Distributed Coordination of Multi-agent Networks

WeiRenYongcanCaoDistributedCoordinationofMulti-agentNetworksEmergentProblems,Models,andIssuesWeiRenDept.Electrical&ComputerEngineerin[e-mail]&ComputerEngineeringUtahStateUniversityOldMainHill412084322-4120LoganUtahUSAISSN0178-5354ISBN978-0-85729-168-4e-ISBN978-0-85729-169-1DOI10.1007/978-0-85729-169-1SpringerLondonDordrechtHeidelbergNewYorkBritishLibraryCataloguinginPublicationDataAcataloguerecordforthisbookisavailablefromtheBritishLibrarycSpringer-VerlagLondonLimited2011MATLABRandSimulinkRareregisteredtrademarksofTheMathWorks,Inc.,3AppleHillDrive,Natick,MA01760-2098,USA.:eStudioCalamarS.L.Printedonacid-freepaperSpringerispartofSpringerScience+BusinessMedia()ThisworkisdedicatedtoMyparents,HongyiRenandLiyingWang,andMyson,Kaden–WeiRenMywife,YumanWei,andMyparents,GuangyaoCaoandYanpinXie–YongcanCaoPrefaceInthepasttwodecades,scientistsacrossdiversefieldhavebeentryingtoidentifytheunderlyingmechanismsofnetworkedsystems.Biologistsusenetworkstostudytheworkingandwiringoftranscriptionalregulatorycircuits.Sociologistsusenet-workstopredictthebehavioroftechno-socialsystems.Physicistsusenetworkstomodelandpredicttheemergenceofbehaviornorms,andusequantitativemethodstoanalyzetheresultingnetworkedsystems.Acrossdifferentfieldsnetworkscien-tistsaremakingadramaticprogressandpushingnetworkanalysistoitslimits.Inengineering,researchersstudyhowtoassembleandcoordinateindividualphysi-caldevicesintoacoherentwholetoperformacommontask.Thisgivesrisetoaveryactiveandexcitingresearchfield—multi-agensystems.Withnumerouscivil-ian,homelandsecurity,andmilitaryapplications,multi-agentsystemsplacehighdemandsonfeaturessuchaslowcost,highadaptivityandscalability,increasedflxibility,greatrobustness,andeasymaintenance.Tomeetthesedemands,thecur-renttrendistodesigndistributedalgorithmsthatrelyononlylocalinteractiontoachieveglobalgroupbehavior.Thisbookintroducesemergentproblems,models,andissuesindistributedco-ordinationofmulti-agentnetworks.Theseproblems,models,andissuesrepresentsomeemergentresearchdirectionsinthefielofmulti-agentsystems.Emergentproblemsincludecollectiveperiodicmotioncoordination,collectivetrackingwithadynamicleader,andcontainmentcontrolwithmultipleleaders.Theseproblemsextendtheexistingapplicationdomainsofmulti-agentnetworks.Inparticular,col-lectiveperiodicmotioncoordinationisappropriateforapplicationsinvolvingagentnetworkswithrepetitivemovements,collectivetrackingguaranteestrackingofady-namicleaderbymultiplefollowersinthepresenceofreducedinteractionandpartialmeasurements,andcontainmentcontrolenablesmaneuveringofmultiplefollowersbymultipleleaders.EmergentmodelsincludenetworkedLagrangiansystemsandnetworkedfractional-ordersystems.Thesemodelsresultfromphysicalconstraintsorcomplexenvironmentsinwhichmulti-agentsystemsoperate.Inparticular,La-grangianmodelsrepresentaclassofmechanicalsystemsincludingautonomousve-hicles,roboticmanipulators,andwalkingrobots,andfractional-ordermodelsaremorerealisticrepresentationsofsystemsoperatingincomplexenvironmentsthanviiviiiPrefaceinteger-ordermodels.Emergentissuesincludethesampled-datasetting,optimalityaspect,andtimedelay.Theseissuesarerealisticandimportantinreal-worldappli-cations.Inparticular,thesampled-datasettingisrelevantwhenagentsinteractwiththeirneighborsintermittentlyratherthancontinuously,theoptimalityaspectplaysanimportantroleindesigningenergy-efficienalgorithms,andthetimedelayeffectisinevitableinnetworkedsystems.Thisbookisdividedintofourparts.Thefirspartintroducespreliminaries(Chap.1)andoverviewsrecentresearchindistributedmulti-agentcoordination(Chap.2).Thesecondpartintroducesemergentproblemsindistributedmulti-agentcoordination,namely,collectiveperiodicmotioncoordination(Chap.3),collec-tivetrackingwithadynamicleader(Chap.4),andcontainmentcontrolwithmul-tipleleaders(Chap.5).Thethirdpartintroducesemergentmodelsindistributedmulti-agentcoordination,namely,networkedLagrangiansystems(Chap.6)andnetworkedfractional-ordersystems(Chap.7).Thefourthpartintroducesemer-gentissuesindistributedmulti-agentcoordination,namely,sampled-datasetting(Chap.8),optimali