Virtual Reality:
Past, Present, and Future
Enrico Gobbetti and Riccardo Scateni
Center for Advanced Studies, Research
and Development in Sardinia
Perspectives on AR: Evolutionary
Build a better UI
Desktop is the primary interface mode right
now
Good for documents and flat stuff
Not so good for 3D input or output
Humans are well designed for a 3D world
Taking the interface into 3D will benefit
certain classes of applications
Perspective on AR: Revolutionary
Presence simulation itself is an interface
metaphor
Build a new reality, providing a new way to work
Virtual prototyping and walkthroughs
Simulation based training
Telepresence and teleoperation
Generalized augmentation of everyday reality
Change the way the work is done at a very low
level
Short History
60’s-70’s: Origin with teleoperation systems
1980’s: MIT creates first limited 3D workspace
1980’s: NASA picks up work, with focus on
teleoperation
1989: Autodesk and VPL Research demo VR in
trade shows
VPL founder Jaron Lanier coins “Virtual Reality”
… and the rest is history
Necessities of AR/VR
Multiple/integrated input and output modalities
Functional fidelity: sensory cues must match task
Responsiveness
Affordances: maintain affordance needed to
create illusion of presence in the perceived
reality
Appeals to mental representation: recognizable by
users through real-world metaphors
Requirements of VR/AR: Input
Simulated presence requires allowance for all
normal modes of interaction with reality
Full body gesture/position/locomotion sensing is
the holy grail
In reality, only a few of the possible channels are
captured
Gesture and voice come in second to position
Guesture/locomotion is sampled at critical
points and rest of data is interpolated
Output: Visual
Human eyes have a horizontal FoV of ~180
degrees without motion
The sensitivity drops greatly, but peripheral
view is needed for sense of realism
Maximum accuracy id defined by size of fovea,
with resolution of arc minutes (°)
Minimum motion update rate is ~15 Hz, below
which the brain no longer interpolates motion
For high motion content simulations, >60Hz is
often necessary
Output: Aural
Hearing is not as privileged as vision, but
necessary for situational awareness
Dynamic sensitivity is max between 1 and 4 kHz
Azimuth (left vs. right) perception is generally
good to one degree
Monaural levels are used for amplitude detection,
which is often mapped to distance using rate of
change
Output: Haptic
Haptic senses are combined sensory and motor
systems
Basic idea is force feedback
Generally, there is a lack of techniques for
simulating the haptic output without having a real
object
In absense of this, stylized gestures are often used
to replace fine haptic interaction
Update rates > 1kHz is needed to maintain realism
Key Issues: Synchronization
VR/AR often causes motion sickness
Key to this is the fact that one or more sensory
subsystem is producing data which is not in
synchronization with either the other sensory
systems, or not matching previous experience
with the same conditions
Each sense has its own detection threshold
Overall tolerance of variation is generally very
low
Hardware issues
Covered in other presentation
Guidelines/Comments
Iterative/user-centric design is key when using a
new methodology
There are no toolkits to abstract away the UI
issues
Synchronization of this sort requires parallel
programming techniques
Latency is especially jarring and should be
carefully considered
There are new problems such as occlusions
2D display techniques often do not apply
This is still a lot about technology and basis
Set reasonable goals for the points to be made
today
Virtual Reality:Past, Present, and Future
Note
Perspectives on AR: Evolutionary
Perspective on AR: Revolutionary
Short History
Necessities of AR/VR
Requirements of VR/AR: Input
Output: Visual
Output: Aural
Output: Haptic
Key Issues: Synchronization
Hardware issues
Guidelines/Comments
