機器人外文文獻

Keywords:forwardkinemaFlexiblemanipulatorrobotshavewideindustrialapplications,withHigh-precisionandhigh-accuracyinrobotoperationsrequirethestudyofrobotkinematics,dynamicsandcontrol[4].Theaimofforwardkinematicsistocomputethepositionandorientationoftherobotendeffectorasafunctionoftheangularpositionofeachjoint[1].Thecontrollerperformancebyconsideringthejoints'motioncollectively.Therefore,theprecisionandaccuracyofsuchinformationisessentialtothecontrollerinordertoincreaseitsperformanceinrealroboticoperationsCommerciallyavailablemotioncontrollersuseasinglesensorforeachjointtoestimatetherobot'sangularposition;themostcommonsensoristheopticalencoder[5-1l],whichprovidesahigh-resolutionfeedbacktothecontroller.However,itonlygivesinformationontheservomotorpositionandanydeformationscausedbyjointmonitored[6,12],decreasingtherobot'saccuracy.Thisproblemismoreevidentinopen-chainrobots.Moreover,theprovidedinformationisrelative,whichmeansthatitisimpossibletoestimatetheinitialpositionoftherobot.Anothersensorthatiswidelyusedintheestimationoftheangularpositionoftherobotjointsisthegyroscope;itprovidesameasurementofangularrateofchange,requiringtheaccumulatedsumovertimetoestimatetheangularposition.Dthequantized,noisysignalcausesaccumulatederrorswhenangularpositionisrequired[13-15].Furthermore,theestimatedangularpositionisrelative,whichdoesnotpermitonetoknowtheoftherobotjoints.Agoodsensorthatprovidesanabsolutemeasurementistheaccelerometeranditcanbeusedtoestimatetherobotangularposition[5,16-20].Nevertheless,thesignalobtainedisnoisyandcontainsmuchinformationthatneedspreprocessingbeforebeingusedrequired:theproblemsofusingasinglesensortoestimatetheangularpositionofthejointsandtheonlineestimationoftheforwardkinematics.Inthisperspective,sensorfusiontechniquesimprovetheaccuracyofthe2whichcomplicatetheonlineestimationoftheforwardkinematics.SomeForexample,in[7]accelerometerandencodersignalsarefusedusingaestimatorforpositionestimation;experimentationisperformedonalinearmotorpositioningsystem,whichrequiresnoforwardkinematicsestimation.Anotherworkthatfusesencoderandaccelerometersignalsispresentedin[16],wheretheforwardkinematicsoftwolinksina6-DOFrobotiscalculated;differentusedforsensorfusion.However,theefficacyoftheproposedalgorithmisdemonstratedoffline.Otherworksattempttofusemorethantwodifferentsensors.In[12]thefusionofencoder,accelerometerandinterferometerthroughaKalmanfilterispresentedtoestimatethepositionofaparallelkinematicmachine,buttheanalysisislimitedtoone-axismovement.In[6]camera,accelerometerandgyroscopesensorsareplanartwo-linkrobottofacilitatetheforwardkinematicsestimation.industrialrobotsisproposed.MultiplePCstoprocessallthedatacollectedfromthesensorsandtocontroltherobotareused.However,forwardkinematicsareestimatedonlyfromtheencodersignals.robots'forwardkinematics.Besides,forwardkinematicsislimitedtoacoupleofjointsduetotheequations'complexity.Therefore,theonlineforwardkinematicsestimationproblemformulti-axisrobotsstillrequiresadditionalefforts.Duetotoffilteringandfusingtheinformationofseveralsensorswouldbedesirable.Also,multi-axisforwardkinematicsestimationinanonlinefashionwouldbeadvantageous.ofthesensingmethodofconventionalmotioncontrollersthroughproposalofanencoder-accelerometer-basedfusedsmartsensornetwork.Furthermore,weproposeasmartprocessorcapableofprocessingallthesensedencoder-accelerometersignalssoastoobtainonlineforkinematicsestimationofeachjointofasix-degree-of-freedom(6-DOF)industrialrobot.Thesmartprocessorisdesignedusingfield-programmablegatearrays(FPGA)owingtotheirparallelcomputationcapabilitiesandreconfigurability.Itisdesignedthroughthecombinationofseveralmethodsandtechniquestoachieveonlineoperation.Thesmartprocessorisabletofilterandtofusetheinformationofthesensornetwork,whichcontainstwoprimarysensors:anopticalencoder,3anda3-axisaccelerometer;andthentkinematicsforeachjointinanonlinefashion.Thesensornetworkiscomposedofsixencodersandfour3-axisaccelerometersmountedontheUniversalManipulationArm)robot,demonstratingtheperformanceoftheproposedfusedsmartsensornetworkthroughthemonitoringofthreeexperimentswerecarriedoutwherebytheforwardkinematicsobtainedwiththeproposedmethodologyiscomparedagainsttheobtainedthroughtheaccuracyisfoundwhentheproposedmethodologyisused.2.MethodologyTheuseofaccelerometerson6-DOFPUMArobotsrequiresplacingthemadequatelyinspecificpositions.Thecombinationofaccelerometersandencodersmakeupthesensornetworkthatneedstobeprocessedinordertoestimatetheangularpositionofeachjointandtheforwardkinematicsaccurately.Inthissection,theplacementofthesensornetworkonthePUMArobotispresented.Then,theFPGA-basedforwardkinematicssmartprocessorisclearlydescribed.2.1.SensorNetworkAsensornetworkisanarrayofdiversetypesofsensorstomonitorseveralvariables[25,26];inthiscasetheangularpositionofthejointflexibilitiesoftherobot.ThesensornetworkarrangedontherobotispresentedinFigure1.Figure1(a)depictsthepositionoftheaccelerometersontherobot.xiA,yiAandziAarethemeasurementsofeachaxisfromtheaccelerometeriA.Figure1(b)isaschematicoftherobotincludingitslinkparameters.irepresentstheangularpositionofjointi.iaandidrepresentstherobotphysicaldimensions.Also,thelocalizationofencodersiEisshown.Figure1.Sensornetworklocationonthero(a)Accelerometerplacement.(b)Schematicshowingtheparametersthatdescribetherobot.4Forthiswork,theforwardkinematicsestimationconsistsontheestimationofthejointangularposition(i),thespatialpositionofeachjoint(iiiZYX,,);andtheroll(i),pitch(i),andyaw(i)angles[1].Suchanglesrepresenttherotationalongthrespectively,requiredtoobtaintheorientationofeachjoint.2.1