文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 工程科技 > 信息与通信 > Convergence Rate of Distributed Consensus for Second-Order Multi-agent Systems
侵权投诉

Convergence Rate of Distributed Consensus for Second-Order Multi-agent Systems

2010 8th IEEE International Conference onControl and Automation

Xiamen, China, June 9-11, 2010

FrD7.6

ConvergenceRateofDistributedConsensusforSecond-OrderMulti-agent

Systems

ZhongYuandLihuaXie

Abstract—Thispaperisconcernedwiththeconsensuscon-vergencerateforsecondordermulti-agentsystems.First,whenbothpositionandvelocityareexchangedbetweenneighboringagents,thefastestconsensusconvergencerateunderagivenprotocolisderivedbasedontheassumptionthatalltheeigenvaluesoftheLaplacianmatrixarereal.Next,inthecasewhenonlythepositioninformationisexchangedbetweenagents,anobserver-basedprotocolisproposed.Weshowthatwithaproperlychosenobserverstructureaseparationprinciplecanbeestablished,namely,thepolesoftheclosed-loopmulti-agentsystemaretheunionoftheobserverpolesandthepolesoftheclosed-loopsystemunderfullstateinformationexchange.Assuch,anobservercanbedesignedtoachievethesamefastestconsensusconvergencerateasthatforthecasewithfullstateinformationexchange.Simulationsareprovidedtoverifythe ndingsandcomparewithexistingmethods.

I.INTRODUCTION

Thedistributedconsensusproblemfornetworkedmulti-agentsystemshasattractedalotofattentionsinrecentyearsduetoitsbroadapplicationsindistributedcomputation,sensorfusion,formation ight,networksynchronization,etc.Consensuscontrolmeanstodesignanetworkedinteractionprotocolsuchthatalltheagentsreachanagreementontheirstatesasymptoticallyorin nitetime.In1995,Vicseketal.[1]proposedamodeltodescribeagents’motioninaplanewiththesamespeedbutdifferentheadingsandJad-babaieprovidedasimpleconsensusprotocolwhichupdatestheheadingofanagentbyaveragingtheheadingsofitsneighbors[2].Olfati-SaberandMurraystudiedtheproblemusinggraphtheoryandprovidedconditionsforconsensusunderswitchingtopologiesandconstanttimedelays[3].RenandBeardthengeneralizedtheirresultsandproposedamorerelaxedconditionforthenetworktopologytoreachconsensus.Theconditionisgivenintermsoftheexistenceofaspanningtree.Otherstudiesfordifferentmodelsandconsensusprotocolscanalsobefound,seee.g.[4]andreferencestherein.

Itisworthnotingthattheconsensusproblemhasprimarilybeenstudiedfor rst-orderkinematics.However,inmanyapplications,bothpositionandvelocityinformationmaybeofinterest,forinstance,inmulti-robotformationcontrol.Thestudyofthesecond-orderconsensusproblemcanbetracedbackto2003,whenOlfati-SaberandMurray[5]investigatedthe ockingproblem.XieandWang[6]alsoproposedaprotocolanddiscussedtheconsensusconvergencerate.RenandAtkins[7]proposedaprotocolsimilartothatof[5]and

ZhongYuandLihuaXiearewiththeSchoolofElectricalandElectronicEngineering,NanyangTechnologicalUniversity,Singapore639798,E-mail:{yuzh0005,elhxie}@ntu.edu.sg.

gaveasuf cientconditionforconvergence.Inmostexistingprotocolsforsecondordermulti-agentsystems,thevelocityinformationisrequired,whichissometimesnotpractical.Forthecasewherenovelocityinformationisexchangedbetweenagents,Ren[8]studiedtheproblemforundirectednetworksandprovidedasolutiontotheconsensusproblem.Amaindrawbackofthealgorithmin[8]isthatitrequiresthenetworktobeundirected.Suetal.[9]modi edtheprotocolforthestudyof ocking.However,thenetworkisstillrequiredtobeundirected.Gaoetal.[10]proposedanewprotocolforbalancednetworkswhichiscomplicatedincalculation.

InthispaperweshallstudythetheconvergencerateforaclassofnetworktopologieswhoseLaplacianmatrixhasallrealeigenvalues.Whenbothpositionandvelocityareex-changedbetweenagents,thefastestconsensusconvergencerateunderagivenprotocolisderived.Forthecasewhenonlythepositioninformationisexchangedbetweenagents,anobserver-basedprotocolisproposed.Weshowthatwithaproperlychosenobserverstructureaseparationprinciplecanbeestablished.Inthiscase,theobserverandcontrolprotocolcanbedesignedseparately.Assuch,thesamefastestconsensusconvergencerateasthatforthecasewithfullstateinformationexchangecanbeachieved.Simulationsareprovidedtoverifythe ndingsandcomparewithexistingmethods.

Theremainderofthepaperisorganizedasfollows.SectionIIprovidessomepreliminariesandde nesnotationsusedinthepaper.SectionIIIanalyzestheconvergencerateforaclassofnetworktopologiesforsecond-ordermulti-agentsystems.SectionIVproposesanobserver-basedapproach.SimulationresultsareprovidedinSectionVtodemonstratetheresults.ConclusionsandfutureworkareaddressedinSectionVI.

II.PRELIMINARIESANDNOTATIONS

Considerthenetworkedmulti-agentsystemwith xedtopologyG=(V,E,A),whichisadigraphofordernwithnodesetV={v1,v2,...,vn},edgesetE V×VandthenonnegativeadjacencymatrixA=(aij).AnedgeofGisdenotedbyeij=(vi,vj)ifnodevjcanreceiveinformationfromnodevi.TheadjacencymatrixAisde nedsuchthataij=1ifeji∈Eandaij=0else.Denotethesetofneighborsofnodevi byNi={vi1,vi2,...,vini}.Let

n

D=diag(dii)wheredii=j=1aij.TheLaplacianmatrixofthegraphisde nedasL=D A.1nisthen×1columnvectorwithallentries1.ObviouslywehaveL1n=0.Inisthen×nidentitymatrix.

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email