Entre 2000 et 2003, un projet de recherche et développement financé par
le Laboratoire universitaire Bell (LUB) a permis de créer un environnement
logiciel distribué qui permet une interaction de personne à personne aussi
bien par l'audio, la vidéo, et la communication tactile, que par la
visualisation et la manipulation d'objets synthétiques.
En 2002-2003, notre travail a permis la réalisation d'une conférence
vidéo à trois personnes entre trois environnements d'immersion 3D
hétérogènes: un à l'École polytechnique (CAVE), un autre au Centre de
recherche en calcul appliqué (CERCA) (ImmersaDesk) et le dernier à
l'Université McGill (CAVE). Cet environnement de conférence vidéo utilise le
graphe de scène WLV et les outils de compression
d'images développés à McGill. Les recherches actuelles se répartissent entre
plusieurs sous-projets correspondant à des applications pratiques du concept
de vidéo-conférence en environnement immersif. L'application MolEdit
(éditeur de molécules) a ainsi été modifiée dans ce cadre et nous
développons également une application indépendante où les participants se
retrouvent dans une salle de conférence virtuelle et peuvent interagir sur
des objets géométriques issus de modeleurs. Dans cette phase du projet,
nous cherchons maintenant à perfectionner l'interface entre l'humain et
l'environnement virtuel ainsi qu'à mieux définir les protocoles de partage
de scène et d'interaction avec les objets.
A Distributed Shared Visualization Environment
Jeremy R. Cooperstock and Benoît Ozell
Centre for
Intelligent Machines and
Centre de recherche en calcul appliqué
Goal
The goal of this project is to create an environment that simultaneously
supports distributed computer-mediated human-human interaction through
audio, video, and possibly haptic communication, as well as shared
visualization and manipulation of synthetic objects.
Objectives
The Distributed Shared Visualization Environment is a project that aims
to explore the challenging research problems associated with distributed
computer-mediated human-human interaction. The project brings together
researchers with a wide range of backgrounds in digital image processing,
audio processing, music, haptics, human computer interaction,
communications, teleoperation, telepresence and engineering. The focus is
not on a single problem domain (e.g. office environments), but rather, seeks
to create a general-purpose environment for collaboration and shared
visualization.
Shared Reality Environment
In the camera-monitor mediated world of videoconferencing, the limitations
of communications bandwidth and equipment capability tend to place a severe
handicap on the senses of sight and sound and eliminate the sense of touch.
As a result, even in state of the art videoconference rooms using the
highest quality equipment, the sense of co-presence enjoyed by individuals
in the same room is never fully achieved. Gaze awareness, recognition of
facial gestures, social cues through peripheral or background awareness, and
sound spatialization through binaural audio, all important characteristics
of multi-party interaction, are often lost in a videoconference. While many
of these issues can be addressed in part by improved display technology and
increased bandwidth, we believe that the result will still be inadequate.
To overcome these limitations, we believe that the computer must play a more
active role as an intermediary in the communications. Furthermore, it is
necessary to move from the restricted videoconference environments of
television monitors and stereo speakers to immersive spaces in which video
fills the participant's visual field and is reinforced by spatialized audio
cues. Haptic feedback should be introduced to help bridge the physical
separation of remote individuals. This feedback could range from reproducing
the floor vibrations in response to a user walking about to the tactile
response of a surgeon's instrument as it moves through different tissue.
Collaborative Framework in Engineering
In the field of computational based design and high performance computing,
engineers and designers in remote locations need to collaborate using
videoconferencing, but also need to exchange virtual (synthetic) models and
analyze simulation results. Simple simulation, visualization and analysis
tasks are usually fulfilled at the same place, on the same computer or on
computers connected together by a local area network. But as projects grow
and become more complex, high-performance computers are needed to solve
problems and high-performance visualization environments are needed to
analyze solutions. These facilities are often not available at the same
location.
In order for participants to maximize the effectiveness of their discussion
and analysis of simulation results, synthetic objects need to be created in
3D as if they were part of the videoconference scene. The participants
should be able to manipulate them as if real, to point out certain parts of
an object to others, and to identify different views on these objects for
others to see. These objects could be displayed on walls or on a table in
3D. Depending on the application, gesture and speech recognition may also
be integrated so that the virtual environment responds to various commands.
Benefits to End-User
This project will facilitate and enhance the human interaction in
multi-disciplinary design and analysis in engineering and scientific
application. It will encourage international collaborations using advanced
high-speed networks, enabling researchers to work together, whether their
colleagues live across the country or across the ocean, and to access
geographically-distributed computing, storage, and display resources.
The usefulness of this type of environment will be greatly enhanced through
its manipulative capabilities and functional links to current CAD and CAE
packages. Teams scattered around a country or around the world could work
together on the same project using the integrated collaborative framework
that is proposed. Using the infrastructure of videoconferencing, several
persons can work together locally or use global communication networks such
as broadband ISDN. This type of collaborative framework for distributed and
distance problem solving and analysis finds applications in other fields of
science, medicine, architecture, education, etc.
Methodology
The planned testbed consists of three small audio-insulated rooms, two based
at McGill University and a third at CERCA, each equipped with high
resolution video projectors, cameras, microphones, and multi-channel audio,
interconnected by an ATM switch. The video will be rear-projected to cover
three walls of each room, thereby encompassing the users' visual field and
creating the illusion of a larger shared space. Multi-channel audio will be
used to produce effective spatialization of sound sources, enhancing the
sense of co-presence. Powerful computer hardware will be exploited to
perform advanced, real-time graphics rendering, supporting three-dimensional
visualization through stereoscopic shuttered glasses.
The individual components of this project include:
- acquisition and translation of high-level object descriptions
from CAD packages (e.g. CATIA or STEP) and the implementation of a fast and
efficient object transport protocol
- development of audio/video transport protocols for high-fidelity,
low-latency communication, based on context-sensitive image capture and
compression
- real-time video blending, background removal, image mosaicing, and
perspective transformation of displayed people and objects in each room,
accompanied by audio spatialization of sound sources
- manual gesture recognition for object manipulation: image feature
selection using attentional mechanism, feature tracking, and development of
a suitable gesture recognition algorithms
- flexible integration of high-bandwidth audiovisual streams with
time-sensitive visualization data: synthetic object generation, rendering,
and manipulation in the context of high-fidelity videoconferencing,
development of a control interface for rendering and manipulation of
synthetic objects