机械手外文翻译_3.docx
机械手外文翻译Proceedingsofthe33rdChineseControlConferenceJuly28-30,2021,Nanjing,ChinaTheRemoteControlSystemoftheManipulatorSUNHua,ZHANGYan,XUEJingjing,WUZongkaiCollegeofAutomation,HarbinEngineeringUniversity,Harbin15000E-mail:sunhuasAbstract:Aremotecontrolsystemofthe5degreeoffreedommanipulatorwasdesigned.Thismanipulatorwasinstalledintoourmobilerobottoconstitutearemoterescuerobot.TheDenavit-Hartenbergmethodwasusedtoestablishthekinematicmodelsandthepathplanningofthemanipulatorwasresearched.TheoperatorcouldremotecontrolthemanipulatorbytheinteractiveinterfaceofPCwhichcoulddisplaymovingpictureandvariousdataofthemanipulator.TheservosofthemanipulatorwerecontrolledbytheslaveFPGAcontroller.Inaddition,theslaveFPGAcontrollercommunicatedwiththePCviathewirelesscommunicationmodule.OwingtotheembeddedNiosIIprogramandIP(IntellectualProperty)coregeneratingPWMwavesinFPGA,thesystemcouldcontrolthemultipleservosfastandflexible.Inordertoachievereal-timeoperationandsimulation,theinteractiveinterfacewasestablishedbythemixedprogrammingofVCandMATLAB.KeyWords:Themanipulator;Remotecontrol;Denavit-Hartenberg;FPGA;Human-computerinteraction1IntroductionWiththedevelopmentofthemicroelectronictechniqueandthecomputertechnology,themanipulatorhasbecomeessentialequipmentinthemanufacturingindustry.Asweallknown,themanipulatorisusuallyappliedtoaccomplishdull,onerousandrepeatedphysicalwork,especiallyusedtosubstitutethemanualoperationunderthedangerousandthehazardousenvironmentsuchasthecorrosionandthehightemperature.Inthispaper,themanipulatorwasinstalledourmobilerobot.Thetele-operationsystemofthismanipulatorwasdesigned.ThewholesystemisonstitutedbyPCandslaveFPGA.TheoperatorcanremotecontrolthemanipulatorbyPC.ThewirelesscommunicationwasusedfortransmittingdatabetweenPCandFPGA.FPGAiscontrollerofthethemanipulatorinthemobilerobot.FPGAhastheabundantinternalresourceandIPcores.AndacentralcontroloptionwasbuiltviaanembeddedNiosIIprogramandanIPcoreinFPGA.Furthermore,VeriloglanguagewasadoptedtodesigntheIPcorewhichgenerateddigitalPWMwavesforcontrollingthemanipulator.Therefore,thissystemcouldreachhigherprecisionandeasytodebug.MATLABsoftwarewasadoptedtobuildthekinematicmodelsofmanipulator.AndusingD-H(theacronymofDenavit-Hartenberg)methodtosolvetheforwardandinversekinematicequationsofthemanipulator,toanalyzethemotivation,toplanandtrackthemotionspath.Inaddition,agoodinterfaceofhuman-computerinteractionwasenhancedintheremotecontrolsystemofthemanipulatorinPC.Moreover,themanipulatorsimulationtechnologywasbuiltbyusingthemixedprogrammingofVCandMATLAB.Thus,themotionchoreographswasgotquicklyandeasily,alsogreatlysavedtimeandcutthecost.2ManipulatorModelandPathPlanningAtfirst,themotionmodelofthemanipulatorwasbuilt.Then,thekinematicsimulationanditspathplanningwereresearched.Theseworksprovidedthefoundationforthedesignoftheremotecontrolsystemofthemanipulator.2.1MotionModeloftheManipulatorThemanipulatorwasregardedasanopenloopkinematicchain.Itwasconstitutedbyfiverotaryjoints.Anditsoneendwasfixedonabasewhiletheotherendwasusedtoachievetheabilityofgrabbing.Therefore,itisbettertoestablishachaincoordinateframeasshowninFig.1.Theterminalpositionandattitudewasdeterminedviausingforwardkinematicequationafterknowingtherotatingangleofeveryjoint.TheD-HparametertableshownasTable1wasestablishedbyusingtheframesinFig.1.Fig.1CoordinateframesofmechanicalarmTable1D-HParametersoftheRobotArmDuetoD-Hmethod:T=An+1n+1n=Cn+1?Sn+1Sn+1CanCn+1Can0an?San?Sandn+1Sn+1SanCn+1San00CanCandn+101WhereCn+1=cosn+1,Sn+1=sinn+1,Can=cosan,San=sinan.Thetransformationmatrixofeveryjointwasgivenbyequation(2).T10=cos1sin1sin1cos1000000001001T21=cos2?sin200001d1?sin2?cos2000001T32=cos3?sin3sin3cos3000000001d201T43=cos4?sin40000?1?d3sin4cos4000001T54=cos5?sin5sin5cos5000000001d401T50=nxnxnynynxnxnynynznz00nznz01=T10T2?1T3?2T4?3T5?4(2)Whereunitvectorn,o,ainequation(2)wasn=normal,n=orientation,n=approac?,n=position.Parametersofmechanicalarmweregivenbyd1=85mm,d2=116mm,d3=85mm,d4=95mm.Thereforetheforwardkinematicequationwasdeterminedbytakingeveryparameterinequation(3).P50=180C1S2+3+116C1S2180S1S2+3+116C1S285+116C2+180C2+3(3)Inpracticalapplication,themanipulatorwasadoptedtograbobjects.Thisrequiredthatthefixedpositionwasgivenfromterminaltotargetlocation.Thatwastheinversekinematicanalysisofmanipulator.Inversetransformationwasusedtodetermineangleofeveryrotaryjointtowardtheestablishedcoordinates.Andtheusedmethodofinversetransformationwasthecommonmethodtosolvesuchproblem(thismethodalsoknownasalgebraicmethod).UsinginversetransformationTnn?1?1separatelytotheleftmultiplicationwithT=50T10T2?1T3?2T4?3T5?4,theangleofeveryrotaryjoint12345wasdetermined.Owingtotheseresults,therotaryangles123atterminalpositionofmanipulatorweretotallydecidedbythetargetpositionPxPyPz.Angle4wasusedtochangeterminalattitudeofthemanipulatoranditwaschangedbytheknownnormalvector.However,angle5,wasdecidedbythesizeoftargetobject.2.2MotionSimulationoftheManipulatorThemanipulatormodelwasbuiltandsimulatedviaMATLABtoolbox.Wecouldverifytherationalityofthemathematicalmodel.WhiletheMATLABmodelwasestablishedbytable1andshownasFig.2Fig.2MATLABsimulationofthemanipulatorComparingtotheFig.1andFig.2,thesimulationmodelofthemanipulatorwascoincidedtothereferenceframemodel.Thatwastosay,thegivencoordinateframewascorrect.TheseresultsalsocouldbeprovedbythedeterminedinversekinematicequationsviaMATLABshowninthetable(2)andtable(3).Thetargetpositionwassolvedbyforwardkinematics.Afterthat,therotaryangleswerecalculatedbyinversekinematicalequation.Itturnedoutthattheserotaryanglescoincidedtothegivenangles.Therefore,theseresultsverifiedthecorrectnessofforwardandinversekinematicalequation.Table(2)ForwardKinematicsAnalyzeTable(3)InverseKinematicsAnalyze3PathPlanningoftheManipulatorThetotaldisplacementofjointwascalculatedbyinversekinematicalequationwhenthemanipulatormovedtonewposition.Thus,themanipulatorcouldmovetonewposition.Althoughthemanipulatorfinallymovedtotheexpectedpositioninsuchcondition,themotionofthemanipulatorbetweenthesetwopointswasunknown.Duetospacelimitations,motionandsomecertainpositionrequirements,themanipulatorwasoftenunabletomoveastheabovementionedmethod.Therefore,themotionpathwasdesignedtocoincidewiththelimitedconditions.Inthispaper,wecouldusethesecertainlimitationstodecidesomeexpectedpoints.Andtheseexpectedpointswereusedtomatchtheplanningpathofthemanipulatorsmovement.Owingtotheplanningpath,coordinateineverypartcouldbecalculated.Therotaryangleofeveryjointwascalculatedviainversekineticalequationandtheseanglesrealizedthemovementofplanningpath.MovementofthemanipulatorwasshowninFig.3(Where?representedthepointswouldbepassedbythemanipulator;*representedtheexpectedpointsofeverysegment;-representedpathplanningofthemanipulator).InFig.3,wecouldseethatthemotionofthemanipulatorpassedeveryplanningpointandthemovementpathcoincidedtotheplanningpath.Fig.3Thepathplanningsimulationofthemanipulator4RemoteControlSystemoftheManipulatorTheremotecontrolsystemofthemanipulatorcontainsthemainPCandtheslaveFPGAcontrollerusingDE2BoardofALTERCompany.ThemotorsofthemanipulatorwerecontrolledbymultipathPWMwaves.AndthePWMwavesweregeneratedbyIPcore.TheFPGAcontrollerCommunicatedwithPCviawirelessserialport.WhileinthePCinteraction,theoperatorcouldobservethemoveofthemanipulatorinreal-timeandtele-controlthemotionofthemanipulator.Alsoeverymovementofmanipulatorcouldbeobservedinadvanceviathesimulationtechnique.ThegeneraldesignofthemanipulatorremotecontrolsystemwasshowninFig.4.Fig.4Theblockdiagramoftheremotecontrolsystem4.1ControlModeoftheManipulatorThereweretwocontrolmodesofthemanipulator.OnemodeisthattheinversekinematicalequationsarecalculatedbyFPGAstraightlytodetermineangleofeveryrotaryjoint.Thus,thecontrolofthemanipulatorwasachieved.Theadvantageofthismodeismoredirectandindependenttofinishthecontrolofthemanipulatorwithouttheexternaldevices.Atthesametime,thismodehaslargequantitiesofcalculations,whichoccupymoreinternalstorageandrunningtimeofFPGA.ResourcesofFPGAarewastedunderthismode.TheothermodeaccomplishedthecontrolofthemanipulatorbyusingVCandMATLABinPC.UsingVCandMATLABfinishedalargenumberofcomplexcalculationsanddeterminedangleofeveryrotaryjoint.AndtheangleresultsweretransmittedtoFPGAinordertoaccomplishthecontrolofthemanipulator.Thismannersavedlotsofinternalstorageandrunningtime.Inaddition,FPGAcouldfinishotherworksunderthismode.Butthemanipulatorwasnotunderfastcontrolinthismode.Inthissystem,anewmodewasadoptedinthemanipulatorremotecontrolsystemdependingontheadvantagesofthetwomodes.Specifically,whenthemanipulatoraccomplishedthespecifiedandrepeatedmovementtheformermodewasadoptedunderdirectcontrolbyFPGA.WhenthemanipulatorwantedtoachievenewmotionsthelattermodewasusedtobecommandedbyordersfromPC.Thisnewmodewasmadegooduseofadvantagesofthetwomodesintheabove.Andthisnewmodelightenedcomputationalburdenandimprovedworkingefficiencyofthemanipulator.4.2SOPCDesignfortheRemoteControlSystemMovementofthemanipulatorwascontrolledbyservos.AndtheservoswerecontrolledbyPWMwaveswiththecycleof20ms.PulsewidthofthesePWMwaveswas0.52.5mscorrespondingtotherotaryangleofservowith-90degreeto90degree.HighprecisionofPWMwavesweregeneratedbyIPcoreviaVeriloginthissystem.TheresultswereshowninFig.5.PWMwavescontrolledrotaryanglesoftheservosviatheservodrivers.Fig.5ThePWMIPcoreMultipleofIPcoreswereabletobedownloadedintoFPGA.AndmultiplePWMwaveswithhighprecisionweregeneratedintheoutput.AsshowninFig.6,thepulsewidthofthesewavescouldbesettledbyprogramofNiosII.Themovementofthemanipulatorwasmoreflexibleandinhigherprecisioninthissystem.Fig.6TheIPcoresgeneratingPWMwaveThemovementofthemanipulatorwasaccomplishedbythedutyratioofPWMwaves.Formula(4)invertedrotaryanglentothecorrespondingamountofthedutyratioofPWMwaves.ThedutyratioofPWMwavescorrespondedtotheNiosIIoutput.PWMn=1000000?n?5000090+750004Wirelessserialof9600baudratewasusedtotransmitthecoordinateandtheangleinformationfromhostcomputertoFPGA.Afterthat,thedataandorderswereanalyzedbyFPGAThenFPGAtransmittedthemovementresultstointeractiveinterfaceofhostcomputerviawirelesstransitionmodel.ThiscommunicationwasrealizedthroughaddingUVRTcommunicationprotocoltoFPGA.4.3TheInteractiveInterfaceoftheRemoteControlSystemTheinteractiveinterfaceoftheremotecontrolsystemwasshowninFig.7.Thereweresomefunctionsintheinteractiveinterface:videoobservation,themanipulatorcontrolandthesimulationmodeling.Atfirst,themanipulatorvideocouldbeseenfromcameratointeractiveinterface.Theoperatorcouldmonitorthemanipulatorinreal-time.Secondly,theangleandthecoordinatecouldbesetincontrolzoneoftheinteractiveinterface.Theangleofthemanipulatorcouldbesetindependentlytoeachsinglejoint.Inaddition,theanglesettingcouldbeshowninreal-timeinthelistofinteractiveinterface(asshowninFig.7).Inthesetofcoordinates,judgingofcoordinatesettingassuredthatthetotalcoordinatescouldachievetothetargetpoints.Thusthemanipulatorcouldbecontrolledtomoveinthesettledpathdependontheangleinformation.Lastly,theMATLABrobottoolboxwasembeddedintothisinteractiveinterface.Oneinterfacewasintegratedboththecontrolandsimulationofthemanipulator.MATLABrobottoolboxwasdirectlyusedbyinteractiveinterfaceinthemanipulatormodeling.Eachgroupofinformationwassimulatedseparatelyinordertodetectwhethereachmovementwascorrect.Andthegeneralsimulationcouldtestwhethermovementarrangementofthemanipulatorwasreasonable.Combiningwithmultiplesimulationmethodsmadethemovementarrangementmoreflexible,theoperationofthemanipulatorsimplerandinterfaceinteractionmoreperfect.Fig.7Theinteractiveinterfaceofthemanipulator5ExperimentandSimulationInordertoverifypropertiesoftheremotecontrolsystemofthemanipulator,experimentsofthesystemwereunderwayandwerecomparingtothesimulationsystem.Tobespecific,manipulatormodelingwasbuiltbyinteractiveinterfaceandagroupofcoordinatescouldbedesigned.ThesecoordinatesweretransmittedtoFPGA,whichcontrolledtheservostoaccomplishthemovementofthemanipulator.Jointangles,theterminalcoordinatesshotbyinterfacevideo.Thesimulationresultswereshownin