa University– Leader & Coordinator• LSIS, Marseille• Liophant SimulationIEPAL16

PIOVRAScenario Definition• PIOVRA includes a detailed scenario definition to beused for metrics and performance evaluation of themodels developed and their fidelity and effectiveness.• A demonstration phase on a complex scenario isexpected to be used as framework for experimentingand analysing the PIOVRA federation; the scenario isexpected to be integrated with war gaming systems (i.e.JTLS) in order to simulate/study specific situations.• PIOVRA VV&A represent a very important andsignificant task in order to guarantee the success of theinitiative.17

PIOVRAPIOVRA Prototype• PIOVRA project main goal is to complete a Prototypefor completing Experimental Analysis based on DOE(Design of Experiments) for the VV&A and Executionof PIOVRA models.• The Project Partner previous experiences guarantee thepossibility to benefit from their skills and libraries inorder to obtain also with the Budget Requested anExecutable System for full Validation of PIOVRAmodels over Operative Scenarios.• Prototype integrated with war gamingsystems in order to simulate/studyspecific situations.18

PIOVRA ObjectsPIOVRAComportment Objects are dedicated to the simulation of actors thatrepresents behaviors of populations, movements or analog entities towhere units on the field belongs.Action Objects are units that have the task to simulate particularelements acting in the scenario like a military unit, a terrorist, a politicalrepresentative. They can also simulate particular events like riots,demonstrations, etc. Part of the Action Objects is generated in functionof a particular state of one or more Comportment Objects.Support Objects include all the objects don't representing human actorsof the Scenario, but influent phenomena; for instance Support ObjectList includes Weather Object as well as Zone Objects.19

PIOVRA ZONESPIOVRAQSTUVWYIn the proposed context, themovement algorithms aredevoted to control any actionobject which moves insidePIOVRA Zone and each ZoneObject is an entity that includesmovement links and groundcharacteristicsAPOCBMNDFLEHGIZJKR20

PIOVRA MovementsPIOVRAjiji1 1513 1 2 20 2 3 4 536421

Psychological ModifiersPIOVRACurrently the research is focusing onthe Following Psychological Modifiersare under Analysis:FearStressFatigueAggressiveness22

FearPIOVRAEventGeneratingFearMontecarlo TechniqueFear Generated LevelSize/ForceVs.Event ImpactList of Object Zones Perceivingthe EventB%+K%+Z%+B%+Y%+X%=1Fear EffectCoolCharacteristicObject ZoneB%0% 100%X%K%FatigueResistanceIncreaseZ%N%Y%StressLevelIncreaseNoEffectParalysisEscapeAggressiveLevelIncreaseStress Level ModifierFatigue Level ModifierAction ObjectAction ObjectAction ObjectAction ObjectAction ObjectAction ObjectList of Action Objectsinside the Zone23

FatiguePIOVRAFL ( tji+1Hf ( FLwhere:⎧⎪y≤ 0⎪( ti), y)= ⎨⎪⎪y> 0⎪⎩( SOP ( t ), zone ( t )) ⋅ ( t − t ) ⋅ Hf ( FL ( t ), Dfc( SOP ( t ), zone ( t ))) = FL ( t ) + Dfc+jjijGsf ⋅ kffjji1Gsf ⋅ kff2ji 1FLj( ti)⋅ + 1bf1FLj( ti)+ 1bf2⋅FL ( t ) + 1FL j (t i ) is the Fatigue Level of the j-th Action Objectat i-th event timeSOP j (t i ) Status of Operation of the j-th Action Objectin i-th event timezone j (t i )Zone location of the j-th Action Object in i-thevent timeDfc(x,y) Unitary Continuous Change in Fatigue Leveldue to Status x in Environment yGff j Fatigue Factor Characteristics of j-thAction ObjectHf(x,y) Function for reproducing Hysteresis andSaturation on Fatigue depending on currentstatus as well as current increasejkff 1 , kff 2 Factors for tuning Hf Function Impactbf 1 ,bf 2 Factors for tuning Hf Function Periodjijiiiij( E ) Hf ( SL ( t Dfe( E)FL ( t ) FL ( t ) + Dfe ⋅ ),ijiiji= E i Event affecting the j-th Object at i-thevent timeDfe(x,y) Event Discrete Change in Fatigue Level dueto Status x in Environment yjjiFatigue is state correspondingto temporary loss of strengthand energy resulting from hardphysical or mental work;physical fatigue (equivalent tophysical debilitation for Pavlov)is generated by anything thatphysically weakens theorganism, this is supposed toincreases the likelihood of crisis(transmarginalinhibition).Fatigue Continuous VarianceEvent Affecting Fatigue24

Different Evolution TrendsPIOVRAFatigue is characterizedby saturation inincreasing and fasterrecovery during earlyphases in our range ofanalysisFatigue IncreasingAggressiveness IncreasingFatigue DecreasingAggressiveness Decreasing0 20 40 60 80 100 1200 20 40 60 80 100 120Aggressiveness increasesbased on exponentialescalation and relaxesback slowly25

Stress ReactionPIOVRACrisisEventPreliminaryEvaluationEvent PerceptionNot SignificantEventStress LevelRecoveryDanger, DamageMenace, ChallengePositiveEventSociologicalCopyingResourcesSecondaryEvaluationand ReactionAssessmentPsychologicalCopyingResourcesNegativeInfluencePositive InfluenceRationalReactionEmotionalReaction26

AggressivenessPIOVRAThe aggressiveness emerges usually as result of: Benefit Opportunity Perception (rationalaggressiveness); this is influenced by individualaggressive capacity, gender difference andbehavioral models Negative Emotions (hostile aggressiveness): this isrelated to a provocation (perception of an hostileattitude), personality difference and other emotionand negative feeling including physical pain,weather stress factor (hot) and irritation.Aggressiveness is heavily influenced by sociologicalconditions in term of size of the group, cultural models,media influence etc.27

Scenario ExampleCOMPLEX SCENARIO DAY X-3PIOVRA1ORANIAN2FOBSAXONIANCIMBRIAPIOVRA28

Scenario ExampleDAY X-2 E X-12 UAV / HELIPIOVRAPIOVRAShow the Flag129

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT HEmbassiesDemonstrationGovernorPIOVRA30

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT H + 1EmbassiesPIOVRAGovernorGovernor decides tostop demostrantionFederal Police closesthe main bridgeAnother platoon is onhigh alert statusGovernor decides toopen the bridgeCrowd starts to getnervous and violentBtg Com, asks toGovernor to let thepeople demostrate31

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT H + 2EmbassiesPIOVRAGovernorThe crowdconcentrates under thegovernor palaceSmall groups go to thewestern embassiesAdditional platoonstarts to move inStationary Platoonready to performevacuation/extractionof GovernorOne explosion withinthe crowd with somecivilians deads andinjuredFederal Police startsto close the area32

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT H + 3EmbassiesPIOVRAGovernorFederal police securingthe area andevacuating injuriedciviliansSmall groupsharassing westernembassiesCrowd dissolved butfuriousAdditional platoon in the area,ready to:-support Governor evacuation- defend western embassiesrStationary platoonstarted Governorevacuation33

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT H + 4EmbassiesPIOVRAGovernorGovernor evacuated inthe FOBWestern embassies stillunder harassmentDisorders all aroundthe cityAdditional platoon in the area,ready to defend westernembassies34

1 Platoon Liason Teams Federal PoliceSequena PoliceDAY X AT H + 5EmbassiesPIOVRAGovernorSituation unstable andunpredictableFederal Police and localPolice doing their best topacificate the cityAllied troops in the FOBUrban areas not under controll35

SIMULATION DAY X AT H00PIOVRASIMULATION AT COI 18 May 200636

SIMULATION DAY X AT H16PIOVRASIMULATION AT COI 18 May 200637

PIOVRASummarizing• PIOVRA Project is devoted to Create New CGFs thatintroduce innovative characteristics in SimulationScenarios• PIOVRA CGFs introduce a Competitive Advantage forusing Simulation in Exercise/Planning & OperationSupport respect existing tools and packages• PIOVRA results will be completely validated andverified as well as integrated in a Prototyperepresenting at least a very advanced stage ofdevelopment of a full capable Simulation Federatewww.liophantliophant.org/piovraF o recc a stsP a rttC o m m ere rcia liN u m b erssF r in e A N NF o rer eca stsO r d in iA rriv i P ror od u z io n eP r o d u zio n i in C o rsor M a g a z zin oB O M… .D I P U n ive r sity o f G e n o ah ttp ://st.itim .u n ig e .itC o n su m p tio n sF r in e S im u la to rF r in e M e tricsS istemm a In fo rmr m a tiv oP erfo rmr m a n cec es38