distributed virtual environments for traiiining and telecsllaboration

8/3/2019 Distributed Virtual Environments for Traiiining and Telecsllaboration

1/4

Distributed Virtual Environm ents for Traiiining and T elecsllaboration

N.D. Georganas, FIEEE, E.M.Petriu, M. Cordea, and D. IonescuSITE .-School of Infona ation Technology and EngineeringUniversi~ty f Ottawa, Canada

A)bstract: This paper presents a brief review o f th e_ .Distributed Krtuul Environment (DVE) applicat,training and telecollaboration. The second partpuper discusses systems architecture and avatartracking aspects b r an experimental DVE.1.. Introductijon

The basic idea of Distributed Krtual Enviro(DVEs) is simplc:: a simulated world runs not 1comp uter system, but on several, usiing a series ofserver applications, [l ]. The computers are corover a network and people using those computlable to interact an d collaborate in re:al time, sharsame virtual w orld. Tele-immersive DVEs recogrpresence and miwements of indwiduals andtrack those imagzs, and then project them in rcmultiple, ge0g;raphcally distributed imenvironments where individuals can interact wilother and with com puter-generated models.Collaborative DVEs require an unconvenlbroad range of iietworking, database, graphics,modeling, real-lime processing ;and user ircapabilities.The interest generated by DVEs is reflectedmultitude of the research projects pursued all oworld, some of which are briefly presented here.Transvision: a Collaborative AugmentedTestbed of Sony Computer Science Lab., SorJapan, [2], is an attempt to use augm ented virtualfor collaborative desig ning . The user can secpalmtop video-see-through display a COIgenerated 3D model superimposed on the realview. The position and orientation of the disptracked by the system such that the computer-gemodel appears to occupy real space. Two 01participants can share the same cornputer modelwere reaI.

0-7803-5276-9/99/$10.000 939 IEEE

y1s inf' th ewner

nentsL onelient-ecteds areg the.e thejects,listic,:rsiveeachinallyworld:rfacen th e:r theealityCo.,ealityon am e r -worldy are:ratedmores i f i t

1847

Distributed Interactnve Virtual E nvironmen t (DIVE)developed by the Distributed CollaborativeEnviroinments group of the Swedsh Institute ofComputer Science, [3], is an internet multi-user DVEwhere participants navigate in 3D space and see, meetand interact with other users and applications. DIVEsupp0rl.s the development of virtual environments, userinterfaces and applications based on shared computergenerated environments. Applications include virtualbattlefields, spatial m odels of interaction. virtual agents,real-world robot control and multi-modal interaction.Social Virtual Reality Research, a project of theMitsubishi Electric Research Laboratory (MERL),Cambridge, USA, [4], focuses on the interaction amongpeople in virtual environments. It envisions grouplearning environments where people learn from eachother, from teachers, and by interacting with computersimulations.SHAVE, a project of the SRI International, U S 4 [ 5 ] ,is a multi-user virtual environment that allows severalusers on the Internet to interact within a shared 3Dworld. The SHAVE system provides a framework forremote collaborative work within interactive andeditable 3D worlds.ViItual Env ironment Technology Laboratory (VETL)of the {Jniversity of Houston and NA SA Johnson S paceCentre, USA, [6] , does research and developmentfocused on virtual environm ents for training, education,and scientiWengineering data visualization.NASA is exploring virtual reality technology to helpinternational crews of the International S pace Station toprepare for their missions The first demonstration ofusing a shared virtual environment across the Atlantictook pllace in 1995. Astronaut Bernard Harris at theNASAJJohnson Space Center in Houston, shared avirtual environment with the astronaut Ulf Merbold atthe Fraunhofer Institute for Computer Graphics inDarmaadt, G ermany.


2/4

COVEN (COllaborative Virtual ENvironments), aproject of the Advanced TelecommunicationsTechnologies and Services (ACTS) consortium led bythe Thomson-CSF Central Research Laboratory, [7],brings together a wide range of expertise oncommunication infrastructures, Virtual Environments,Human-Computer Interaction and VR animation toprovide support for a European distributed virtualenvironm ent. This project aims at developing acomputational service for teleworking and virtualpresence. The overall objective of the project is toprovide the facilities needed to support futurecooperative tele-working systems and to demon strate theadded value of networked VR or both professional usersand home users.DVEs are well suited for collaborative design andinteractive training of the industrial personnel forequipm ent operation and maintenance. Collaborativedesign work in VDE typically involves a small gro up ofusers, either synchronously or asynchronously, engagedin the construction, and manipulation of objects in thevirtual world. Since the interfaces for three-dlmensionalmodeling in VR are still relatively imprecise comparedto 2.5D CAD packages, most of the collaborative tasksin collaborative design involve evaluations of the design,and to a lesser degree, redesign or brainstorming fornew d esign possibilities, [SI.As an example, the National Computational ScienceAlliance (NCSA) and Caterpillar Belgium S.A. havedeveloped a system to allow remotely located engineersto work together on veh icle design review and redesign,[9]. Such a collaboration is needed for Caterpillarengineers in the U.S. an d Europe who jointly designCaterpillar vehicles so that they meet customer demandsand safety requirements for both m arkets.2. Experimental DVE

An experimental DVE is currently underdevelopment in the MCRLab at the Un iversity of Ottawaas a test bed for training an d telecollaboration, Figure 1.It will be used to study human-machine interfaces,object modeling and collaborative object manipulation,and the multimedia telecollaboration using Internet toolsand protocols, [lo]-[12 ]. Other issues of interestinclude: how participants should be represented in thecollaborative environment; how to effectively transmitnon-verbal cues that real-world collaborators so casuallyand effectively use; how to best transmit video and audlovia a c hannel that allow s both public addressing as wellas private conversations to occur; how to deal withdifferent time-delays and how to sustain a virtualenvironm ent even when all its participants have left.

The human users in a collaborative environmen t arerepresented by virtual alter egos called avatars. We areusing avatars based on muscle-based articulatedmodels of the human ace and body, [13]. Vision,force and tactile interfaces are used to provide theavatar-owners with a feeling about the interactionswhich their avatars are experiencing in the virtualenvironment, [141-[16].As an expedient approach to the 3D objectrecognition and tracking problem we are using apermanent pseudo-random encodmg of the visiblesurfaces of the of the avatar-owners. This encoding hasthe notable advantag e of a very compact encoding of theobjects requiring only one symbol of code perquantization interval. It is a technique well suited forcontrollable environments which allow for objectencoding and a priori mapping of all encoded objectmodels in a data base, [16]. The window property of thepseudo-random codes allows identfication of theabsolute coordinates of the encoded points on objectsfrom a relatively small subset of symbols recognized onthe object surface using monocular vision. Kno wing howdifferent surfaces of the geometric models of theencoded objects were originally m apped on the encodingarray it is possible to recover the identity and theposition, orientation and scale estimation parameters ofthe object containing the recovered window. Therecovered movements of these actors are u sed to builda KDB for the dynam ic behavior of the avatars evolvingin the DVE.We are also using a feature-based technique for thereal-time tracking of the human face position andorientation using vision, Figure 2. It consists of threesteps (i) extraction of a set of 2D features from a set ofmonocular images, (ii) tracking the found features fromone frame to another, and (iii) motion estimation. Th efeature detection is based on the shape and colorconstancy. An ellipse will search for the largest skin-blob, and classifl it as a face. The maximum size ofthe ellipse is obtained, by searching for skin inside,non-slan outside. The advantage of using a shape fordetection and further for tracking is that the tracker isnot attracted by skin-like background regions. Awireframe mesh is then fitted on the 2D face usingdetected features as anchor points and least-squareinterpolating for the rest of the geometry. The modelprovides the depth parameters of a Kalman filter used torecover structure from motion.Most of the models used in this DVE areimpleme nted in Java and VRML 2.0. Issues of causalityand multimedia synchronization, areas that we havesignificant know-how and experience, are also studied.Sharing was solved using the Java-EnabledTelecollaboration System (JETS) developed at theMCRLab, [171.

1848


3/4

k o r n p u t e r I Animat ion ~ j c dObjec t Shape &Behavior Models

Motion Tracking

FTRTIIAL SCENE / Th

,

6._ - -Pseudo-Random Encoding - -

1

II

iI

I

I

Figure 1. The architecture of the distributed viirtual environment.

1849


4/4

F i m e 2. Real-time trackingof the avatar-owner'sface position and orientation.

AcknowledgmentThis work was funded in part by the STENTOR NewMedia Fund.References[ l] R.C. Waters and J.W. Barms, "The Rise of Shared

Virtual Environments," IEEE Spectrum, Vo1.34, No. 3;pp. 18-25, March 1997.transvision.html;

[2] http://www. sl.sony.co.jp/person/rekimoto/[3] http://www.sics.se/dive/dive.tml;141 http://www.merl.codthreads/social/index.html;[5] http://os.sri.com/research/shave.html;[6] h t t p : / / w . vet1uh. du/overview/overview2. html;[7 http://chinon.thomson-csfG/projects/coven/;[X I J. Leigh, A. Johnson and T.A. DeFanti, "CALVIN: anImmersimedia Design Environment UtilizingHeterogeneous Perspectives," Proc. IEEE International

Conference on Multimedia Computing and Systems'96,Hiroshima, June 1996.[9] V.D. Lehner and T.A. DeFanti, "Distributed VirtualReality: Supporting Remote Collaboration in VehicleDesign," IEEE Computer Graphics and Applications,1997.

[101 N.D. Georganas, "Multimedia Applications DevelopmentExperiences", J . Multimedia Tools and Applications,

[I ll J.W. Wong, K.A. Lyons, R.J. Velthuys, G. v.Bochmann,E. Dubois, N.D. Georganas, G. Neufeld, M.T. Oszu, J.Brinskelle, D.F. Evans, A. Hafid, N. Hutchisosn, P.Iglinski, B. Kerherve, L. Lamont, D. Makaroff, and D.Szafkon, "Enabling Technology for DistributedMultimedia Applications", IBM Systems J. , Vol. 36,

[121 W. Robbins and N.D. Georganas, "Collaborative MediaSpace Architectures", Proc. CCBR'98, Ottawa, June

[13] H.J.W. Spoelder, E.M. Petriu, T. Whalen, D.C. Petriu,M. Cordea, "Knowledge-Based Animation of ArticulatedAnthropomorphic Models for Virtual RealityApplications," Proc. IMTC/99, IEEE Instrum. Meas.Technol.Con$, Venice, Italy, 1999.[14] G. Burdea,, E. Roskos, D. Silver, F. Thibaud., and R.Wolpov. "A Distributed Virtual Environment withDextrous Force Feedback, Proc. Infomatique '92,International Conference Interjiace to Real and VirtualWorlds, pp. 255-26, Nanterre, France, 1992.

[15] E. Petriu, W.S. McMath, S.K. Yeung, N. Trif, "ActiveTactile Perception of Object Surface Geometric Profiles,"IEEE Trans. Instrum. Meas., Vol. 41, No. 1, pp.87-92,1992.[16] E. Petriu, D. Ionescu, D.C. Petriu, S.K. Yeung, Ph.

Lavoie, N. Trif, "Absolute Position MeasurementApplications of PseudeRandom Encoding," Proc.ETM'96 IEEE Intl. Workshop on Emee en t Technol. fo rInstrum. Me as. , pp.119-126, Como, Italy, 1996.[17] SShirmohammadi and N.D.Georganas, "JETS: Java-Enabled TeleCollaboration System", Proc. EEEMultimedia Systems'97., Ottawa, June 1997

Vol. 4,No. 3, pp.313-332, 1997.

NO.4, pp.489-507, 1997.

1998.

1850
http://www/http://www.sics.se/dive/divehttp://www.merl.codthreads/social/index.htmlhttp://os.sri.com/research/shave.htmlhttp://w/http://chinon.thomson-csf/http://chinon.thomson-csf/http://w/http://os.sri.com/research/shave.htmlhttp://www.merl.codthreads/social/index.htmlhttp://www.sics.se/dive/divehttp://www/

distributed virtual environments for traiiining and telecsllaboration

Documents