ORIGINALPAPER
Benchmarkingofvirtualrealityperformanceinmechanicseducation
MauraMengoni·MicheleGermani·MargheritaPeruzzini
Received:27April2011/Accepted:29April2011/Publishedonline:27May2011©Springer-Verlag2011
AbstractThepaperexploresthepotentialitiesofvirtualreality(VR)toimprovethelearningprocessofmechanicalproductdesign.ItisfocusedonthedefinitionofaproperexperimentalVR-basedset-upwhoseperformancematchesmechanicaldesignlearningpurposes,suchasassemblabilityandtolerancesprescription.Themethodconsistsoftwomainactivities:VRtechnologiesbenchmarkingbasedonsensoryfeedbackandevaluationofhowVRtoolsimpactonlearningcurves.Inordertoquantifytheperformanceofthetechnol-ogy,anexperimentalprotocolisdefinedandantestingplanisset.Evaluationparametersaredividedintoperformanceandusabilitymetricstodistinguishbetweenthecognitiveandtechnicalaspectsofthelearningprocess.Theexperi-mentalVR-basedsetupistestedonstudentsinmechanicalengineeringthroughtheapplicationoftheprotocol.KeywordsMechanicalproductdesign·Virtualreality·Experimentalprotocol·Learningcurve·Mechanicseducation
1Introduction
Modernsocietyisdominatedbycontinuousscientificandtechnicaldevelopments.Specializationhasbecomeoneofthemostimportantenablersforindustrialimprovement.Asaresult,nowadayseducationismoreandmorejob-orientedandtechnicaleducationisassuminggreaterimportance.Inthiscontextbothuniversityandindustryarecollaboratingtocreatehighprofessionalcompetencies.Thefirstdisseminates
M.Mengoni(B)·M.Germani·M.PeruzziniDepartmentofMechanicalEngineering,PolytechnicUniversityofMarche,
ViaBrecceBianche,60131Ancona,Italye-mail:m.mengoni@univpm.it
knowledgeandinnovativemethodswhilethesecondpro-videsapracticalbackgroundforgeneralprinciplestraining.Themainproblemdealswiththeeffortandtimerequiredtoimprovetechnicallearning,whilemarketcompetitivenessforcescompaniestodemandyoungandhigh-qualifiedengi-neersinshorttime.Therefore,theentireeducationalprocessneedstobefastandefficient.Novelinformationtechnolo-gies(IT)andemergingvirtualreality(VR)systemsprovideapossibleanswertotheabove-mentionedquestions.Someofthemostimportantissues,inmechanicaldesignfield,aretheinvestigationofsuchtechnologiespotentialitiesandtheevaluationofachievablebenefitsintermsofproductdesignlearningeffectivenessandquality.WhileIThasbeendeeplyexploredindistanceeducation,i.e.e-learning,VRstillrep-resentsanovelty.
VRreferstoanimmersiveenvironmentthatallowspow-erfulvisualizationanddirectmanipulationofvirtualobjects.Itiswidelyusedforseveralengineeringapplicationsasitprovidesnovelhumancomputerinterfacestointeractwithdigitalmock-ups.Thecloseconnectionbetweenindustryandeducationrepresentsthestartingpointofthisresearch.Insteadoftraditionalteachingmethods,virtualtechnolo-giescansimultaneouslystimulatethesensesofvisionbyprovidingstereoscopicimagingviewsandcomplexspatialeffects,oftouch,hearingandmotionbyrespectivelyadopt-inghaptic,soundandmotiondevices.Thesecanimprovethelearningprocessinrespectwithtraditionalteachingmeth-odsandtools.Theobservationofstudentsinterpretingtwo-dimensionaldrawingshighlightedseveraldifficulties:theimpactevaluationofgeometricanddimensionaltoleranceschains,thedetectionoffunctionalandassemblyerrors,therecognitionofrightdesignsolutionsandthechoiceofthepropermanufacturingoperations.Theselimitationsforcetutorstoseekforinnovativetechnologiesabletoimprovestudents’perception.
123
104InvestigationsintotheuseofVRhaveindicatedthatitmayimprovethelearningprocessofferingamoreusefulproduct’srepresentationandcreatinganaugmentedenvironmentforthemodelsinvestigationanddescription.ThemainproblemdealswiththeeffectiveapplicationofaVRsystemintoedu-cationalsituationsandtheappraisalofitsimpactonlearning.ThesearecrucialpointsifweconsiderthatthedefinitionofaproperVRarrangementforspecificlessonpurposeshastobecorrelatedtothelearningprocessastheresultofindi-vidualskills(proceduralaspects)andinstrumentalpractice(declarativeaspects).
ThescopeofthisresearchistheexperimentationofVRinmechanicalproductdesignteaching,theassessmentofthelimitsandadvantagesofavailabletechnologiesand,finally,theevaluationoftheachievedlearningprocessperformance.Inthiscontext,thepaperaimsatdefiningbothaproperVRset-upformechanicalproductdesignteachingandanexper-imentalprotocolforvalidationtests.AmethodisproposedtobenchmarkcurrentVRtechnologies.Itisbasedonthestudyofproductdesignlessonsbytraditionalmeansofrep-resentation,ontheidentificationofthemaincriticalactivitieswherepaper-basedtoolsandCADsystemsusuallyfailandonthecorrelationbetweentheidentifiedactivitiesandtoolsusability,achievedpresenceanddepthofsensationsthatareperceivedintotheexperienced-basedlearningenvironment.Thenanexperimentalprotocolbasedonspecificcontextmet-ricsisdefined.ItaimsatevaluatingboththelearningprocessperformanceforthespecificpurposeandtheusabilityoftheadoptedVR-basedset-up.
2Background:VRtechnologiesforlearningpurposes2.1LearningenvironmentsformechanicalproductdesignAlearningenvironmentischaracterizedbyactiveinterac-tionsamongallinvolvedindividuals.Kaye[13]inparticularstatesthatlearningisanindividualprocessbutitisalwaysinfluencedandstimulatedbytheexternalcontext.Onlybyconversationandcomparisonwithpeersandexperts,stu-dentsreachasolidknowledgeofthespecifictopics.Mul-tidisciplinaryenvironmentsareparticularlyappropriateforeducationalpurposes:scientificstudiesshowedthathumanlearningusuallyhappensfor83%bysight,only10%byhearingandtherestbyothersenses[17].Ontheotherhand,collaborativeenvironmentsprovidelearnerswithseveraladvantagessuchastheopportunitytoexperiencethemulti-plestandpointsofotherlearnerswithdifferentbackgroundsandtheabilitytodevelopcriticalthinkingskillsthroughtheprocessofjudgingandvaluing.Inordertoimprovecol-laboration,advanceddigitaltechnologiesmayhelpstudentsforincrementinglearningbysimulatingrealdesignprocessoperationsandbysharingthedesignoutcomes[10].
123
IntJInteractDesManuf(2011)5:103–117
Arecentstudyshowsthatdesigneducationcanbeimplementedbytwomaindifferentapproaches[22]:•face-to-faceeducation,thatimprovesthelearners-learnersandlearners-instructorsinteraction.Themainproblemsinexperience-basedlearningapplicationarerelatedtoretrievinginformation,developingcollaborativeworkandexperiencingdesignsolutionsinrealtime;
•distanceeducation,thatimplieslearnersandinstruc-torsingeographicallyseparatedsites.Threedifferentapproachesandrelatedcommunicationmediaarepro-posed:(a)one-wayinstructionbymail,radioandtelevi-sion,(b)singletechnologyinstructionbycomputer-basedorweb-basedlearning,and(c)blendedlearningthatcom-binesface-to-facewithasynchronousand/orsynchronouscomputertechnology.Thesamestudydemonstratesthatface-to-faceismoresuc-cessfulthandistancelearninginmechanicalproductdesignwheretheunderstandingofdesigntopicsrequiresaconcreteexperimentationofgeneralprinciples.Theconceptofexpe-rienced-basedlearninghasbeenwidelyexploredindesignteaching.Itbasicallyconsistsofthefollowingsteps:con-creteexperienceoftechnicalproblemsandsolutions,obser-vationoftheachievedresultsandformulationofabstractconceptsandgeneralizations[14].Otherresearchesclaimedthatlearningwithoutexecutionofactionremainsatthestateofmentalactionandthereforedistantfromrealaction[21].Thesepreliminaryconsiderationspointouttheimportanceofadoptingeducationalmethodsandtoolsthatmakestudentsexperiencedesigntopicsinaneffectiveway.Asuitablelearn-ingset-upoughttoanswerstudents’needsandtoenablethemtoreacheducationalgoals.
2.2PotentialitiesofVRinmechanicalproductdesign
teachingTraditionalteachingisalmostentirelycentredon2Drep-resentationsthatmakedifficulttheinterpretationofdiffer-entdesignsolutionsanderrorsdetection.NowadaysCADsystemsoffer3Dmodelsvisualizationandanimationthatsupportassemblycomprehensionbyabettervisualcharac-terization.Otherwiseperceptionisstilllimitedonlytosightanditdoesnotsignificantlyincrementthelearningprocess.ItisworthtonoticethatCADsystemsfunctionalitiesarenotabletosupporttheidentificationofawkwardreachanglesorrelevantassembly/disassemblyissues.
Theuseofexperimentallaboratoriesmayovercometheabove-mentionedproblems.Theyallowstudentstoexperi-encemechanicalequipmentsandmanufacturingoperations.Highcostsofmaintenance,greatinitialinvestmentsandtheincreasingnumberofclassroomstudentsmakelaboratoriesnoteasytobekeptup.
IntJInteractDesManuf(2011)5:103–117VR-basedenvironmentsseemtobeavalidsolutiontoimprovemechanicaldesigntopicslearning.Howeverthisintuitiveassumptionneedstobeobjectivelydemonstrated.ResearchesintotheuseofVRhaveindicatedthatitmayoffermoreusefulartefactsrepresentationsandsimulatetherelevantcharacteristicsofaproductsuchasengineering,manufacturingandmaintenanceaspects[6].Inparticular,inassemblyandtolerancesanalysisVRenvironmentscansup-portrealisticinteractionbetweenpartsbyreal-timesimula-tionofphysicalconstraintsandanintuitiveinterface,whichallowsnaturalmanipulation.Somestudieshavesuggestedalsothataddingforcefeedbacktoassembly,virtualanal-ysisincreasestaskefficiencyandperformance[1,25]whileeliminatingphysicalprototypesgivessubstantialcostsavings[23].Finally,virtualenvironmentsmayeasilycreatecollabo-rativespaceswherestudentslearnmultiple-levelinformationabouttheproduct,listentodifferentinterpretationsandsharetheirlearningexperiencetodeveloptheirownpracticalandcognitiveskills.
Mostoftherecentstudiesonface-to-facedesigneducationhighlightthepotentialitiesofVRtechnologiestoimproveperceptionineducationandtrainingapplications[4,20,24].ThemainadvantagesrecognizedtoVRapplicationsare:•theimprovementofthespatialabilityoflearnersastheyallownotonlythevisualizationof3Dmodelsbutalsotheirexperimentation.Furthermorestudentsfindthemmucheasiertounderstandthingsfromdiagramsormod-elssimplylookingatgraphsormathematicalalgorithms;•theknowledgesharingfacilitationandthecollaborationinmultidisciplinaryteamwork;
•theachievementofsenseofpresenceinsteadoftraditionalvisualizationtechnologies;
•theinteractionwithvirtualmodelsinaveryintuitiveandnaturalmanner.Althoughtheabove-mentionedadvantages,noexperimen-talresultshavebeenachievedinselectingthemostsuitabletechnologiesformechanicseducationtasks.ThisismainlyduetothehighcostsofVRtechnologyimplementation,thedifficultytoidentifytheproperVRtechnologiescombina-tionforthespecificlearningpurposesandthecomplexitytounderstandhowitimpactsonthelearningcurve.2.3EvaluationofVRtechnologiesinthelearningprocessTheabilityofamechanicaldesignerisearnedbyexperienceandpracticeandcouldbeconsideredtheoutcomeofhis/herlearningprocess.Overtheyears,severalmodelsweredevel-opedtocapturethehumanperformanceincarryingoutatask.Theyareusuallybasedontwomainlearningcurvesforms:time-basedorperformance-basedrepresentations.Outlininglearningcurvehasbecomeaverypopularmethodthanksto
105
itssimplicityofuseanditsabilitytofitempiricaldatawell[15].
Fromthelearningpointofview,mechanicalproductdesignimpliesasequenceofmentalandpracticalactivi-ties,suchasthecreationofsolidmodels,theirassembly,thecomprehensionoftheirinterrelationsandtheirparticularfunctions.Thetransferofacquiredexpertisedependsonboththeindividualaptitudeforlearningandtraducingactionintoexperience(cognitivelevel)andthesupportandeasetouseoftools(technicallevel).Consideringthecaseunderinves-tigation,thethinkingprocessesallowstudentstoelaboratetherightprocedureforproductdesign,butthemerecourseofactionsdoesnotleadtoasuccessfulresult.Furthermore,tools’usabilityandlearnabilityinfluencethelearningpro-cess.Severalstudiesstressedtheneedtounderstandhownewtechnologiesaffectlearningcurvesinordertoestablishtheappropriatetrainingandassessment[8].Recentworksarenotablyorientedtoconsiderfinallearningasthesumofcog-nitiveandtechnicalcomponents.Hamadeetal.[7]developamethodtoassessthespeedandproficiencyofCADsys-temslearning,basedontestexercisesandonthemeasureoftwoobjectivefactors,performancetimeandfeaturecount.Thestudyinterestliesnotonlyintheobjectivityandporta-bilitytoothercontexts,butalsointhedistinctionbetweenproceduralanddeclarativeknowledge.Totallearningcurveissetupanddecomposedintoitscognitiveandbehaviouralcomponentsthroughadual-phaselearningmodelapproach.Howeverthemethodleavesouttheusabilityofequipmentanditusesonlyobjectivemetricsunderestimatingthesub-jectivecomponents.
DistinctionofcognitiveandtechnicalaspectsiscitedalsobyBlavieretal.[3].Theyachieveamorecompleteevalu-ationandincludetheperceptualandthetechnologyimpactonthelearningperformance.Theresearchestimateslearningcurvesinacomparativestudybetweenclassicalandroboticlaparoscopyandappealsbothtoobjectivedatacollecteddur-ingexperimentaltestsandsubjectivedatacollectedbyques-tionnaires.
BothobjectiveandsubjectivedataarerequiredforanaccurateevaluationofaVR-basedset-upformechanicallearningpurpose.Thisisthefocusofthepresentresearch.ThiswayallowsmeasuringperformanceadvantagesandestablishinghowmuchtheVRsystemcontributestostu-dents’learningexperience.
3TheVR-basedset-upformechanicalproductdesignlearning
3.1VRtechnologiesclassificationbasedonperceptionPerceptionindesigneducationplaysacrucialroleinincrementingknowledge.Althoughseveralresearchersare
123
106IntJInteractDesManuf(2011)5:103–117
concentratedonvisualperception[18],itisworthtonoticethatindividualsperceiveobjectsandthespacesurroundingthemalsobyallothersensorialmodalities.
Humanbeingsdependonfivesensestoexperiencetheirsurroundingsandinferfromthephysicalobjectsandenvi-ronmentaroundthem.Usersinteractwithobjectsandtheirinterfacebyexperiencingthemwiththeirsensorialmodali-tiesthatgenerateasetofstimuli.Theyarefirstelaboratedatacognitivelevelandthentransformedintoactions.InordertoobtainsimilarconditionsbyVR,itisnecessarytoidentifywhichtechnologybetterstimulateseachsenseandprovidesdeepsensationsintheusers.
TheproposedclassificationstartsfromtheBurdea’sdef-initionofVRas“ahigh-enduser-computerinterfacethatinvolvesrealtimesimulationandinteractionsthroughmul-tiplesensorialchannels”[2].AvailableVRtechnologiesaredividedintofourclasses(visual,sound,hapticandmotion)thatcorrespondtothefoursensorialchannelsinvolvedinthevirtualexperience(vision,hear,touchandmotion).Eachofthemprovidesthecorrespondingsensorialfeedbacktotheusers(Fig.1).
Thesenseofmotion,sometimescalledthesixthhumansense,isusuallyachievedviasomemeansofpositionandorientationtrackingthatmediatetheuser’sinputintotheVRsimulation.Theyincludeallnavigationandmanipulationtechnologies.Thesenseoftouchisperformedbywhatarecalledhaptictechnologies.Tactileandforcefeedbackdevicesseektosimulatetactilecues.Thesenseofvisionisprovidedbyvisualizationtechnologies.Theycanbedividedintosin-gle-userdisplaysandmulti-usersdisplays.Thefirstsaregen-erallydesk-supportedwhilethesecondsarelargevolumedisplaysthatallowthesimultaneouscollaborationofseveralindividuals.Theycanbeflatorcurved,frontofrearprojec-tion-based;theymayprovidepassiveoractivestereoscopicexperience.Thesenseofhearingisprovidedbysoundtech-nologies:stereosound,specializingmultiplesoundsourcesand3Dsound.Theydifferfromeachotherinthelevelofrealisticfeedbacktheysupplytotheuser.
Inordertoachieveafullyimmersiveenvironmentthatimprovesstudents’perceptionofthevirtualscene,alsosmellandtastefeedbackmustbesimulatedbutthecomplexityofthesesenseshasmadethemdifficultforavailableVRtech-nologytoconquer.
Allthesetechnologiesarealsocombinedwiththecom-putinghardwareforVRreal-timesimulationandinteractionsupportandwiththesoftwaretoolkitstomaptheinput/outputdeviceswiththedigitalscene,model3DobjectsandcreatelibrariesforoptimisingVRsimulations.
Fig.1TheproposedclassificationofVR
technologiesandthecorrelationwiththehumansenses
123
IntJInteractDesManuf(2011)5:103–117107
3.2AmethodforVRbenchmarkingtoachievethelearning
goalsInordertoimproveexperimental-basedlearningbyVRtech-nologies,studentsshouldfeelbeinginvolvedinthevirtualenvironmentandbeallowedtotestprinciplesbytouching,hearingandmovingtheobjectstheyareworkingwith.Ithasbeendemonstratedthatthecommunicationmediuminflu-encestheformofinteractionandknowledgeperceptionandcognition,particularlywhenlearnersareunfamiliarwiththecommunicationtechnologiesusedtodeliverinstructionandperformdesigntasks[19].ThereforewestatethatinVRapplicationsforeducation,itisimportanttoevaluatenotonlythelevelofinvolvementperceivedbytheuserbutalsothesystem’susability.
Basedontheseconsiderations,thebenchmarkingofVRtechnologiescombinationisbasedonthreedifferentclas-sesofheuristics:usability,presenceanddepthofsensations.UsabilityconcernsthecapacityoftheVRinterfacestomeettheusers’needs.Thedegreeofthesystemusabilitydependsondifferentcharacteristicssuchastheadoptedtoolsbar-rierfree,ease-to-use,intuitiveness.Presencemeans“beingimmersed”andreferstoanemotionalandmentalstateofbeinginvolvedinthevirtualscene;itdenotesthelevelofengagement.Thesenseofpresenceisdeterminedbysomecharacteristicsofthesystemsuchasinteractivity,collabo-ration,nonconstrainingornavigationsupport.Finally,thedepthofsensationreferstothedegreeofthesensoryfeed-back(visual,tactile,auditory)thatusersfeelwhileexploringthevirtualspace.
Inordertomanagethecomplexityofallpossiblecombi-nationsofVRtechnologies,amatrix-basedmethodisintro-duced.Two3Dmatricesareusedtoconnecttechnologiesandhumansenses:thefirstallowsthemanagementofthecombinationbetweenhaptic,visualandnavigationtechnol-
ogies,whilethelattermatchesthefirstachievedarrange-mentwithsoundtechnologiesandassignsavalueforeachfinalcombinationtoeachheuristicsforassessingthesystem’sperformance(Fig.2).InordertoidentifywhichVRcombina-tionissuitedtoteachspecificmechanicaldesignsubjectsweintroduceanadditionalmatrix,whichcorrelatestheactiv-itiesnecessarytoperformexperienced-basedlearningandthelevelsofsensoryfeedbacknecessaryforperceptionandcognition.Onthecontraryofpreviousmatrices,thatareful-filledbyVRexpertsastheyrelatetotechnicalandfunctionalperformancesandarenotdesignlearning-oriented,thisaddi-tionalmatrixneedstobefulfilledbyprofessorsofmechanicalproductdesignfortheirdeepexperienceoflearningenviron-ments.
Themethodapplicationpreliminarilyrequirestheanaly-sisofwhichactivitiesshouldbeperformedtogainagoodunderstandingofthelessonsubjects.Inthestudycontext,twodifferentmechanicalproductdesigntopicshavebeenexploredandforeachofthemthenecessaryteachingactivi-tieshavebeentraced.
•Productfunctionaldesignandassemblyprinciples.Thetopicaimstodevelopacriticalattitudeintheinterpretationofmechanicalcomponentsandassemblies,infunctionalerrorsdetectionandintheidentificationofassemblyproblems.Infunctionaldesignthetutorgenerallyshowsdifferentfunctionalalternativesinordertoclarifytheseconcepts.Theuseofexamplesisconsideredfundamentaltoimprovegeneralprinciplesunderstanding.
•Dimensionalandgeometrictolerances.Thetopicaimstodeveloptheabilityofidentifyingassemblyandmanufacturingproblemsindifferentdesignsolutionsandtherelationbetweentolerancesandfinalproductquality.Inparticularstudentsshouldbeabletover-ifytheconsequencesoftolerancesprescriptiononthe
Fig.2Synthesisoftheproposedmethod
123
108IntJInteractDesManuf(2011)5:103–117
Table1Assessmentofthesensoryfeedbacknecessarytoundertakethenecessaryactivitiesforlearningproductfunctionalandassemblyprinciplesandgeometricanddimensionaltolerancesprescription
Valuation
DefinitionofthelevelsofsensoryfeedbackforteachingpurposesAssessmentofdesignsolutionalternativescombiningdifferentstandardcomponentswithsimilarfunctions
Assessmentoftheimpactofdesigndecisionsonmanufacturingandassemblyoperations
IdentificationofmanufacturingandassemblyproblemsbyanalyzingmanufacturingoperationsandequipmentsIdentificationoftheright
functionaldesignsolutionbyanalyzingthemanufacturingandassemblycyclecostsofdifferentalternatives
Detectionoffunctionalandassemblyerrorsindifferentdesignsolutions
UnderstandingcomponentsinteractioninassembliesandidentificationofthepropersequenceofassemblyoperationsAbsoluteImportance(bij)
RelativeImportance(ai*bij)Requestedvalueofsensoryfeedback(ai*bij/5*bij)
Interpretationofadesignsolutionandidentificationofthetolerances
Identificationofthedimensionalandgeometrictolerances
referencesandofthedifferencebetweenthem
Understandingofthedifferencesbetweentolerancesaccordingtothecontroltechniques
IdentificationofthemanufacturingandassemblyproblemsderivingfromwrongtoleranceschainAssessmentoftheimpactof
toleranceschainontheassemblyoperationsAbsoluteImportance(bij)
RelativeImportance(ai*bij)Requestedvalueofsensory
feedback(ai*bij/5*bij)
Senseofsight5
LevelsofsensoryfeedbackSenseoftouch5
Senseofhearing1
Senseofmotion
3
52
35
5
1
3
3411
53
55
55
55
612530.8733
55
2
24
455441880.7
355
11
335
manufacturingandqualitycontrolprocessesandoftol-erancesvariationonproductfunctionalities.Studentsshouldalsobeabletorecognizetolerancechainsandidentifytheproperreferences.Professorsinmechanicalproductdesignhavebeeninvolvedinthedefinitionofthemostsignificantactivitiesandthelev-elsofsensoryfeedbacknecessarytoperformthem.Table1showsthenecessarylearninglevelofproductfunctional
123
IntJInteractDesManuf(2011)5:103–117designandassemblyprinciplesanddimensionalandgeo-metrictolerancesprescription.Theevaluationvaluesinthegreycolumnarerelatedtotheimportancethateachactivityrepresentsforthelearningprocess.Thevaluesinthesidecolumnsrepresentthesensoryfeedbackrequiredforperformingeveryactivitysuccessfully,accordingto1–5scale.Forexample,itcanbestatedthatthesenseofsightisveryimportant(thelevelis5)inassessingdesignsolutionalternativeswhilethesenseofmotion(3)andthesenseoftouch(1)aresecondary.Hearingisusefulinnoway.Onthecontrarythefunctionalandassemblyerrorsdetectioninvolvesthesamethreesensorialchan-nelsbuttheyallarekeyfactorsandgainanequalfeed-backlevel(5).Absoluteandrelativeimportancevaluesarecalculatedapplyingmathematicalfunctionsshowedintable.FinalvaluesassessthelevelofsensoryfeedbackrequestedtotheproperVRset-uptechnologyinordertosupporttheconcerningtopic.Incasestudyobtainedfeed-backvaluesareequalrespectivelyto0.87forfunctionalandassemblyprincipleslearningand0.7fortoleranceslearning.
AnexhaustivedescriptionofthebenchmarkingmethodcanbefoundinapreviousresearchworkwhereitwasappliedforidentifyingtheVRsystemthatbetteranswerstodesignreviewsactivitiesrequirements.ExperimentalresultspointedoutthatmostadvantagesintheuseofVRareachievedwhenthetechnologyisselectedaccordingtospecificprocesschar-acteristicsandneeds[19].
AtthispointthreedifferentVRtechnologiescombinations(C1,C2andC3)havebeenchosenfortheassessment.C1isasingle-userlearningenvironmentthatconsistsofahead-mounteddisplay(HMD)coupledwithacommonmouse.C2consistsofalargevolumedisplayprovidedwithste-reoimagingandanoptictrackingsystemthatbothimprovevisualperceptionofdesignsolutionsandcreateacollab-orativeenvironment.C3associatesasimilarlargevolumedisplaywithafinger-basedhapticdevicethatgivesforcefeedbackandtactileperceptionofmaterialsandcompo-nentsinteraction.Theevaluationmethodisbasedonasetofheuristicsthatareconvenientlydefinedforvirtualrealitysystemcontext.Heuristicsaregroupedintothreedifferentdimensions(usability,presenceanddepthofsensation)aspreviouslydescribed.EachVRcombinationisevaluatedbyafive-pointLikertscale[16].
Themethodapplicationallowsthedefinitionofarelativetotalvalueforeverytechnologicalset-up(Fig.3).Comparingtheobtainedtotalvalueswiththesensoryfeedbackrequiredbymechanicaldesignlearningtopicsitispossibletorec-ognizewhichtechnologicalcombinationfitsbetterforthespecificpurpose.
ThecasestudyresultshighlightthatC3combinationisthebestsuitedinsupportingproductfunctionaldesignandassemblyprinciplesteachingwhileC2combinationisthe
109
bestsuitedintoleranceslearning.Ifwewouldidentifythecombinationthatbetteranswerstobothtopics,C3shouldbeselected.Itsfinalvalueisthenearesttobothrequestedfeedbacklevels.
4AstructuredexperimentalprotocoltoassessthelearningprocessinmechanicalproductdesignInordertoverifytheapplicabilityoftheproposedmethodandtotestthemainadvantagesconnectedwiththeuseoftheachievedVRtechnologiescombinations,astructuredexperimentalprotocolisset.Accordingtotheeducationalstandpointitisworthtonoticethatlearners’performancesalwaysdependontwomainfactors:thepersonalaptitudeforspecificsubjectsandtheinfluenceoftheadoptedhard-wareandsoftwaretechnologies.Thesetwodimensionsinmechanicseducationarestrictlyinterconnected:itisdifficulttodistinguishtheirspecificimpactonthelearningprocess.Thereforeunderstandinghowtechnologyaffectslearningisanotatrivialtask.
Inordertofacethisissuetwoevaluationlevelsarecon-sideredinsidetheprotocol:performanceandusability.Thefirstanalysisaimsatmeasuringtheperformancesachievedbystudentsduringlessonsandtoquantifytheminanobjec-tiveway.Inthiscontexttimemonitoring,errordetectionandlearningcurvescouldbeusefulindicators.Thesecondanal-ysisaimsatassessingtheusabilityoftheexploitedsupportsystem.ItcanbestatedthatforlearningscopeahighlyusableVRset-upisneeded.4.1Performanceanalysis
Theperformanceanalysisisbasedonasetofobjectivemet-ricsanddirectevaluationmethods,suchasusertaskanalysis.Performancemetricsaredefinedconsideringthemaindiffi-cultiesofstudentsinmechanicalproductdesignlearning.Theyareexpressedbytangibleentities,suchastimeandpercentagemistakes,relatedtothefollowingaspects:••assembly•assemblyrecognitioncapability,•assemblabilitysequence•surfaceerroridentification,detection,•design•dimensionalalternativesfinishingcomprehension,evaluation,•geometric•matingtolerancestolerancesdetection,•
tolerancestypedetection,geometricalchainrecognition,
tolerancescomprehension,impact.
Eachaspectrepresentsapossibletaskforeachmechanicseducationtopicthatisidentifiedduringthebenchmarking
123
110IntJInteractDesManuf(2011)5:103–117
Fig.3ThestudiedVRcombinationsforsimulatingassemblyoperations.TheyconsistrespectivelyofHMDcoupledwithacommonmouse(C1),alargevolumedisplayprovidedwithstereoimagingandoptictrackingsystem(C2)andasimilardisplaywithafinger-basedhapticdevice(C3)
activity.Metricsareusedtoassessthestudents’performancewhileexecutingthespecifictasks(Table2).
Performanceanalysiscanbeachievedbytheso-called“formativeevaluation”,thatiscommonlyusedinassess-ingtaskperformance.Invirtualenvironmentstheformativemethodissuccessfullyappliedinapplication-specificevalua-tion[9].Itisauser-centredtechniquethatusesaspecifictask-basedscenarioanddefinesactivitiesandtasksflow.Usersworkthroughtheproposedscenarioandevaluatorsobservethemandcollectdata.Notonlyquantitativedatacouldberecordedbutalsoqualitativeones,suchasuser’scommentsandreactions.Inordertovaluateperformancejustquantita-tivedataneeded.Howeverusersmonitoringisusefulforthefollowingusabilityanalysis.
Repeatingmulti-userandtime-spacetestsitispossibletodrawnthetotallearningcurve.Itsslopesuggeststhequal-ityofthelearningprocess[15].Suchperformanceanalysiscanbesuccessfullyappliedinordertoevaluatemechanicaldesignlearningprocessingeneral,withtraditionalmediaorVR-basedmedia.4.2Usabilityanalysis
Performanceisnottheonlyworthwhileaspectinlearningevaluationbutalsotheinvolvementandthetoolsmasteryperceivedbystudentsareimportantfactors.TheusabilityanalysisaimsatinvestigatingtheimpactoftheadoptedmeansofrepresentationonlearningandatassessingthequalityoftheinteractionbetweenstudentsandtheVRsys-temintermsofefficiency,effectivenessandsatisfaction[12].
Itisbasedonusabilitymetricsasquantitativeandqualita-tiveestimatesofstudents’responseandindirectevaluationmethods,suchaspost-hocquestionnaireandobservationsbyexperts.MetricsaredefinedforeachusabilitydimensionconsideringaVRtechnologycontextandthespecificeduca-tionalpurpose.Thereforeincanbestatedthateffectivenessisrelatedtovisibility,interactionsupportandsensoryfeed-backprovidedbythesystem.Efficiencydependsoninfor-mationavailability,easeofuse,mentalworkloadandsystemreactivity.Satisfactionisconnectedwithperceivedcomfortandreliability,simplicityofactionsandinformationquality(Table3).
Tomeasureusabilitymetricsweadoptedtwodifferentinvestigationtechniques[9],inordertocollectqualitativeandquantitativedata:users’observationandpost-hocques-tionnaires.Subjectiveimpressionsarecollectedbyquestion-nairesandstudentsconsidertheirownlearningexperienceandscoreeachmetricafterperformingthetaskscenario.Afive-pointLikertscale[16]ischosentoassignavaluetomet-ricsindicatorsandeachscoreisweightedaccordingasstu-dentsgetusedtovirtualmodellingtools,i.e.traditionalCADsystems,ornot.Westatethatthemultipliercorrespondsto0.5ifthestudentisconfidentwithvirtualmodellingandto1ifhe/sheisnot.Quantitativedataarecollectedbyusersdirectmonitoringorbyvideorecording.Theobserverjudgesuser’sbehaviourandexpressesavalueforeachusabilitymetricbyfollowingthesameLikertscale.
Usabilityanalysisbecomesfundamentalincomparingdif-ferentVRset-upandestablishingwhichisthebestforthespecificpurpose.
123
IntJInteractDesManuf(2011)5:103–117Table2Proposedperformancemetricsformechanicaldesignlearning
111
TopicTaskDescription
Itisaskedtorecognizetheform,thetotalvolumeandthemainfunctionoftheproduct
MetricsTime(min)Mistakes(%)
ProductfunctionalGeneraldesignandassemblyassemblyprinciplerecognition
Assemblysequencerecognition
ItisaskedtorecognizetherightassemblysequenceTime(min)andimpossiblemountingconfigurations
Mistakes(%)
AssemblabilityItisaskedtodetectpositionalerrorsinassemblyerrorsdetectionandrecognizechangesinfunctionSurfacefinishingItisaskedtocomprehendsurfacefinishingcomprehensionandappropriatemanufacturingprocessesDesign
alternativesevaluation
Dimensionalandgeometrictolerances
DimensionaltolerancesdetectionGeometrictolerancesdetectionMatingtyperecognition
Itisaskedtoevaluateassemblysolutionalternativesanddetectwrongsolutions
Time(min)Mistakes(%)Time(min)Mistakes(%)Time(min)Mistakes(%)
Itisaskedtoidentify’dimensionalfeatureswheretolerancesareprescribed
Time(min)Mistakes(%)
Itisaskedtoidentifygeometricfeatureswheretolerancesareprescribed
Time(min)Mistakes(%)
Itisaskedtorecognizeifgaporinterferencematingsaredemanded
Time(min)Mistakes(%)
ToleranceschainItisaskedtorecognizetoleranceschainsandrecognitionreferencedimensionsGeometrictolerancesimpact
Itisaskedtocomprehendthedatumfeature,thetolerancesvaluesandtheirimpactonfunctionandmanufacturingprocess
Time(min)Mistakes(%)Time(min)Mistakes(%)
5Experimentaltests
OneoftheselectedVRcombinationsisevaluatedbythepro-posedprotocol.Theexperimentalset-uphasbeenchosenonthebasisofthebenchmarkingillustratedinSect.3.More-over,thewidespreadofavailableVRtechnologiesandtheirlowcostmakethearrangementofapropervirtualenviron-mentdifficulttoachieve.ItisnecessarytorecognizethemainlimitsandpotentialitiesofdifferentVRset-upsforspecificeducationpurposes.Theproposedmethodfortechnologicalbenchmarkingaimsatansweringsuchquestionsandiden-tifyingtheVRcombinationthatisbestsuitedtoimprovemechanicaldesignlearning.
Thespecificset-upconsistsofalargevolumedisplayprovidedwithstereoimagingandcoupledwithafinger-basedhapticdevice(C3).Actually,benchmarkingresultshavehighlightedthatC3combinationisthebesttocarryouttheproposedmechanicaldesigntasks.C3relativetotalvalue(0.78)isthenearesttobothrequestedfeedbacklevels(0.7and0.87).Thevisualizationtechnologyisrepresentedbyalargerearprojection-baseddisplaythatprovidespassivestereoscopy.BarcoProjectorsareusedtoprovidehighreso-lution.ThehaptictechnologyisrepresentedbyaPHANTOMdevice(PersonalHApticiNTerfaceMechanism)bySensA-bleTechnologies.Aphantomisahandedforcefeedbacktoolwhereastylusissupportedbyanelectromechanicalarm.Theuserholdsthestylusexperimentingforce,motionandtactilefeedback.ThehearingtechnologyconsistsofastereodigitalsystembyPioneerthatisconnectedwiththecentralworksta-tion.TheworkstationismadeupofaHPseriescomputer,anAMDOpteronprocessor,250GBharddisk,4GBRAMandtwoNVIDIAgraphiccards.ConcerningtheCADapplica-tion,SolidEdgeandVisMockupbyUGSareinstalledonthemachine.Thecrucialpointdealswiththehardware/software
123
112
Table3ProposedusabilitymetricsformechanicaldesignlearningUSdimEffectiveness
MetricsModelvisibilityInteractionsupportSensorialfeedback
Efficiency
InformationavailabilityEaseofuseMentalworkloadSystemreactivity
Satisfaction
PerceivedcomfortSimplicityofactionsPerceivedreliabilityInformationquality
Description
Itdefineswhethermodel’sparts,areclearlyvisibleandaccessiblefromthestudent’sperspective
ItmeasureshowthevirtualmodelcanbemanipulatedandmodifiedinallitsfeaturesItmeasuresthesystemcapacity’togivefeedbacktothestudent’sactionsbyinvolvingallsensorialchannels
Itdeterminesthesystemabilitytogetalltheinformationrequiredforperformingthespecifictask
Itmeasuresthelevelofphysicalandmentalcommitmentrequiredforperformingthespecifictask
Itindicatesthelevelofmentalstressandstrainduringtests,thatisrelatedtohowsystemsupportslearnability
Itmeasuressystempromptnessinfollowingstudent’sactionsandupdatethesceneinreal-time
Itmeasuresthelevelofcomfortperceivedbythestudentmmovingandmanipulatingobjectsinvirtualscene
Itquantifiesthecontrollabilityandease
providedbythesysteminactionandhandleparts
Itmeasuressafetyandreliabilityimpressionsperceivedbythestudentabouttechnologyandtools
Itmeasurestheclarityofthereceivedinformationandifitisgoodenoughforperformingthespecifictask
IntJInteractDesManuf(2011)5:103–117
QualitativemeasurePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.scorePost-hocQ.score
QuantitativemeasureStatements,requestsforexplanation
Statements,requestsforfurthersupport
Statements,requestsformorefeedbackStatements,requestsforexplanation
Statements,requestsforexplanation
Statements,requestsforexplanation
Statements,requestsforfurtherreactionStatementsStatementsStatementsRequestsforfurtherinformation
connectioninordertointegratethehapticdevicewiththevirtualscene.Inthematterofcreatingthisconfiguration,avalidsolutionissuggestedbyarecentresearch[11].ItshowshowCADgeometrycanbetranslatedintotheproperdigitalformatreadablefromthephantomhapticdevice.Indetailasoftwareapplicationschainforconverting.jtfilesin.objfileshasbeendeveloped.
Thepresentedexperimentalset-upallowsuserstointeractwiththeVRinterfaceinanaturalmanner.Theycanalsoexpe-rienceamultimodallevelofperceptionbysimultaneouslyhavingvisual,auditoryandtactilefeedbacks.Inparticularthehapticdeviceintegrationmakesuserstoexperienceobjectsphysicalitybyenhancingthestiffnessperceivedthroughthestylus.Bycollectingperformanceandresponsefromusers,itispossibletoquantifywhetherandhowtheVRsystemsupportsthemintheirowngoals(Fig.4).
Theexperimentalarrangementhasbeentestedbythepro-posedprotocolinSect.4.Testshaveinvolvedtenmechanicaldesignstudentsattendingthefirstyearofamechanicalengineeringcourse.Thetestgrouphasbeendefinitelynonhomogenousinpracticeanddesigncompetenciesbecauseofdifferentsecondaryschoolbackgrounds.Inthestudygroup
Fig.4Proposedexperimentalset-upforevaluatingtheselectedVRcombination
fivestudentshaveatechnicalcollegeeducation,threeofthemattendedasecondaryschoolfocusingonsciencesandtwoofthemhadanon-technicaleducation.Studentshavebeenval-uatedbyindividualtestingsessionsinsteadofclasstestinginordertocollect20diverseresponsesandavoidrecipro-calinfluencesinanswering.Eachstudenthasbeenasked
123
IntJInteractDesManuf(2011)5:103–117Fig.5Exampleoftestingassemblyusedforperformanceanalysis
113
toexaminefivedifferentmechanicalassembliesofmedium-levelcomplexity.Theyallaremadeoutofabout20–25parts.Thetwo-dimensionaldrawinginFig.5depictsoneofthefiveassembliesselectedforthetrainingsessions.
Eachsingleassemblyisanalysedviathreedifferentrep-resentationalmeansforsupportinglearning:bytraditionaltwo-dimensionaldrawings,bytri-dimensionalCADmodelsandbyvirtualmock-ups.Tutorshavepreviouslypreparedthesameexercise.Itispresentedtostudentsbyadoptingthethreedifferentmeansofrepresentation.Infirstcaselecturesaresupportedbybi-dimensionaldrawingsandgeneralprin-ciplesareillustratedbyslides,inthesecondcaseaparametricmodelhasbeengeneratedinUGSSolidEdgeenvironmentanddisplayedbyacommondesktopcomputer.Finally,inthethirdcaseavirtualmock-uphasbeencreatedandvisualizedintheidentifiedvirtualset-up.Thedifferencebetweenthethreedifferentmeansallowscheckingprotocolvalidityandinvestigatingthequalityofthedifferenttechnologiessup-portingmechanicaldesigneducation.
Studentshavebeensubmittedtoperformanceanalysisasscheduledbyprotocolinthethreelearningchannels.Dur-ingtrainingsessionstheyhavebeenobservedbyexpertsandvideorecorded.Recordingstudentsatworkbyvideointer-actionanalysis(VIA)techniquehelpsexpertsincollectingobjectivedata,definingperformancemetricswithprecisionandgatheringusers’commentsorreactions.Nousabilityanalysishasbeencarriedoutbecauseofitsinherentnature.Meaningfulresultsarepresentedinfollowingtables.Foreachselectedtaskandforeachlearningchannel,metricsvaluesarecollected.Tables4and5respectivelyshowexamplesofresultsintermsoftimeandnumberofmistakes.Testshavebeenperformedinvolving10personson5assemblyexer-cises,totallyhaving50trials.Thenumberofsampleusershasbeendeterminedbyanalysingotherprotocolstudiescar-riedoutbyresearchersinpsychologyforstudyingcognitivebehavioursandthoughtprocesses[5].Asthepresentresearchisapilotstudyitistheminimumnumberrequestedtoper-formconcurrentprotocolbutshouldbeincreasedinordertovalidatethemethodandmakethestatisticresultssignificant.Meanvaluesintimehavebeencalculatedconsideringthetimeperformanceofeachstudentandapplyinga5%cut-offonthefinalGaussiandistribution.Percentagevaluesformistakesarecalculatedbycountingtotalwronganswersanddividingbythetotalattempts.A10%resultmeansthatfivetrialshavefailed,withoutspecifyinghowsomestudentsgivenoncorrectanswers.Inordertoproperlyinterpretingtheresults,itisworthytonoticethattasksareperformedinthesamesequencetheyarepresentedinTables4and5.There-foretimedataarenotabsoluteinvaluesbecausethefirstcomprehensionphasesinfluencethefollowingones.Tablesneedtobereadhorizontally,comparingdifferentlearningtoolsonthesametask,notvertically.Inthiswayitispos-sibletoidentifywhichactivitiestakeadvantagebytheVRlearningset-up.
Aspredictedintheprevioussections,theanalysishigh-lightsageneralpositivetrendusingtheexperimentalVRset-upinsteadoftheothermeansofrepresentation.Severaltestsresultsshowareductionintimeandmistakesnumber.Focusingonsinglevalues,itcanbestatedthatVRmakesgeneralassemblyrecognitionandsequencecomprehensioneasierandfasterthanotherrepresentations.Weconsiderthatismainlyaconsequenceoftheenhancedvisualandtactileperception.TheadoptedVRalsosupportssurfacefinishingcomprehensionandmatingtyperecognition.Highqualityrenderingandhapticfeedbackaredirectlyresponsibleforitalthoughtheadoptedtechnologyisnotabletoprovideafinetactilefeedbackaswellasrequested.
OtherVRtechnologiescombinations(C1,C2)couldnotprovidesimilarresults.Thefirstdoesnotsupportpartici-pationandinteractionbetweenmultiplestudentsandtutors
123
114IntJInteractDesManuf(2011)5:103–117
Table4Experimentalresultsofperformancetimeondifferentlearningchannels
TopicTaskLearningchannelsTime(ms)2Ddrawings
3DCADmodels10.3005.0007.1006.2010.2007.4010.1005.4010.0010.40
VRmockups08.1503.3006.5004.0010.5008.0011.0002.3010.2009.30
Functionaldesignandassemblyprinciples
GeneralassemblyrecognitionAssemblysequencerecognition
Asseinblabilityerrorsdetection
SurfacefinishingcomprehensionDesignalternativesevaluation
Dimensionaltolerancesdetection
Geometrictolerancesdetection
MatingtyperecognitionToleranceschainrecognition
Geometrictolerancesimpact
15.2006.3010.3008.4012.0005.1015.3006.0016.3020.10
Dimensionaland
geometrictolerances
Table5Experimentalresultsofperformancemistakesondifferentlearningchannels
TopicTaskLearningchannelsMistakes(%)2Ddrawings
3DCADmodels6181410382634262430
VRmockups4402302828202424
Functionaldesignandassemblyprinciples
GeneralassemblyrecognitionAssemblysequencerecognition
Asseinblabilityerrorsdetection
SurfacefinishingcomprehensionDesignalternativesevaluation
Dimensionaltolerancesdetection
Geometrictolerancesdetection
MatingtyperecognitionToleranceschainrecognition
Geometrictolerancesimpact
30333442503850365254
Dimensionaland
geometrictolerances
thatseemtobefundamentalforgeneralprinciplescompre-hensionafterexperiments.Thesecondonestimulatesonlythesenseofvisionwhileassemblyrecognitionneedalsoanhapticfeedbackasdemonstratedbystudentsrequestsandquestionnairesanswers.Usabilityanalysismayenforcethesestatementsoncecarriedoutonthedifferentmeansofrepre-sentation.WecouldacknowledgethattheadoptedVRset-updoesnotcreategreattimeadvantageifcomparedtodesktop-basedsystemssupportingbyCADtechnologies.Assemblabilityerrorsdetectionisavalidexample.Studentsneedsimilartimetounderstandbytrialanderror.HoweverVRmock-upsdeterminesignificantreductionsinmistakes.Thisisduetothefactthatstudentscandirectlyexperienceinterference
123
IntJInteractDesManuf(2011)5:103–117115
matingbythehapticphantomandthatthesystemitselfpre-ventsforerroneousmounting.
TheselectedVRcombinationset-upappearstosupportalotofdesignactivities.Howevernogreatimprovementsareachievedinfewreallyhardissuesincomparisonto3DCADmodels.Themeasuredvaluesoftimeandmistakesmet-ricsforeachrepresentationaresimilarinthefollowingtasks:designalternativeevaluation,tolerancechainrecognitionandgeometrictolerancesimpact.Thesepreliminaryresultswilldirectthechoiceofapropersoftwaretoolkittomodelandmanagevirtualmock-upsinordertomeetalltasksneces-saryformechanicsdesigneducation.Theadoptedsoftwaresystemactuallyallowsthesolerepresentationofnominalgeometry.Dimensionalandgeometrictolerancesprinciplesrequiresmorecomplexrepresentationsabletomakestudentsawareofthecomponentsandassembliesvariationsfromthenominalgeometryrespectivelyintermsofsizeandofform,orientation,locationandrun-out.
Analysiscouldbeimprovedbyassociatingtheinput/outputdeviceswithappropriatesoftwaretoolkitsinordertomanageassemblymatingandtoleranceschainsinamoreeffectiveway.Toachievethisissue,additionalinformationrelatedtomaterialpropertiesandreal-timedeformationisnecessary.Twomainapproachesarebeingpursuedtosupportinteractionsbetweenparts:physically-basedmodellingandconstraint-basedmodelling.Theybothprovidethesemanticinformationnecessaryforsimulatingphysicalassemblyandtolerancesinvirtualenvironment.Currentcommercialsys-temslackofageometricconstraintdetectionmanagement,soaspecificarchitectureshouldbearrangedandconnectedwiththecentralworkstation.
Statisticaltestsareperformedtobetterhighlighttheachievedresultssupportingtheabovementionedconsider-ations.
Histogramrepresentationscanbeusedtopointoutthetimeandmistakesmetricsvaluesdistributionforeachiden-Fig.6Histogramoftimemetric
tifiedtask(Figs.6,7).Bylookingatdiagramsit’seasytounderstandhowVRtechnologieshelpstudentsbyreduc-ingbothperformancetimeandmistakes.Howeveritcanbenoticedthatdropinmistakesratioismorerelevantthantimereduction.Almosteveryactivityhashighlightedgreatimprovementsandthemajoradvantageshavebeenreachedingeneralassemblyrecognition,assemblysequencerecog-nition,assemblabilityerrorsdetectionandsurfacefinishingcomprehension.Asanexample,assemblyrecognitionhasshowna30%ofmistakesin2Ddrawingmodality,insteadof6%with3DCADmodelsand4%withVRmock-ups.Thesurfacefinishingcomprehensionhasrevealedthebig-gestimprovement,froma42%ofmistakeswith2Ddraw-ingstoa10%with3DCADmodelsandtoa2%withVRmock-ups.Achievedadvantagesarelesssignificantfortol-erancesanalysis,suchasgeometrictolerancesdetectionandtoleranceschainrecognition.Thatstrengthenswhatstatedbeforeaboutthenecessityoffurthersupportsintolerancesanalysisactivities.
6Conclusionsandfuturework
ThepresentworkisastepforwardstheexplorationoftheadvantagesofVRinexperienced-learningapproachappli-cation.Itaimstodefineaproperevaluationprotocolandanadequateexperimentalset-upfortestingifmechanicalprod-uctdesignlearningisincrementedbyimmersingstudentsinavirtualenvironment.AlthoughtheuseofVRinteachingandlearningisnotnewinitself,experimentationismainlybasedonavailabletechnologiesnotreallysetforthespecificpurposes.AbenchmarkingmethodisproposedtoidentifytheproperVRtechnologiescombinationtosupportmechanicaldesigneducation.Thiscombinationisthenusedtostructurethevirtualset-upforexperiments.
123
116IntJInteractDesManuf(2011)5:103–117
Fig.7Histogramofmistakesmetrics
Performanceanalysishasbeencarriedoutonthreediffer-entmeansofrepresentation.Experimentalresultsshowthat:•2Ddrawing-basedteachingmethodrequiresalotofeffortintimeforcompletingallproposedtasksandimpliesalotofmistakes;
•3DCADmodelssupportstudentsbothingeneralassem-blyprincipleslearningandindesigntolerancesunder-standing.Mainprofitsarerepresentedbyagoodreductionbothintimeandmistakesnumber;
•VRmock-upsareabletobettersupportmechanicaldesignlearningandreachextrabenefits.Itisconfirmedbyfurtherreductionincompletiontimeandmistakesratio;•Virtualrealityallowsanenhancedvisualandtactileperceptionandagreatersensorialfeedback.Itmakesgeneralassemblyrecognition,sequencecomprehension,surfacefinishingcomprehensionandmatingtyperecog-nitioneasierandfasterthanotherrepresentations;
•VRmock-upsadvantagesaresignificantforthesimpleractivitiesifcomparedtoothersrepresentationtechniques,whiletheydecreaseforthemorecomplextasks;
•ThetestedVRcombination(C3)quiteproperlysupportsfunctionaldesignandassemblyprinciplesteachingbutneedtobeimprovedinordertobettersupporttoleranceslearning.Thescientificcontributionofthisworkconsistsintheselec-tionofaVRsystemaccordingtotheanalysisofperceptionandcognitionmechanismsinthelearningprocessandinthedevelopmentofagoal-orientedprotocolbasedonuserperformanceandsystemusability.Theproposedapproachcanbeadoptedtoothereducationapplication:benchmark-ingcanbeusedtoidentifywhichVRset-upbetteranswertospecificlearningpurposedwhiletheprotocolcansupportperformanceandusabilityanalysisofdifferentVRtechnol-
ogies.Thedifferenceconsistsintheactivitiesthatshouldbecarriedoutbysampleuserstoimprovelearningandthatdeterminethespecificcontext.
Moreover,themethodapplicabilityisnotlimitedtolearn-ingcontextsbutitcouldbeconsideredvalidforevaluatingtheuseofVRinindustrialapplicationsingeneral.Metricsjustneedtobenormalizedandweightscalibratedfornewspecificpurposes.
Futureworkscouldbeorientedtosetupamorecomplexexperimentalsystem,wherethepresentVRcombinationisupgradedbyamorecomplexsoftwaretoolkittoprovidebet-terperformances.FurtherdevelopmentscouldbeachievedbyjoiningVR-basedsystemwithcommercialCAT(Computer-AidedTolerances)softwareforstatisticaltolerancesanaly-sis.InsuchcontextCADmodelrepresentsthecommondatasourcefromwhichbothVR-basedandCATsystemsobtainthenecessaryinformation.Atthispointanalysiscanfollowtwodifferentapproaches:VR-basedassemblysimulationorCATtechnologyvariationsimulation.
Students’performancescouldbeevaluatedinbothcasesbyapplyingtheproposedprotocolandtheidentifiedmetrics.Thecomparisonofexperimentalresultsallowstopointoutstrengthsandweaknessesofeverytechnologyinsupportingmechanicseducation.References
1.Burdea,G.C.:Invitedreview:thesynergybetweenvirtualrealityandrobotics.IEEETrans.Robot.Autom.14,400–410(1999)2.Burdea,G.C.,Coiffet,P.:VirtualRealityTechnology,2thedn.Wiley,USA(2003)
3.Blavier,A.,Gaudissart,Q.,Cadiere,G.B.,Nyssen,A.S.:Compar-isonoflearningcurvesandskillstransferbetweenclassicalandroboticlaparoscopyaccordingtotheviewingconditions:implica-tionsfortraining.Am.J.Surg.194,115–121(2007)
4.Creem-Regehr,S.H.,Willemsen,P.,Gooch,A.,Thompson,W.:Theinfluenceofrestrictedviewingconditionsonegocentricdis-
123
IntJInteractDesManuf(2011)5:103–117
tanceperception:implicationsforrealandvirtualindoorenviron-ments.Perception34,191–204(2005)
5.Dorst,K.,Dijkhuis,J.:Comparingparadigmsfordescribingdesignactivity.Des.Stud.16,74–261(1995)
6.deSaGomes,A.,Zachmann,G.:Virtualrealityasatoolforverificationofassemblyandmaintenanceprocesses.Comput.Graph.23(3),389–403(1999)
7.Hamade,R.F.,Artail,H.A.,Jaber,M.Y.:EvaluatingthelearningprocessofmechanicalCADstudents.Comput.Educ.49,640–661(2007)
8.Hernandez,J.D.,Bann,S.D.,Munz,Y.:Qualitativeandquantita-tiveanalysisofthelearningcurveofasimulatedsurgicaltaskonthedaVincisystem.Surg.Endosc.18,372–378(2004)
9.Hix,D.,Hartson,H.R.:DevelopingUserInterfaces:EnsuringUsabilityThroughProduct&Process.Wiley,NewYork(1993)10.
Hakkarainen,K.,Lipponen,L.,Jarvela,S.:Epistemologyofinquiryandcomputersupportedcollaborativelearning.In:Koschmann,T.,Hall,R.,Miyakem,N.(eds.)CSCL:CarryingFor-wardtheConversation,pp.129–156.LawrenceErlbaum,Mahwah(2002)
11.Howard,B.M.,Vance,J.M.:Desktophapticvirtualassemblyusingphysicallybasedmodelling.VirtualReal.11,207–215(2007)12.ISO9241-11.ErgonomicRequirementsforOfficeWorkwithVisualDisplayTerminals(VDTs)—Part11:GuidanceonUsability(1998)
13.Kaye,A.R.LearningTogetherApart.In:KayeA.R.(ed.)Col-laborativeLearningandComputerConferencing.ComputerandSystemSciences(90).Springer,Berlin(1992)
14.
Kolb,D.A.:ExperientialLearning:ExperienceastheSourceofLearningandDevelopment.PrenticeHall/EnglewoodCliffs,NJ(1984)
117
15.Lieberman,M.B.:Thelearningcurve,diffusionandcompetitive
strategy.Strateg.Manag.J.8(5),441–452(1987)
16.Likert,R.,Hayes,S.:SomeApplicationsofBehavoiural
Research.UNESCO,Paris(1957)
17.Murgio,M.P.:CommunicationsGraphics(1969)
18.Mohler,J.L.:Methodologyonspatialproblemsolutionsamong
highandlowvisualachievers.J.Ind.Technol.24,1–9(2008)19.Mengoni,M.,Germani,M.,Bordegoni,M.:Virtualrealitysystems:
amethodtoevaluatetheapplicabilitybasedonthedesigncontext.In:ProceedingsoftheInternationalDesignEngineeringTechnicalConferences&ComputersandInformationinEngineeringCon-ference,September(2007)
20.Moran,S.,Rubio,R.,Gallego,R.,Suarez,J.,Martin,S.:
Proposalofinteractiveapplicationstoenhancestudent’sspatialperception.Comput.Educ.50,772–786(2008)
21.Oxman,R.:Educatingthedesignerthinker.Des.Stud.20,105–
122(1999)
22.Savall,J.,Borro,D.,Gil,J.J.,Matey,L.:Descriptionofahaptic
systemforvirtualmaintainabilityinaeronautics.In:ProceedingsofIEEE/RSJInternationalConferenceonIntelligentRobotsandSystems,Lusanne,Switzerland(2002)
23.SoH.,BrushT.A.:Studentperceptionofcollaborativelearning,
socialpresenceandsatisfactioninablendedlearningenvironment:relationshipsandcriticalfactors.Comput.Educ.(2007,inpress)24.Towle,E.,Mann,J.,Kinsy,B.:Workinprogress—development
oftoolstoimprovethespatialabilityofengineeringstudents.In:FrontiersinEducation,FIE’05.Proceedings35thAnnualConfer-ence(2005)
25.Volkov,S.A.,Vance,J.M.:Effectivenessofhapticsensationforthe
evaluationofvirtualprototypes.J.Comput.Inf.Sci.Eng.123,123–128(2001)
123
因篇幅问题不能全部显示，请点此查看更多更全内容