Biomedical Engineering Reference
In-Depth Information
these factors. The result, of course, is that the utility of full body-based sensory input
during navigation becomes much like users of a CAVE—full body-based sensory
input is available, as long as the user does not take more than a few steps. These space
constrains have been addressed in much the same manner as CAVEs. While users
may control their orientation by turning their head or body, linear movements are
relegated to a joystick control or accomplished by walking in place (e.g., [
35
,
104
]).
Altered HMD-Systems
A variety of methods have been employed to allow users to navigate through HMD-
based VEs in a way that provides sensory information from all modalities. Perhaps
the simplest of these methods is to increase the available physical space and untether
the user. If sufficiently large physical space is available (e.g., a gymnasiumor airplane
hangar), recent advances in motion tracking technology have made it is possible to
track user motion over much larger volumes with sufficient resolution and accuracy.
Optical tracking systems, for example, are now available with high resolutions that
can differentiate between distal objects, fast update rates, and on-board graphics
processing capabilities. Software advances have also made it possible to chain a
large number of cameras together. Likewise, it is possible to untether the HMD by
either transmitting rendered images to the user wirelessly (e.g., [
65
]) or by having
the user wear a high-powered but portable rendering computer (e.g., [
109
]). Our own
large-scaleHMD facility [
109
] for example, uses these approaches and provides users
more than 1,100m
2
of tracked space in which to walk. It is even possible to situate
users in large, open, outdoor spaces while wearing rendering and tracking equipment
in order to simulate very large VEs without sacrificing naturalistic navigation and
full idiothetic sensory feedback by employing a combination of inertial and GPS
position tracking (e.g., [
6
]).
An alternative, but much more common approach has been to employ specialized
hardware navigation interfaces. Indeed, a wide range of devices has been created
to permit navigation in a tethered HMD, including omnidirectional treadmills [
25
],
roller skates [
49
], unicycles [
78
], stepping platforms [
51
], robotic floor tiles [
50
], or
discs of ball bearings [
48
]. Many of these navigation interfaces can be used inter-
changeably with a CAVE or HMD display, with the important consideration that an
HMD will occlude users' view of the navigation device. Hollerbach [
46
] has written
a review of many such devices, along with their advantages and shortcomings for
virtual navigation, and the implications of permitting (as in a CAVE), or not permit-
ting (as in an HMD), immersed users to view the navigation interface device. In such
a system, natural gaits are possible, providing accurate proprioceptive and efference
information to the user. Inertial information, however, will be in conflict in such
cases as the user remains relatively stationary in the treadmill's center. Consider the
recently developed Virtusphere [
65
], for example. Untethered users walk inside of a
hollow sphere that sits on a base of rollers. Because the sphere has its own mass, it
will not stop, start, or change directions with a high degree of responsiveness, and
