基于微型六足仿生机器人的多机器人群体协调控制

上海交通大学硕士学位论文基于微型六足仿生机器人的多机器人群体协调控制姓名：戴毅申请学位级别：硕士专业：精密仪器及机械指导教师：颜国正20040201-4-ResearchonCoordinationandCooperationofMultiMicroBiomimeticHexapodRobotsMarkovMarkov-5-AbstractInthispaper,weintroduceourresearchonsomekeyproblemsaboutMultiMicro-robotsCooperationSystem.Ourmultirobotssystemisbasedonanovelbiomimetichexapodmicrorobotthatisdevelopedinourlab.Thisrobotimitatesthelocomotionmechanismandgaitsmodelofhexapodinsects.Therobot,whichisdrivenbytwohigh-powersteppermotors,canperformsomebasicmovementswithsatisfiedvelocity,suchasmovingforward/backward,turningleft/right.Comparedwithtraditionalmobilerobots,ourrobotisblessedwithsomenotableadvantages,suchassmallgeometricsize,highadaptationandlowenergyconsumption.Besidesthese,ourrobotcanperformtasksefficientlyinun-structuredenvironments.Inthispaper,wealsodiscussourresearchcontributiononmultimicrorobotslocalization.Inoursystem,wedevelopedanovelprobabilisticapproachtomultimicrorobotslocalization,whichisbasedonPartialObservableMarkovProcess.Comparedwithtraditionallocalizationmethods,ourapproachutilizesthedrivinginformationfromeachrobotandfusesitwithmeasurementfromexternalvisionsensor,andthusproducesmorerobustresultswhileloosestherequirementonsystem’shardware.Inaddition,ourlocalizationmethodistotallydistributedandthushashighscalability.Attheendofthispaper,problemsonmultirobotscooperativeexplorationandmapbuildingarediscussed.Weintroduceoursolutionstosomekeysub-problemsinthisfield,suchasmodelingofstructuredenvironment,mapmerging,localizationofotherrobotinpartialmapsandarchitectureformultirobotsexploration.Forexample,weusehierarchicalBayesianNetworktomodelthestructuredenvironment,suchasartificialmaze.Basedonthisenvironmentmodel,wealsopresentanewtechniqueforlocalizationinpartialmap.Thislocalizationmethodcanlocalizerobotsthatareinthecertainrobot’spartialmap.Inaddition,itcangivefinehypothesisontheposesofrobotsthatareoutsidethepartialmap,too.Becauseofthis,theefficiencyofmapmergingtaskisenhanceddramatically.KeywordMicrorobots,Bio-inspiredhexapod,Multirobotssystem,Robotslocalization,MultirobotsMarkovlocalization,Multirobotsexplorationandmapping,Environmentmodel,Mapmerging,Multirobotscoordination,Bayesiannetwork,Robotics,Artificialintelligence,Probabilitytheory-6-1961UnimationUnimate1.1.-7-agentagentWebster’sDictionary1)2)3)1DAI–DistributedArtificialIntelligence231940GreyWalterWienerShannon19702080CEBOTSWARMACTRESSGOFERCEBOT-8-ACTRESSGOFER3CEBOT80Y.UnyCao,AlexS.FukunagaAndrewB.Kahng1997controlagent-9-leaderL.E.Parkertaskcoverage1)2)3)-10-McFarlandeusocialbehaviorcooperativebehaviorReinforcementLearning-11-Fujimura1991PlanninginDynamicEnvironment1.2.lllllll1.2.1.-12-1.2.2.1.2.3.1.2.4.-13-1.2.5.1.2.6.1.2.7.-14-20CEBOTACTRESSSWARMGOFERALLIANCE/L-ALLIANCECEBOTCEBOTCellularRoboticsSystemCEBOTCellCEBOTMasterCellsCEBOTMARS–MicroAutonomousRoboticSystem20mmACTRESSACTRESSActor-basedRobotandEquipmentsSyntheticSystem-15-ContractNetSWARMSWARMSWARMCEBOTSwarmIntelligenceSWARMSWARMcellSWARMGOFERGOFERAIGOFERCTPSCTPSAIGOFERALLIANCE/L-ALLIANCEALLIANCEL.E.ParkerL-ALLIANCEALLIANCE-16-ALLIANCE/L-ALLIANCE1.3.1101011010l10ll-17-40010-18-1.4.1.3-19-1.5.MarkovBayesian-20-2.1.2.2.2-1Illinois-21-2-2MIT2.3.2.3.1.-22-abcdefa-23-2.3.2.(2-5)2-4-24-3mm2-53mm2.4.2.4.1.ABCABCOyABCA(xA,yA)B(xB,yB)C(xC,yC)zABCd1d2d3AB()AABABAyxxxxyyy+---=BABAABxxyyK--=OAABBAOyyxxK--=′A2-5-25-xyyxxyABBA--=ABOAAA()()()()⎟⎟⎠⎞⎜⎜⎝⎛----22,ABBABABAABABBABAdxxxyyxdyyxyyx2ABdABOA221AAyxd′′+=d2d3{}321,,mindddd=2.4.2.yA(xA,yA)B(xB,yB)xC(xC,yC)A’B’C’d1d2d3o2-7FxFyMoyxmgl-26-MFxFygl⎪⎩⎪⎨⎧+-=-==MlFlFImgFymFxmyxyxqqqcossin&&&&&&⎩⎨⎧==qqsincoslylx⎥⎥⎥⎦⎤⎢⎢⎢⎣⎡+--=⎥⎥⎥⎦⎤⎢⎢⎢⎣⎡⎥⎥⎥⎦⎤⎢⎢⎢⎣⎡--mgmlmlMFFmlmlllIYX22sincos10cos01sincossinqqqqqqqqq&&&&2.5.2.5.1.2-8FN2-9FN-27-2-10M8SP-GX21AllegroA3967SLBH-sREFMAXRVI8/=-28-2.5.2.PIC16F87A3967SLBPIC16F87A3967SLB1a1b2a2b2-12-29-2.6.VisualC++VisualC++AppWizard2-132-132.7.11802-30--31-3.1.3.1.1.l3-1-32-lLandmark3-1210,,ZZZ10,ZZ1j21,ZZ2j)(rXArtificialLandmarkllsensorfusionKalmanMarkovMonteCarloParticleFilter-33-3.1.2.VSnn3.2.RobocupHISRGBHIS-34-KalmanMarkov3-1KalmanMarkovDataAssociation3-13-1KalmanMarkovMarkov1)CCD2)3)CCD4)5)-35-3.3.Markov3.3.1.MarkovMarkovMarkovHypothesisProbabilityDistributionMarkovMarkov3-23-2MarkovMarkovUniformDistribution3-2Markov3-2Belief0-36-Markov3-23.3.2.Markov)',,(qyxl=xyqtltTLd},,,{10TddddL=TdTsaaMarkovdTL),,,|()|(10TttdddlLPdlLPL===(1)MarkovtltlLddPdddlLddPttttttt∀===++++)|,...,(),...,,|,...,(211021(2)-37-Markov)(lLBelT=),,,|()(10TttdddlLPlLBelL===(3)TTSd=T(3)),|(),,|(),,|(),,,|()(10101010---======TTTTTTTTttddSPddlLPlLddSPdddlLPlLBelLLLL(4)Markovtl(4)TS},,,{10TdddLTL(4)),|(),,|()|(),,,|()(101010--======TTTtTTTttddSPddlLPlLSPdddlLPlLBelLLL(5)),,|(10TTdddSPLTL)(),|(10lLBelddlLPTTT===-L)()()(1lLBellSPlLBelTTTT====-a(6)-38-110),,|(-=TTTdddSPLaTTad=∫=====---')',|()|'(),,|(1110dllLdlLPdlLPaddlLPTTTTTTL(7)Markov)',|()',,,|()',|(11101lLalLPlLaddlLPlLdlLPTTTTTTTTT========----L(8)Ta1-TL(7)∫=====---')',|()|'()|(111dllLalLPdlLPdlLPTTTTTT(9)∫=====--')',|()'()(11dllLalLPlLBellLBelTTTTT(10)∑=====--'11)',|()'()(lTTTT