基于增强现实ar装配系统研究

AcomprehensivesurveyofaugmentedrealityassemblyresearchAbstractprovideaconciseoverviewofthetechnicalfeatures,characteristicsandbroadrangeofapplicationsofARbasedassemblysystemspublishedbetween1990and2015.Introduction1.traditionalcomputer-aidedassemblytechnology：requirtoidentifyconstraintinformationbetweenthematingparts；difficulttoforeseetheimpactofindividualmatingspecificationsonotheraspectsoftheassemblyprocess；lacksdirectinteractionbetweenusersandthecomponents.2.VRassemblysystem：lackingofrealisticexperience，decreasingthefidelityandcredibility；noteasytofullyandaccuratelymodelthereal-timeandactualworkingenvironments.3.ARassemblysystem：doesnotneedtheentirerealworldtobemodeled；enhancestheinteractionbetweenthesystemsandtheusers，manipulatetheobjectsnaturally.SixfunctionalmodulesofARassemblysystemThreemaincategoriesARassemblysystem120articles，39.47%116articles，38.16%68articles，22.37%ARmulti-mediaandinteractiveinstructionsDisadvantage：focussolelyonprovidingstep-basedinstructions，overlookinguser’sneedoftimelyguidanceinassemblyoperations；couldnotkeeptrackoftheassemblystatusinreal-timeandrecognizeerrorsandcompletionstates；whenassistingcomplexassemblyprocesses，time-consumingauthoringprocedures,integrationwithenterprisedata；theinteractionschememaydisturborinterruptuser’son-goingassemblytask.Target:detectingandrecognizingtheusers’actionsinordertoprovideanindustrialrobusthandsfreeinteraction.examinetheappropriatenessofARguidanceformorecomplex,multi-stepassemblytasks.Context-awarenessandauthoringARassemblysystemkeepstrackofthestatusofusersinreal-time,automaticallyrecognizingmanualerrorsandcompletionateachassemblystepanddisplayingmultimediainstructionscorrespondingtotherecognizedstates.ARassemblyauthoringsystemshavebeenenabledtocopewithmanualassemblyoperationsefficiently;Disadvantage：time-consumingtobuild‘‘adhoc’’systems.needmanualinput.Target:tobemoretask-adaptive,suiteforawiderangeofdifferenttasks.considertheimplementationissuesofsensors.becomesmartenoughtounderstandauser’sintentandgeneratetimelycontext-awareguidanceduringassemblyoperationswithoutmanualinput.EvaluationofARassemblyguidance1.Effectivenessevaluation:concernsthecapabilityofthesystemtoproduceadesiredresultforacertaintaskoractivity.2.Usabilityevaluation:involvesinvestigatingtheeaseofuseandlearnabilityoftheARassemblysystembasedonneedsanalysis,userinterviewsandexpertevaluations.Disadvantage：differentfactorsthatinfluencetheeffectivenessandusabilityofanARassemblysystemarestillobscure.suchasvariousinteractionmethods.designedonlyforaspecific‘‘adhoc’’assemblytask.Trend：evaluatingnovelARinterfaces,especiallyhowtheseinterfacescanimprovehumanresourcecapabilitiesandtheattendanthumanfactors.ARassemblytrainingResearchinthisareahaslargelyinvolvedproceduraltaskswhereauserfollowsvisualorothermulti-modalcuestoperformaseriesofsteps,withthefocusonmaximizingtheuser’sefficiencywhileusingtheARsystemDisadvantage：provideminimalfeedbackwhentheusersmakeamistake,whichisnotconducivetolearning.Target:theadaptivityofvariousmethodsanddifficultylevelsofassemblyoperationswouldbeexpectedtobeintegratedtotheARassemblytrainingsystem.handlecurrenttrendsofagileproductdevelopmentandmanufacturing.ARassemblyprocesssimulationandplanningmanyofthecurrentARassemblysimulationsystemsseemtobelimitedtotheinteractionbetweenvirtualobjects.ignoreanimportantissueofincorporatingrealobjectsintotheARenvironmenttoprovidethecorebenefitsofAR.suchashapticfeedback,naturalinteractionandbettermanipulationintuition.Hand-basedHCIhasbecomeapopularresearchtopicinARassemblysimulation.Disadvantage：theuserneedtoundertakespecificsimpleactionstoimplementtheinteractionwiththeARenvironment.theactionlackintuitiveness.onlyalimitedtypesofgestureshavebeenusedandthegesturesaredesignedbyresearchersforoptimalrecognitionratherthanfornaturalness.themappingsbetweengesturesandcommandstothecomputerareoftenarbitrary.assemblydesignandplanningTheimplementationofassemblydesignandplanningintheARenvironmentispromisingandagrowingnumberofresearchershavestudiedthisarea.Disadvantage：accuracyissue,intuitivenessanddependencyonCADsoftware.Target:toconstructanintegratedARassemblyplatformtomodeltheinteractionbetweenmatingcomponentsaccurately.toproposeaninformationmanagementmethodtoextractsinformationfromCADsystem,storesandreasonstheassemblyrelationsindependently.Marker-basedtrackingTheuseofmarkersisverycommoninARassemblybecauseitincreasestherobustnessandimprovesthecomputationefficiencyduetothestableandfastdetectionofmarkers.Disadvantage：sufferfrommarkerocclusion.industrialcomponentsaretoosmalltoattachmarkers.Sensor-basedtrackingThetrackingprincipleisbasedondifferenttypesofsensors.Visionbasedtrackingreliesoncomputervisiontechniques.Marker-lesstrackingbedeemedtobeadesirablesuccessorofmarkersrecently.Disadvantage:thetrade-offsbetweenthetrackingaccuracyandcomputationcostneedtobeconsideredcarefully.CollaborativeARinterfaceatechnologyaimmingatassemblydesigners,plannersaswellasoperatorscanparticipatetogethertofacilitatecommunicationbetweenthemduringassemblyprocess.Rana