PDF-湍流烟羽环境下多机器人主动嗅觉实现方法研究

3410Vol.34,No.10200810ACTAAUTOMATICASINICAOctober,200811211(/)..;.,.,/.,,,.,,,,,,/TP24StudyonRealizationMethodofMulti-robotActiveOlfactioninTurbulentPlumeEnvironmentsMENGQing-Hao1LIFei1ZHANGMing-Lu2ZENGMing1WEIXiao-Bo1AbstractThispaperpresentsacooperativesearchstrategybasedonmultiplemobilerobotstorealizetheactiveolfaction(alsocalledodour/gassourcelocalizationorchemicalplumetracing).Antcolonyalgorithm(ACA)combinedwithupwindsearchisusedtocoordinateswarmrobots0motion.TheACAcane®ectivelymakemostrobotsmovetowardthehigh-pheromoneareaandguaranteethedistancesamongrobotsnottoofar.Theupwindsearchcandecreasetheprobabilityofrobotstrappinginlocaloptimizationtooearly.Tojudgethecorrectmotiondirection,thehistorypheromoneisalsoconsideredintheACA.Themethodofcombininghigh-concentrationmaintenancecriterionand°uidmass°uxdivergencecalculationisputforwardtorecognizetheodour/gassource.Computersimulationsshowthattheproposedsearchstrategycanworkinturbulentenvironments.Usingthisstrategy,therobotscanescapefrombothlocalhighconcentrationandeddyareas.Inaddition,theodour/gassourcecanbecorrectlydeclared¯nally.KeywordsRobot,activeolfaction,plume,turbulence,plume¯nding,plumetracing,odour/gassourcedeclaration1/[1].,\,/.,/,[2]../,/2007-06-012007-10-10ReceivedJune1,2007;inrevisedformOctober10,2007(60475028),(863)(2006AA04Z221,2007AA04Z219)SupportedbyNationalNaturalScienceFoundationofChina(60475028),NationalHighTechnologyResearchandDevel-opmentProgramofChina(863Program)(2006AA04Z221,2007AA04Z219)1.3000722.3001301.SchoolofElectricalEngineeringandAutomation,TianjinUniversity,Tianjin3000722.SchoolofMechanicalEngineer-ing,HebeiUniversityofTechnology,Tianjin300130DOI:10.3724/SP.J.1004.2008.01281,\./[3¡4][5¡6].,,.,/.,()[7¡8],(),,.,,.,,,.128234Ishida[9]Russell[10]../.,./,.,,./,250.119cm2/s,20.7cm/h[11],/./,.,.,/\(Filaments),,[12¡13].Farrell[14],Farrell[15](HiddenMarkovmodel,HMM),,.Nielsen[16]Farrell[14]..Russell[17],/./,,;,/;/,/./,.,.,Z(Zigzag)[9][18].,(Chemotaxis)[19](Anemotaxis)[9].,/.Russell[19](Reactive),.,(20cm/s)/.,/,.,([20])/,()()/.,./,.,().,.Hayes[18],.,/.,,/.(Particleswarmoptimization,PSO)Jatmiko[21]Marques[22].[21]Farrell[14],[22]Nielsen[16].Cui[23].,../.,Fluent.,.../(),.2,.Flu-10:1283entCFD(Computational°uiddynamics).Fluent,,.Fluent.,,[24]@½@t+@(½u)@x+@(½v)@y+@(½w)@z=0(1),½,t,uvwxyz.,,(Navier-Stokes)[24]@(½u)@t+@(½uu)@x+@(½uv)@y+@(½uw)@z=@@x(¹@u@x)+@@y(¹@u@y)+@@z(¹@u@z)¡@p@x+Su(2)@(½v)@t+@(½vu)@x+@(½vv)@y+@(½vw)@z=@@x(¹@v@x)+@@y(¹@v@y)+@@z(¹@v@z)¡@p@y+Sv(3)@(½w)@t+@(½wu)@x+@(½wv)@y+@(½ww)@z=@@x(¹@w@x)+@@y(¹@w@y)+@@z(¹@w@z)¡@p@z+Sw(4),¹(Dynamicviscosity),Su,Sv,Swp./,.1(a)1(b)()10m10m().Window1Window2,Door,./(8,8).,.,,.,mol/L.t,,.1(a)1.6m/s,x37.4o;1(b)1m/s,x90o.(a)(a)Instantaneousdistributionofturbulentplumeinthe¯eldwithoutobstacles(b)(b)Instantaneousdistributionofturbulentplumeinthe¯eldwithobstacles1Fig.1Theinstantaneousdistributionofturbulentplume:1)1,/,;2),/,.3,/.,.,128434,,.,..,,,.,./,/.,,/./,./,/,/./.3.1M£N.(x,y),n,c(x;y)=(Pni=1ci)=n,ci.,s.ik(5).8:pkij(t)=[¿ij(t)]®[´ij(t)]¯Ps2max(t)[¿is(t)]®[´is(t)]¯;j2max(t);½i(t)½avg(t)MOVEUPWIND;(5),pkijikj,[25];¿ij=ecj(t)=C,´ij=e¡dij=D;cj(t)tj;dijij;CD®¯;¿ij(t)´ij(t);®¯;½i(t)ti;½avg(t)t5;max(t)t.(5),,,;,,,,,,.,..tt+¢t(¢t),q/,cq=(cq(t+¢t)+cq(t))=2(6).T,,c(t+T)Ã(1¡±)c(t);±2(0;1](7),.3.2/(),///.../:1),;2)/,./.1),Max-time.,,Maxtime,.,.,/,.2).[26]¡@½@t=r¢(½V)(8)10:1285,½/,V,½V.,,r¢(½V)0;r¢(½V)0;r¢(½V)=0,.[27]ZWr¢(½V)dW=IS(½V)¢dS(9)W,S,/.2-D,,.,,R(2(a)).,(),2(b).2(c)n,V,(a)R(a)TherobotsmovetothecircleofradiusR(b)(b)Therobotsmovealongtheedgesofthepolygon(c)(c)Thesinglerobotcalculates°uxdivergence2Fig.2Theprocessofcalculating°uxbyrobots(½V)¢dS=(½V)¢ndS=(½V)¢n(Vr¢dt)(10),Vr,dt./,/,/;,/.4:1),;2);3);4);5)/;6)Ad-Hoc,10.5),.5),..1286344.13»5,5(±¤),0.7m/s.,t=3,,,.,(5),.t=15,15,/.4,t=22,/,.5,t=33,.5,R,..3Fig.3Thephasesofplume¯ndingandplumetracing4Fig.4Thephaseofmaintenancecriterion5Fig.5Thephaseofodour/gassourcedeclaration10:12874.26»8.,.,90,.6,.,,,.,,,,,7.8.4.3.,,,.5,6Fig.6Thephasesofplume¯ndingandplumetracing7Fig.7Thephaseofmaintenancecriterion8/Fig.8Thephaseofodour/gassourcedeclaration128834,9(55).9(a),9(b),.,.9(c),/.(a)(a)Therobotsenterintothephaseofplumetracing(b)(b)Therobotsgetintothelocaloptimum(c)(c)Therobotscannotdeclarethesourceaccurately95Fig.9Thesearchresultbasedon¯verobotsformation,,.,,/,/.4.41),.,,.,,,.10.,.5,,.5,.7,.10Fig.10Thein°uencebythenumberofrobots2)/,.,/,,,.1158,.128.8(0,0)11Fig.11Thein°uencebythedi®erentstartingpositions10:1289,,.600,,.,,.12Fig.12Eightgridnumbersandthecorrespondingstartingpositions3),,,..,.,/,,.,10,30.4),(5),®,¯.,,.,,,,.5/:,,,.,.,(/).,.,,,.,/.,.,;,.References1MengQing-Hao,LiFei.Reviewofactiveolfaction.Robot,2006,28(1):89¡96(,..,2006,28(1):89¡96)2MurlisJ.MechanismsinInsectOlfaction.Oxford:OxfordUniversityPress,1986.27¡383IshidaH,HayashiK,TakakusakiM,NakamotoT,Mori-izumiT,KanzakiR.Odour-sourcelocalizationsystemmim-ickingbehaviourofsilkwormmoth.SensorsandActuatorsA:Physical,1996,51(2):225¡2304LytridisC,KadarEE,VirkGS.Asystematicapproachtotheproblemofodoursourcelocaliz