PerceptualMotor Recalibration on a Virtual Reality Treadmill

1
Perceptual-Motor Recalibration on a Virtual
Reality Treadmill B. J. Mohler, W. B. Thompson,
S. H. Creem-Regehr, P. Willemsen, J. J. Rieser
H. L. Pick, Jr. University of Utah,
Vanderbilt University and University of
Minnesota
Introduction
Sarcos Treadport the Endless Hallway
Discussion
Perceptually guided actions require that the
appropriate scaling be maintained between visual
and proprioceptive information about body
movements. Rieser et al. 1995 demonstrated that
this scaling is adaptive and that recalibration
can occur rapidly as a result of changing
circumstances as a person interacts with the
world. Their methodology involved a test of
subjects walking without vision to a previously
viewed target before and after an adaptation
period with vision. The manipulation used in
Rieser et al. 1995 involved an adaptation phase
in which subjects walked on a treadmill in the
real world while experiencing visual flow
consistent with either faster or slower self
motion than the walking speed on the treadmill
belt. (The experimental apparatus involved an
exercise treadmill towed on a trailer behind a
tractor, thus allowing control of a subjects
motion through the world to be independent of
walking speed.) Durgin et al. 2002 replicated
this recalibration effect using a head mounted
display (HMD). The adaptation phase consisted of
walking up and down a hallway while wearing the
HMD, with visual flow adjusted to be either
double or half the actual walking speed. Verbal
distance judgments were not altered reliably by
the VE adaptation, while blind walking was
affected in the same directions as for Rieser et
al. 1995, though by a larger magnitude. Our
work had two goals. First, we aimed to determine
whether a treadmill VE could produce the same
effects as seen using the towed treadmill in the
real world or the HMD-displayed VE. Second, we
asked whether the recalibration effect created in
a VE would transfer to walking behavior in the
real world.
Our results show that a mismatch between visual
flow of a VE and walking on a treadmill will
recalibrate visually directed locomotion in the
real world. The effects are similar to those
found using real-world visual flow and treadmill
walking. These results are consistent with the
explanation that subjects spatially update their
representation of the environment as a function
of the learned relationship between visual flow
and locomotor activity. There was a larger
effect for the visually slower condition compared
to the visually faster condition. This asymmetry
is consistent with the small but significant
overshoot found in the visually same condition.
There are several possible explanations for the
visually-same effect. It is possible that
subjects perceptually compressed distances while
viewing the endless hallway in the VE. A
systematic underestimation in distance judgments
has been found in numerous VE studies using
head-mounted displays Loomis and Knapp 2003
Thompson et al. in press. Although we have not
directly tested distance estimations in the
treadmill VE, a compression of distance would be
predicted to lead to an overshoot in blind
walking, similar to the visually slower
condition. A second possible explanation is the
distinction between walking on a treadmill belt
and walking on the real ground. On the treadmill,
while the subject has information from their
motor and visual systems that they are moving,
they have conflicting vestibular and cognitive
cues that they are staying in the same place
Pelah et al. 1997. If subjects were to
calibrate to the perception of reduced
self-motion, then we would also expect to find
error in blind walking in the direction of the
visually slower condition.
Left Center
Right

• 3 8 by 8 screens
• 180 degree FOV
• 8 long by 6 wide walking surface
• Viewing was binocular.
• Translational position sensed body position
• Eye-height was fixed at subjects eye-height
• Frame rate gt 30fps.

Conclusions
General Method

• A mismatch between visual flow of a virtual
  environment and walking on a treadmill will
  recalibrate visually directed locomotion in the
  real world.
• These results demonstrate both the effectiveness
  of treadmill-based VEs in simulating the
  perceptual-motor effects of walking through the
  real world and the value of such systems in
  addressing basic perceptual questions that would
  otherwise be difficult to explore.

• Participants
• 24 University of Utah undergraduates
• 8 subjects participated in each condition

Results

• Results by Trial
• Possible variation due to the change from
  treadmill walking to real walking
• Important to look at more than one trial

• Procedure
• Pre-test
• Blind walking to targets on floor
• View a target, form a good image of the
  environment, and walk with eyes closed to the
  target
• 9 trials to 6, 8 10 meters

Percent Change (Post-Pre)
References
DURGIN, F. H., FOX, L. F., LEWIS, J., AND
WALLEY, K. A. 2002. Perceptuomotor adaptation
More than meets the eye. In Abstracts of the
Psychonomic Society, vol. 7, 103-104. LOOMIS,
J. M., AND KNAPP, J. M. 2003. Visual perception
of egocentric distance in real and virtual
environments. In Virtual and Adaptive
Environments, L. Hettinger and M. Haas, Eds.
Erlbaum, Hillsdale, NJ, 21-46. PELAH, A.,
DURGIN, F. H., MILLER, C. M., WASHINGTON, T. A.,
AND NELSON, M. 1997. Adaptation to running
depends on runner's frame of reference.
Investigative Ophthalmology Visual Science 38,
S1007. RIESER, J. J., PICK, JR., H. L., ASHMEAD,
D., AND GARING, A. 1995. Calibration of human
locomotion and models of perceptual-motor
organization. Journal of Experimental
Psychology HPP 21, 480-497. THOMPSON, W. B.,
WILLEMSEN, P., GOOCH, A., CREEM-REGEHR, S. H.,
LOOMIS, J., AND BEALL, A. in press. Does the
quality of the computer graphics matter when
judging distances in visually immersive
environments? Presence Teleoperators and Virtual
Environments.
Blind-walking to targets in real hallway
Percent Change

• Walk on the Treadport with one of 3 conditions
• Visually slower hallway (0.5X)
• Visually same hallway (1.0X)
• Visually faster hallway (2.0X)

• Visually Faster undershot 6
• Visually Same overshot by 3
• Visually Slower overshot by 11

Acknowledgments
Statistics Paired t-tests between pre
and post-tests visually slower t(7) 6.57 p lt
.01 visually same t(7) 4.34 p lt .01
visually faster t(7) -4.92 plt .01. There was
a significant difference between each visual
condition, F(2 23) 437 p lt .01 (p lt .01 for
all contrasts).

• Post-test
• Blind walking to targets on floor
• Same 9 trials as pre-test (same order)

This work was supported by NSF grants IIS-0080999
and 0121084.
bmohler_at_cs.utah.edu http//www.cs.utah.edu/bmohle
r
Apparent distance to target