Biomedical Engineering Reference
In-Depth Information
the user wears several accelerometers. From the relative acceleration of the limbs a
model of the skeleton of the user can be built. However, over time, the position of
the user will drift if only the accelerometers are used to calculate his location.
Although we have already seen some crossover, we distinguish position trackers
from the sensors already discussed because they return an absolute 3D position in
a fixed coordinate system. That is, if the tracked device is moved and returned to
the same position, the tracker reports the same position up to its own tolerance for
precision and accuracy. This is unlike an accelerometer that will drift over time.
Position trackers are a well-studied component of augmented reality and virtual
reality. The technology changes relatively slowly, so previous surveys [
21
,
41
,
44
]
give good overviews that are still very relevant. We thus detail the most common
technologies in use and highlight their strengths and weaknesses.
For small spaces, the stalwart of the field has been the magnetic tracking tech-
nologies epitomized by the Polhemus Fasttrak. This can return six DoF of a sensor
that is attached to or embedded in a control device. A standard unit might track 1-4
such sensors, and a common set up would be to attach one to a HMD and embed a
second in a hand-held controller. Because the tracking is magnetic, there is no need
for the sensor to have a line of sight to the base station, as with optical tracking (see
below). Thus the trackers are commonly used in situations where occlusion is likely.
However, the sensors typically do not work over spaces of more than 3 m by 3 m (the
space is often less than this), and are affected by metal in the environment.
A cheaper, but more limited technology is the visual tracking systems that can
track the head, hands, or other body parts in a small volume. In particular, the Wii
Remote contains an IR sensor that can track a bar of LEDs and thus can estimate the
relative position of the WiiMote from the bar. This is used to allow direct pointing
at the screen. The Sony Move controller uses a camera near the display to track the
3D position of a large light source on the controller. Microsoft's Kinect uses a depth
camera to track the skeletons of one or more people in front of the display. Alongside
these three currently popular technologies, there are quite a few others that have the
aim of giving a limited range of direct movement control in front of the display (e.g.,
NaturalPoint TRACKIR, Logitech Head Tracker). Most of these track only position,
not orientation (although the Wii Remote and Sony Move controllers add pose and
movement sensors to measure orientation indirectly), and all of them assume that
the user is facing in the direction of the display.
In situations where larger tracking volumes, integrated position and orientation
tracking, and greater flexibility of movement are needed, high-end optical track-
ing systems similar to those used in motion capture are frequently used (Fig.
7.12
).
Common systems include Vicon, OptiTrak, ARTrack and PhaseSpace. The first three
systems use passive markers that are mounted on a device, an item of clothing or
a full body suit. The passive markers are retro-reflective, and the camera has an
infrared light source to illuminate the markers. The PhaseSpace system uses active
markers similarly arranged. An important part of the technology is that the system
tracks the positions of the individual points in 3D space, but does not itself track the
orientation of the marker. Thus multiple markers on a rigid or near-rigid object are
needed to determine the orientation of the object. Position and orientation tracking
