Summary
Problem: The misperception of motion is a major problem for VR interfaces. It can cause nausea (the reason many VR designers completely avoid motion) and in driving and flight simulators users misjudge distances and become disoriented.
Objective: Test if adding physical motion cues to VR interfaces can improve users' motion perception.
Methods: Prototyping, Quantitative Analysis (Multivariate Testing)
Insights: We found that adding real rotations to a VR simulation didn't improve navigation in a visually rich environment (a city), while previous research found that motion cues did improve navigation in a visually poor environment (a field). This aligns with psychological research about spatial perception - the brain emphasizes information-rich sensory channels and ignores information-poor channels.
Impact: Our research suggests that physical motion cues are important for simulations where visual cues are poor (flight simulators), but less important where visual cues are rich (car simulators). These results were published in two peer-reviewed papers.
Prototyping
I worked on a team that prototyped a flexible VR interface that included a motion-tracked headset for visual information, fully spatialized sound, and physical motion cues like rumbling, seat movements, and 360-degree rotations.
Multivariate Testing
We tested users' ability to orient themselves by having them travel down five different curved streets, then asking them to point back to their starting position. The accuracy of their pointing was used as a quantitative measure of how orientation accuracy.
The task was repeated several times with and without physical motion cues. The impact of the physical motion cues on orientation accuracy was analyzed using the ANOVA (analysis of variance) statistical method.
On average, our users' sense of orientation wasn't significantly more accurate when physical motion cues were added to the simulation.
This contradicted a previous study that found physical motion cues helped with an identical task. The major difference between the two experiments was that users moved through a realistic city in our experiment, but moved over a grass field in the previous experiment. Research on sensory combination has found that the brain emphasizes visual channels when they are rich in information, but when visuals are lacking it relies on body cues from physical motion.
From this, we concluded that motion cues help orientation only when visual information is lacking - which is common in some settings, like flight simulators.
We found that physical motion cues are critical for VR simulations where navigation is important but visual cues are poor - common in flight simulators, naval simulations, and driving simulations in environments like deserts. In simulations with rich visual information, it may not be worth investing resources into physical motion cues.
We published our findings in the journal of Cognitive Processing and presented the results at the Institute of Electrical and Electronics Engineers' conference on Virtual Reality.
Publications:
Moving through virtual reality without moving? Cognitive Processing · Jul 1, 2012
Can physical motions prevent disorientation in naturalistic VR? IEEE Virtual Reality Workshops · Mar 1, 2012
The process demonstrated the value of negative results. When interpreted along with other research results from the field, our experiment yielded an important insight for the design of VR interfaces.
Using the quantitative pointing task was essential for our statistical analysis - to get a reliable signal out of the inherent noise between attempts at the task - but it made the task less realistic. In the future, I would work to make the task of measuring orientation more natural.