P1: IXL/GEG P2: IXL
WL040A-22 WL040/Bidgolio-Vol I WL040-Sample.cls June 20, 2003 17:34 Char Count= 0
COLLABORATIVEVIRTUALREALITY 595Figure 3: Remote participants sharing collaborative virtual worlds. (Left) Users explore the structure of the human heart at much-
larger-than-human scale. (Right) The name tags that help to identify the generic avatar bodies are obvious.between the new user and one of the users already in
the space. This dialogue often deals with how to navigate
around and interact with the shared environment, with
the remote tour guide showing the new user around their
shared space. This way, the new user’s attention stays fo-
cused on the virtual, rather than the physical, world.
For international collaborations when English is used
as the default language, foreign speakers whose first lan-
guage is not English may find it difficult to converse nat-
urally, and hence these participants tend to be less vocal.
Although the tracker is able to transmit gross gestures, it
is harder to spot more subtle gestures. What is normally
considered a clear nod in the real world usually amounts
to a suggestion of a nod in the virtual world. Cultural dif-
ferences also impact the degree to which a participant
gestures. For example, Americans tend to gesture consid-
erably while speaking, whereas the Japanese tend to ges-
ture very little. In situations where collaborators are from
different cultures, it may be useful to include video to help
mediate discussions so that the faces of the participants
can be clearly seen.Audio and Video
Knowing where your collaborators are and what they are
looking at is important, but to really be able to collabo-
rate, a stable high-quality audio connection is needed. If
it is difficult to hear the other participants or if there is
much delay (latency) in the audio, then the collaboration
quickly breaks down. Unlike prerecorded audio over the
Internet, where the local computer can buffer the audio
and video for smooth playback, here the audio is live, so
a delay of more than 200 ms can hurt the collaboration,
and high variability (jitter) in the delay will hurt it even
more, in much the same way that a bad connection on
an international phone call can make a conversation im-
possible. When multiple people are in the space it can be
difficult to tell who is speaking if the audio does not come
from a particular point. Adding mouth movements, evencrude lip synching, to the avatars gives cues. Directional
audio, where the audio appears to come from a specific
location in the space, is very helpful, as the voice appears
to come from the speaker’s avatar. Audio that is quieter
with distance can be helpful if there are multiple groups
at different places in the virtual space performing differ-
ent tasks. By sending the participants audio over the net-
work, we are able to alter the volume based on that user’s
position in the space. Using 8-Khz audio, which gives rea-
sonable quality for conversations, the bandwidth required
is 64-K bits per participating site. To reduce the number
of encumbrances on the user, we prefer to mount high-
quality ambient microphones on the virtual reality hard-
ware itself, though these mikes can have problems in noisy
environments.
Sending video information requires much more band-
width, but it does help in settings where negotiation and
recognition are important. Mounting a camera (or sev-
eral cameras) on the virtual reality hardware aimed at the
primary user allows the computer to capture video of the
participant while in the space. Sending 15 entire video
frames of 720×486 RGBalpha data per second would re-
quire 160 Mbps, but much of the information that the
camera captures is not relevant. We may be interested in
the live video participants face only for mapping onto an
avatar at the other end. By reducing the area of interest
but maintaining 15 frames per second, we can reduce the
bandwidth to roughly 15 Mbps. We can reduce this further
through the use of hardware and software compression
tools, but the additional latency incurred through com-
pression and decompression can become distracting to
the user.Other Types of Data
Computers running the virtual reality display devices in-
volved in a collaboration need to ensure that the vir-
tual world remains consistent for all users. Application
data about the state of the virtual environment is also