• No results found

MAESTRO - a tool for interactive assembly simulation in virtual environments

N/A
N/A
Protected

Academic year: 2022

Share "MAESTRO - a tool for interactive assembly simulation in virtual environments"

Copied!
12
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

simulation in virtual environments

RolandStean 1

andTorstenKuhlen 2

1

VirtualRealityCenterAachen

c/oInstituteofTechnicalComputerScience,AachenUniversityofTechnology

2

(correspondingauthor)Virtual RealityCenterAachen

c/oComputingCenter,AachenUniversityofTechnology

[email protected]

Abstract. WeintroduceMAESTRO,aVirtualRealitybasedassembly

simulation toolthat comprisesphysically-based modeling,haptic feed-

back and articialsupport mechanisms. Thefocus of this paper is on

the developed support mechanisms and the evaluation of the system.

Theexperimentalresultsshowthatallthreefeatures{haptics,physics,

andarticialsupport{considerablyimproveuserperformanceanduser

acceptanceduringthecompletionofassemblytasksinavirtualenviron-

ment.

1 Introduction

Duringthelast years,Virtual Reality(VR)hasprovenitspotentialfor thevi-

sualizationandmanipulationofcomplexdatalike3-Dgeometriesgeneratedby

meansofCADapplications.Aninterestingandpromisingareaofapplicationfor

virtualenvironmentsisassembly simulation. Thoughvirtualenvironmentscan

alreadyprotably beintegrated into engineers' dailywork,there are stillalot

of disadvantages in existing VR-based assembly simulation systems, requiring

trainedengineers forperforminginteractivesimulations.Especially,the lackof

satisfactoryfeedbackmechanismsinexistingVR-basedsimulationenvironments

complicatesexecutionof interactivemanipulationsof virtualobjects.Indetail,

suitable modeling of user-objectinteractions, realistic simulationof object be-

haviour,and intuitive presentation of informationare missing.Duringthe last

2years,wehavebeendevelopingasoftwarecalledMAESTRO(MultimodalIn-

teractionTechniquesforAssemblySimulationinVirtualEnvironments),where,

forthersttime,thefollowingfeaturesarecombinedinasingle,comprehensive

tool:

{ Realistic,physically-basedbehaviourofvirtualobjects,includingautomatic

calculationoftheobjects'inherentmechanicalcharacteristics

{ Integrationofforcefeedbackintotheinteractionwithvirtualobjects,oper-

atingonauniquescenegraphforboth,graphicsandhaptics

{ Articial support mechanisms like sensitive polygons, virtual magnetism,

(2)

oftheMAESTROfeaturesin detail,butinsteadtogiveinsightintotheoverall

MAESTROfunctionality,and todemonstratebymeansofasystemevaluation

thatthecombinationofallfeaturescancontributesignicantlytoabetteruser

performance anduser acceptanceduring thesimulationof assembly tasksin a

virtualenvironment(VE).

The remainderof the paperwill start with a short surveyof existing VR-

basedassemblysimulationtools.Insection3,ourapproachiscomparedtoother

simulationsystems.Furthermorethephysically-basedmodelingandthehaptics

component of MAESTRO are briey described here. Section 4 illustrates the

developedarticialsupportmechanisms,andsection5givesanoverviewofthe

MAESTROhardwareand software. Finally, section 6evaluatesthe systemby

meansoftwoexperiments.Thepaperendssomeremarksaboutfuturework.

2 A Brief Survey of Assembly Simulation in VE

Basically,available systemsfor interactiveassembly simulationin virtualenvi-

ronments can besubdivided into knowledge/rule-based approaches and physi-

cally-basedapproaches.

R.Heger[1]introducedaknowledge-basedsystemthatallowsaninteractive

executionofmanualassemblytasksinavirtualenvironment.Thesystemworks

withtheconceptofreferencepoints,i.e.,singleverticeswithinthepolygonmod-

elswhichcanbeplacedatconnectionelements(e.g.,screws)ortargetpositions.

Forthesinglereferencepoints,assembly-specicobjectcharacteristicsarestored

and takecarethat objectsareplacedautomaticallyand exactlywhentheuser

approachessuchareferencepoint.

A similar approach was chosen by M. Grafe [2] for a virtual construction

system that uses basic elements. Functional nodes can be attached to single

surfacesoftheseelementscausing,e.g.,asnap-inwhentwocorrespondingnodes

areapproaching.

B. Jung [3] describes a knowledge-based system that allowsan interactive

assembly of virtualbasicelementsto complexmodules. Here,apolygonaland

alogicaldescription arestored in twoseparate knowledge databases.Therst

one contains object characteristics which are relevant to assembly, and which

arerepresentedbysocalledports.Thesecondoneservestodescribeaconstruc-

tion goal, i.e., akind of plan howto create modules from singleobjects. This

knowledge allows the system to recognize assembled objects as modules and,

by inferencefrom the knowledgedatabase, to deducethe specic useof single

elementswithin amodule.

All three knowledge-based systems are using a tracked instrumented glove

and3-Dmices asinteractiondevices anddonotprovideforce feedbackmecha-

nisms.IntheVADEsystemdevelopedbyS.Jayaram[4]andanassemblysystem

introducedbyM.Buck[5],bimanualinteractionisincludedbymeansoftwoin-

strumentedgloves.Bothsystemscontainaphysical,constraint-basedapproach

(3)

intersect,theobjectsaresimultaneouslypresentedtotheuserattheirphysically

plausiblepositionsand,inawireframerepresentation,attheirintersectedposi-

tions.Whilein thesystemofBuckreactionforcesarevisualizedasvectors,the

VADEsystemshouldprovidearealforce feedbackbymeansof anexoskeleton

glove.

R. Guptaet al. [6]introduce an assembly simulation systemthat provides

forcefeedbackbymeansoftwoPHANToMHapticDevices[7].Here,thegraph-

ical representation of the scene and the physically, constraint-basedmodeling

are completelyseparated. Incontrastto all other systemsmentionedhere,the

systemofGuptaisrestrictedtotwodimensionsanddoesnotsupporttheuseof

immersivedisplays.

3 The MAESTRO Concept

MAESTRO combines a physically-based and a knowledge-based approach in

order to prot from the advantages of bothstrategies, i.e., realisticobject be-

haviour on the one hand and eective manipulation of virtual objects on the

other hand. The inexibility of rule- and knowledge-based systems is avoided

here,becausemostof thenecessaryknowledgeis generatedautomatically, and

becauseonlyafewassembly-relevantrulesareapplied.Physically-basedmodel-

ing(PBM)andarticialsupportmechanisms,basedonprioriknowledgeabout

the assembly goal, are completed by haptic feedback functionality. Our ap-

proachesfor PBMand hapticsare only briey introduced in theremainder of

this section, whereasthe support mechanismswill be described in moredetail

in thenextsection.

Physically-Based Modeling Toachievea realistic behaviourof virtual ob-

jectsbymeansofphysically-basedmodeling,mass,frictioncoeÆcient,centreof

gravity,andinertiatensorofthevirtualobjectsmustbeknown.InMAESTRO,

these inherentmechanicalcharacteristicsarecalculatedautomatically fromthe

polygonaldescriptionofthemodelgeometry.Sincethegeometriescannormally

beeasilyexportedfrom aCAD application,MAESTROpossessesahigh exi-

bility.Inprinciple,theproblemcanbereducedtovolumedeterminationofthe

polygonalmodels.Here,wehaveimplementedandcompareddierentexactand

approximativealgorithms[8].

Starting from themechanical object characteristics calculated in step one,

we have, in contrast to other systems (see section 2), preferred an impulse-

basedapproachtoaconstraint-basedapproach,becauseitallowsforaconsistent

methodical treatment of all kinds of contact between virtual objects. Most of

thealgorithmswehaveintegratedhere,arewelldocumentedinthepublications

of B.Mirtich and D. Bara (see, e.g.,[9{11]). Whenever possible, reactionsto

collisionsarecalculatedalgebraicallyinordertofullltherealtimerequirements

(4)

asimple,rule-basedapproachwasfound.

Haptic Feedback A severedecitoftoday'ssystemsforinteractiveassembly

simulationisthelackofmultimodal,especiallyforcefeedback.Thefewexisting

systems likethose of Jayaram or Gupta (see section 2) operate with aredun-

dant haptic scene graphin order to provide stability of the haptic rendering.

In contrast to these systems, our approach is based on a single scene graph

for graphics aswell ashaptics,thus dramatically reducingmodelingcosts and

gainingahighersystemexibility. Inorder to achievestability evenwithouta

hapticdescriptionofthescene,weareoperatingwithaninterimrepresentation

of contact situations.Wewill decribethis approachin detail in aforthcoming

paper.

4 Articial Support Mechanisms

In order to compensate for the problems which inevitably exist in VR-based

assembly operations,arisingfrom thenon-exactmodeling ofgeometry and be-

haviour of virtual objects as well as from inadequacy of available interaction

devices, wedevelopedand integratedarticial support mechanismsinto MAE-

STRO.Figure1illustrateshowtheimplementedmechanisms{guidingsleeves,

sensitivepolygons, virtualmagnetism, and snapin { areassigned to the three

classicalphasesofanassemblyprocess.

Fig.1. Assignment of the integrated support mechanisms to the single interaction

(5)

prioriknowledgeaboutpossibletarget positions andorientationsof objects.In

thecongurationphaseofanassemblysimulationwithMAESTRO,theguiding

sleeves are generated in form of a scaled copy or a scaled bounding box of

the objects that are to be assembled, and then placed as invisible items at

thecorrespondingtargetlocation.Whenaninteractivelyguidedobjectcollides

with an adequate guiding sleeve, the motion path necessary to complete the

assemblytaskisanimatedasawireframerepresentation.Inaddition,thetarget

positionandorientationisvisualizedbymeansofasemi-transparentcopyofthe

manipulated object(seeFig.2).

Fig.2. Guiding sleeves supportthe user during the transport phase of anassembly

taskbymeansofawireframeanimationoftheassemblypath

SensitivePolygons Sensitivepolygonsarecreatedbeforesimulationstartand

positionedatadequatepositionsonobjectsurfaces.A functioncanbeassigned

to everysensitivepolygon, which is automaticallycalled when aninteractively

guided element collideswith thesensitive polygon. Wehave designedsensitive

polygonsprimarily to support theuser during thecoarse positioning phase of

an assembly task. Among others, the polygons can be used to constraint the

degrees of freedom for interactively guided objects. For instance, a sensitive

polygonpositioned attheportof aholecanrestrictthemotionofapintothe

axialdirectionoftheholeandthusconsiderablyfacilitatetheassemblytask.In

case of this classical peg in hole assembly task, MAESTRO creates a copy of

theguidedobject,orientatesthiscopyintothedirectionof thehole'saxis,and

switchestheoriginalobjectintoasemi-transparentrepresentation(seeFig.3).

Virtual Magnetism Inassembly procedures,itisoftennecessaryto position

objectsexactly,i.e.,paralleltoeachotheratarbitrarylocations,withoutleaving

anyspacebetweenthem. Sincein avirtualenvironmentthistaskis nearlyim-

possibletoaccomplishwithoutanyarticialsupportmechanisms,wedeveloped

theprincipleofvirtualmagnetism.TheleftpartofFig.4illustratestheeectof

virtualmagnetismonmovableobjects.Aswithsensitivepolygons,virtualmag-

netismisanessentialcomponentofthetheknowledge-basedmodelingofvirtual

objects'behaviour. In contrast to PBM and the other support mechanisms,a

(6)

Fig.4.Eectofvirtual magnetism(left), andforcecomponentsbasedonobjectdis-

tanceandobjectsurfaces

Beforesimulationstart,anobject'sinuence radiusis calculatedasafunc-

tion of its volume. Whenever two of such inuence areas overlap during the

simulation,anattractionforceisgeneratedthatconsistsoftwocomponents(see

Fig. 4). The rst component is calculated from the distance betweenthe two

objects'centers of gravity,and thesecond component isafunction ofpolygon

surfaces. Tocalculate thesecond component, asearch beamstarts from every

polygonmidpointintodirectionofthesurfacenormal.Ifsuchabeamintersects

with a polygon of the other object, a force directed to the surface normal is

created,whichisproportionaltothedistancebetweenthepolygons.Finally,the

sum ofthe to force componentsis applied to movethe smallerobjecttowards

thelargerone.

Snap-In AsallotherVR-basedassemblysimulationtools(seesection2),MAE-

STROprovidesasnap-inmechanismforthenalphaseofanassemblyprocess.

The snap-inis activated when position and orientation dierence betweenthe

guidedobjectandthetargetlocationfallsbelowaspecicthreshold.

5 The MAESTRO Prototype

Hardware Recently, we have accomplished a rst prototype of MAESTRO,

in which the force feedback system is a crucial component. We are using the

PHANToM 1.5fromSensableTechnolgies[7],allowingasixdegreesoffreedom

input andasimple force vectorasoutput. Due to therather small interaction

volumeof thePHANToM, we hadto choosean indirectinteraction paradigm,

where the pencil-like manipulator("gimbal") of thePHANToM isrepresented

(7)

data to and receivesforce data from the simulation computervia a 100MHz

FastEthernet.

Inprinciple,MAESTROcanworktogetherwithanyimmersivedisplaytech-

nology.IncombinationwiththePHANToM,atable-likedisplayismoresuitable

than a HMD or a fully immersive display like the CAVE. To accomplish the

evaluationofthesystem(seenextsection),wevisualizedthesceneonadisplay

consistingof ahorizontal andaverticalprojectionsurface("TAN HoloBench",

seeFig.5).

Tofacilitateinteraction, wehavefurthermoreintegrated speech recognition

thatallowstheusertopickupandreleasevirtualobjects,ortoactivatesupport

mechanismslikesnap-inorvirtualmagnetism,byafewsimplespeechcommands.

Sofar, MAESTROhas been implementedon aSun MicrosystemsE450 (2

Expert3D graphics boards) and on a SGI Onyx2 (1 Innite Reality graphics

pipe) workstation,bothcapableofdrivingaHoloBenchinstereomode.

4

5 1

2

3

6 7

2 6

9

8

Fig.5. Hardware setup of the MAESTRO assembly simulation system 1: Shutter

glasseswithreceiverofelectro-magnetictrackingsystem,2:Infraredemitterforshut-

tersynchronisation,3:PHANToMHapticDevice,4:Immersivedisplay"HoloBench",

5:PC withhaptics serverand speechrecognitionsoftware, 6:Stereoloudspeakers,7:

Microphoneforspeechrecognition8:Gimbal9:Graphicalrepresentationofthegimbal

withattachedvirtualobject

Software MAESTROisbasedonthecrossplatformVRsoftwareViSTA(Vir-

tualReality SoftwareUniversityof Technology Aachen[12]) and issubdivided

intothreeparallelprocesses:Thegraphicalprocessisresponsibleforprocessing

speechcommands,applyinginteractionmechanisms,andrenderingofthescene.

It is typically running at a frequency of 30 Hz. The PBM process processes

(8)

250Hz. Both processesrunondierentprocessorsof thesimulationcomputer

andcommunicatewith eachother viasharedmemory.Finally, thehapticspro-

cess runsonthePC atanupdate rateof1200 HzandcommunicatesviaUDP

protocolwiththesimulationcomputer.

6 Evaluation

InordertoevaluatetheMAESTROprototype,wehavecarriedoutexperiments

that are documentedin detailin [8]. Twoof these experimentsshall be intro-

duced and discussed here, with regard to eectiveness and user acceptance of

the articialsupport mechanisms,force feedback, and physically-basedmodel-

ing. 12 right-handed subjectsin theageof 24to 31participated at theexper-

iments. They were asked to complete the required manipulation tasks quickly

andaccurately.

6.1 Experiment1

The setup of experiment 1is shown in Fig. 6. Two large and two small nails

mustbebroughtintothecorrespondingholesofavirtualblock.Theexperiment

consistsof3phases:Inphase1,subjectshadtocompletethetaskina"native"

virtualenvironment,i.e.,basedonthe3-Dvisualizationofthescenewithoutany

furthermodelingorfeedbacktechniques.Inphases2and3theyweresupported

byguidingsleevesandsensitivepolygons,respectively.Everyphasewascarried

outtwice.Duringthersttrial,objectssnappedinasfarasthesubjectplaced

themintothetargetpositionbelowacertaintolerance.Thetrialendedwhenall

objectssnappedin,andthetimesubjectsneededforcompletionofthetaskwas

measured.Thesecondtrialwas interruptedautomaticallyafter axedamount

oftime, andthemismatchesbetweentheactualand therequiredobjecttarget

positionswere documented.

Results Theresultsofexperiment1showasignicanttimebenetwithguiding

sleeves in comparison to a native VE, and in turn a signicant benet when

using sensitive polygons instead of guiding sleeves.In trial 2, the positions of

the nails are signicantly more exact for guiding sleeves than for native VEs

(see Fig.7). The subjects' judgementof task diÆculty and quality of theuser

interface conrm thequantitativeresults(see Fig. 8). The snap-inmechanism

has great importance with regard to the judgement of task diÆculty. In case

thesnap-inisnotavailable,e.g.,becausethetargetpositionofanobjectisnot

known, subjects judge the manipulation task as morediÆcult. In such cases,

testpersonsregardthemechanismsofsensitivepolygonsandvirtualmagnetism

(9)

Fig.7.Quantitativeresultsof experiment1{ placingvirtualnailsintoablock.The

left diagram shows the average amount of time needed until snap-in, the right dia-

gramshows theaverage distanceto thetarget positionsixseconds afterstart of the

experiment.

(10)

Figure9depictsthesetupofexperiment2.Here,threevirtualelementsinformof

thelettersI,L,andT,havetobepositionedintoaframe.Again,theexperiment

consists of three phases. Inthe rst phase, subjects had to complete the task

in a native VE, where collisions between a letter and the frame are optically

signaled by a wireframe presentation of the letter. During the second phase,

thesubjectsweresupportedbyhapticfeedback.Finally,inthethirdphase,the

subjectswereaskedtomovethelettersabovetheirtargetpositionsandletthem

fallintotheframe.Thefallingprocedure,aswellasthebehaviouroftheobjects

whentouchingtheframe,havebeenphysicallymodeled.

Fig.9.Setupof experiment 2{start situation(left) of virtualletters(I, L, T), and

targetposition(right)inavirtualframe

Results A major result of experiment 2 is that the test persons had severe

diÆculties to complete the task in a native VE. They needed a dramatically

largeramountoftimethanwithsupportofhapticsorPBMtopositionletterI.

ForlettersLandT,itwasevenimpossibletocompletethetask(seeFig.10).An

interviewwiththesubjectsrevealedthat inanativeVE,theywerenotableto

comprehendwhyandwhereexactlyacollisionappeared.Asaconsequence,they

didnotchangetheirassembly strategyin awell directedmannerand, instead,

followed a trial and error principle, resulting in signicant longer interaction

times.

Fig.10.Quantitativeresultsofexperiment2{placingvirtualobjectsI,L,andTinto

(11)

sults.Both,hapticfeedbackaswellasPBM,gotapositiverating,althoughthe

testpersonsclearly preferredthe hapticsupporttocomplete thetask(seeFig.

11).Nearly allsubjectspraisedtheintuitivehapticrepresentationofcollisions.

Furthermore, they pointed out that the simulated gravitational forces had a

stabilizingeect ontotheir hand-arm-system.In adetailed judgement,haptics

andPBMwerenearlyidenticallyconsideredasimportant,realistic,helpful,and

non-distracting(seeFig.12).

Fig.11.AveragejudgementoftaskdiÆcultyanduserinterfaceforexperiment2

Fig.12.Averagejudgementofhapticfeedbackandphysically-basedmodeling(PBM)

7 Future Work

SinceMAESTROstartsfromthepolygonalrepresentationofascenewhichcan

easilybeexportedfrommostCAD systems,andsincemostparametersneeded

for objectbehaviour,haptics, andarticial supportmechanismsarecalculated

automatically before simulation start, MAESTRO is more exible than most

otherassemblysimulationsystems.Thisbenetofahighexibilitycomesalong

with the drawback, that MAESTRO canonly handle virtual scenes of rather

low complexity. Thecollisiondetection update rateis crucial to the quality of

physically-based modeling and the stability of haptics. Since in MAESTRO,

collisiondetectionandallmodelingarebasedonthepolygonalobjectrepresen-

tation,thesystemperformanceisverysensitivetothenumberofpolygonsinthe

scene.InordertomakeMAESTROapplicabletoassemblysimulationscenarios

thatarerelevantforindustrialapplications,weareworkingontheintegrationof

improvedcollisiondetection algorithmsandon acomprehensiveparallelization

oftheMAESTROsoftwarearchitecture.

TheMAESTROprototypemakesuseofaforcefeedbackdevicethatmerely

produces a force vector. However,for many manipulation tasks it is desirable

(12)

devicesareavailableandshouldbeintegratedintotheprototype,whichproduce

sixdegreesoffreedomforces.

Besidesgraphics and haptics, 3-Dacoustics could furthermoreimprovethe

realismofaninteractiveassemblyprocessandleadtoabetteruserperformance

andacceptance.Therefore,wejuststartedtoworkonanintegrationofbinaural

acousticstechnology into MAESTRO. Here,wefollowan innovative approach

thatisbasedononlytwoloudspeakers,producingastablebinauralsoundeven

whentheusermovesin frontoftheHoloBench.

A considerable extension of MAESTRO'sfunctionality is still necessary to

makeitattractiveforreal lifeindustrial applications. Future versionsof MAE-

STRO should allow bimanual interaction, the use of virtual tools like screw-

driversand wrenches,andprovideareal-time,realisticmodelingofdeformable

objects.Furthermore,itis highlydesirableto establishabidirectionalinterface

between MAESTROand product data managementsystems (PDMs)in order

to integrateVR-basedassemblysimulationintotheoverallproductionprocess.

References

1. Heger,R.:EntwicklungeinesSystemszurinteraktivenGestaltungundAuswertung

vonmanuellenMontagetatigkeitenindervirtuellenRealitat.Springer-Verlag,Berlin

(1998)

2. Gausemeier, J.,Grafe, M., Wortmann, R.:InteractivePlanning ofManufacturing

SystemswithVirtualConstructionSets.24thAnnualConferenceoftheIEEEIndus-

trialElectronicsSocietyIECON'98,Aachen(1998)2146{2151

3. Jung, B., Lenzmann,B., Wachsmuth,I.:InteraktiveMontage-Simulationmitwis-

sensbasierterGrak.ASIMMitteilungen49(1995)20{32

4. Jayaram, S., Jayaram, U., Wang, Y., Tirumali, H., Lyons, K., Hart, P.: VADE:

A VirtualAssembyDesignEnvironment.IEEEComputerGraphics&Applications

19(6)(1999)44{50

5. Buck, M.: Simulation interaktiv bewegter Objekte mit Hinderniskontakten.PhD

thesis,UniversitatdesSaarlandes(1999)

6. Gupta, R., Sheridan,T., Whitney,D.: Experiments onusingMultimodal Virtual

EnvironmentsforDesignforAssemblyAnalysis.Presence 6(3)(1997)318{338

7. Massie, T.H., Salisbury, J.:ThePHANToM HapticInterface:A Device for Prob-

ing VitualObjects. Symposium onHaptic Interfacesfor Virtual Environment and

TeleoperatorSystems,Chicago(1994)295{302

8. Stean, R.: Multimodale Interaktionstechniken fur die Simulation von Mon-

tagevorgangen in virtuellen Umgebungen.PhD thesis, AachenUniversity of Tech-

nology (2001)

9. Bara,D.:DynamicSimulationofNon-PenetratingRigidBodies.PhDthesis,Cor-

nell University(1992)

10. Bara,D.,Witkin,A.:Physically-BasedModeling:PrinciplesandPractice.ACM

SIGGRAPH1997CourseNotes,LosAngeles(1997)

11. Mirtich,B.:Impulse-BasedDynamicSimulationofRigidBodySystems.PhDthe-

sis, UniversityofCalifornia,Berkeley(1996)

12. vanReimersdahl,T., Kuhlen,T., Gerndt,A.,Henrichs,J.,Bischof,C.: ViSTA{

a multimodal, platform-independant VR-toolkit based on WTK, VTK, and MPI.

Referanser

RELATERTE DOKUMENTER

The posture variability of VirtualANTHROPOS and the implementation of different ergonomic categories, such as reach envelope, visual fields and force isodynes, comprise a very

❚ ®¯° ±B²³´ µw¶ ´ ²· A Digital Signal Processing Primer with applications to Digital Audio and

In this paper we present new tools for user interaction with virtual worlds, to bring more natural behavior into the manipulation of objects in virtual environments.. We present

In this paper, we propose a physically based model for the printing process in the virtual woodblock printing for improvement of printing quality.. Virtual printing is a simulation

Real-time simulation of elastic objects in Virtual Environments using finite element method and.. precomputed

Distributed virtual environments offer an efficient way for a number of users to create complex 3D virtual worlds.. However, navigation within these virtual environments can

component based development of interactive audio-visual simulation applications in the domain of VR/AR with par- ticular focus on virtual character simulation technologies. It

As a high-speed, serial communication link, Station Protocol supported transfer of data from multiple sensors (inertial, acoustic) and user inputs (buttons, joysticks,