Biomedical Engineering Reference
In-Depth Information
(assuming that the center of mass is close to the Pelvis). It's thus possible to deduce
the instantaneous velocity and to drive a virtual camera in the virtual world in navi-
gation tasks [
35
]. However this approach has two main limitations. Firstly it assumes
that acceleration is noise-free which is not true with common accelerometers. As a
consequence velocity computed this waymay become false after amoment. Secondly
the initial velocity is required when computing
V
(
t
)
:
n
t
V
(
t
)
=
a
(
t
i
)
∗
Δ
+
V
(
0
)
i
=
0
where
is the initial velocity, n is the number of samples (
t
n
corresponds to
t
) and
a
stands for accelerations provided by the accelerometer. Any
error in estimating
V
Δ
t
is the sample time,
V
(
0
)
. One has to notice here
that users generally have a limited space to move whereas the virtual environment
could be very large. To solve this problem, one of the most famous solutions consists
in using an instrumented treadmill. In that case, forward speed could be delivered by
the treadmill while the other components could be given by the accelerometer.
An alternative consists in detecting footstrikes in the signal delivered by the
accelerometer and to deduce step frequency S
F
. If we assume that the step length is
constant and relative to the user's size, it is thus possible to deduce speed. The result-
ing speed is less noisy than integrating acceleration and is not subject to deviations
in time. However it does not provide accurate speed as the step length is not actually
measured.
Other systems such as the GAITRite (see
http://www.gaitrite.com
)
have been intro-
duced to measure the step length and frequency when walking along limited dis-
tances. It consists of a cable which is attached to the user's ankles and which length
is measured at a predetermined sampling frequency. It is widely used in medicine
because of its simplicity (especially no calibration is required). In addition to S
L
and
S
F
, the system returns the instantaneous distance between a fixed reference frame
and the two ankles. It is thus possible to analyse the longitudinal trajectory of the
ankles within the gait cycle. However, it is limited to straight line walking in a limited
space (generally a few meters).
Recording step length and step width is more difficult in curved walks. As
described in Chap.
3
these parameters are still difficult to define in a strict manner.
(
0
)
could thus lead to an error in
V
(
t
)
8.2.2 Curvature and Non-linear Walking
Retrieving the curvature of non-linear walking is still a complex problem. Indeed,
the instantaneous global orientation of the body is difficult to define. As explained
in Chap.
3
some authors focuses on the footprints, the orientation of the pelvis, the
torso or the head. It leads to different results for determining the global trajectory
of non-linear walking. Hence, positioning a unique sensor on the body to accurately
analyze the instantaneous orientation of the body in curved walking is still debated.
