TNO Human Factors
Virtue: Overview of human factors experiments; concertation meeting 10-11 September 2002
Michael Holewijn (the Netherlands)
2Human Factors
Partners
• BT Exact Technologies: Management• Heinrich-Hertz Institut: Real-time
algorithms• Sony Europe: Hardware systems• Heriot-Watt University: 3-D computer
vision• Technical University Delft: Image processing• TNO Human Factors: Human Factors
3Human Factors
User centered design
Environment
Technical issues
Human Factors
4Human Factors
Background
• 5 user centered
experiments
• Experiments used a mock-
up with the features of the
Virtue system
• Goal: test user specs
described from literature
(D01) & contribute to
requirements final system
5Human Factors
Experiment 1: Effects of shared virtual background
Do subjects perceive a seam in the display?
• task: information sharing• measures: telepresence, information exchange,
cohesion, satisfaction, task performance
• Results: no difference between conditions
6Human Factors
Experiment 2: Value of stereoscopic visualization and motion parallax
• Goals: determine• if stereoscopy and motion parallax are recognizable?• How much freedom of movement is necessary?
• Delft University’s Panorama system
7Human Factors
Results: exp stereoscopic visualization and
motion parallax
• Motion parallax is assessed (48%) more often correctly
than stereoscopy (25%).
• Subjects have to have at least 31 degrees moving
space in order to notice a difference in motion
parallax.
Explanations:• People have experience watching 2D television and are
used to perceive depth based on other cues, such as
interposition.• Panorama system has a limited range of stereo-scopic
and motion parallax effect
8Human Factors
Experiment 3: Effects of delays / synchronization on subjective measures
• Goal: Determine user acceptance of desynchronized
audio-video signals by reducing the audio delay
• Conditions:• video roundtrip delay fixed at 700 ms• audio roundtrip delays: 300, 500 (audio ahead of video), 700
ms
• Measures:, information exchange, cohesion, satisfaction
telepresence, task performance
• `
9Human Factors
Results experiment “Effects of delays / synchronization”
• No effects of audio delays on any performance
measure
• With delays perceived telepresence was significantly
lower
10Human Factors
Experiment 4: Effects of deviations in selective gaze
• Goal: to provide information on the
gaze deviations allowed
• Gaze deviations toward A: 0, 5, 10
degrees• Gaze deviations B<->C: 0, 15, 30
degrees• Measures:
• presence• communication
B C
A
B C
11Human Factors
Results experiment “Effects of deviations in selective gaze”
• Deviations in gaze toward oneself lead to lower spatial
presence than situations without gaze deviations.
• A gaze deviation for the other two participants is often not
visible for the person behind the VIRTUE system. Large
deviations (up to 30 degrees) have no effect on perceived
presence or group communication.
12Human Factors
Design implications
• No seamless background is required to produce a
convincing impression of telepresence.
• Large gaze deviations (up to 30 degrees) have an effect
on perceived presence nor on group communication.
13Human Factors
Design implications
• Stereoscopy and motion parallax features do not have to
be taken into consideration for increasing telepresence.
People are used to using different cues to perceive depth.
• The round-trip delay for audio can be set up to 0.4 s
below that for video (audio may be sent up to 400 ms
ahead of video).
• Different set of tasks and more time pressure may show
beneficial effects of stereoscopy and motion parallax, and
set round-trip delay for audio to 0.3 s below that for video.
14Human Factors
• Goal: to investigate whether and how users adapt
their videoconferencing behavior over time• Conditions:
• 4 types of tasks (brainstorm, negotiation, choice, execute)• Face-to-face (FF) versus Videoconferencing (VC)• 4 times in 2-week intervals
• Measures of performance• task performance• communication structure (e.g., # turns, interruptions, pauses)• communication content (e.g., questions, answers, support)• subjective measures (e.g., presence, cohesion, workload)
Experiment 5: Long-term adaptation
15Human Factors
Conclusions
Performance• VC = FF• Adaptation effect
Structure• FF > VC: more
formal
communication in
VC• No adaptation effect
Content• VC > FF: more positive
expressions in VC• FF > VC: more questions
and fewer answers in VC
VC > FF VC = FF FF > VC
• Performance• Communication structure• Communication content
VC=video conf.FF= face to face
16Human Factors
Assessment of final VIRTUE videoconferencing system
User requirements & acceptance check (September-
October)