Part V - Applications and Acknowledgements
- Human Tracking
- Clinical Gait Analysis
- 2D Classification System
- Time Dependant Values
- Application to Real Data
- Correction Mechanisms
- First Real Application
- Various Points
A. Human Tracking
When a person walks in a
desert-like scene, the lack of visual direction cues makes it more likely
that natural left/right differences in walking pattern will cause the
overall direction of travel to show regular, minor changes. The path often
shows as a large arc.
Since a person wandering even a short distance
can lead to relatively large direction changes, the analysis of how people
walk while wandering could develop into a much more significant factor in
the human tracking arena, now that direction changes over a single step are
Every possible 2D path characteristic can be
recreated using the Step Model, so it allows for the creation of correlation
tables to relate changes in one or more parameters to variations in field
available measurements, like step length (line), or distance and direction
from a standard point or line(s), etc.
This method will facilitate
the creation of a system to predict wander paths for lost people (path
deviation), given a few standard physical measurements and a current sample
footfall pattern. Since a person walking is a vector system, it's perfect
for a computer program.
The study of limb dominance, and its effect
on walking pattern, would be a large part of this application. But, how
parameters change with external stresses, such as fatigue or carrying
something heavy, would be the most substantial area, as well as correction
mechanisms (see below).
Co-operation with 3D gait analysts should
provide much of the data required to investigate path deviation, and there
could be a lot done with what’s already on file.
Unfortunately, the 4
points of gait (and foot-line), at the proper time point of heel-contact,
are not measurable from a real footprint pattern, since they’re in the air.
But, that doesn’t mean the system can’t be applied. At least now we know the
correct body segments and joints which are involved in distance and
direction when walking, and have an idea of their individual variations.
And, the direction changes over a step may be large enough were
estimated positions of heel-point pelvic joints are adequate. Since the
parameters are fundamentally related due to the physical connection of the 4
points of gait, muscular and/or skeletal limitations may make it possible to
limit the possibilities for each step, perhaps with error estimations.
This method is important to human tracking for 2 main reasons. It allows
the production of realistic footfall plots, in order to study distance and
direction relationships between footfalls which couldn’t be studied without
it, and it shows how the different body segments and joints contribute, via
the parameters, to the observed footfall pattern.
This means that
tables can be created which show how each parameter affects the position of
the footprint, so variations for each can be correlated in order to produce
a “most likely” Step Model for each step. A part of this is lab study of how
people walk, in order define limits of motion and show how each parameter is
related to the others.
The relationship of pelvic stretch and
straddle-line doesn’t have to be studied, since they’re the sides of a right
triangle with the pelvis-line as hypotenuse, but all others have to be
determined. The step-out-line and rear-leg-line, for eg., have a
relationship through ground contact and the pelvis-line. How one changes wrt
the other should provide greater insight into basic walking controls, and
help determine what simplifications can be applied, for analysis.
Application to SAR should involve 2 main areas, one of which requires real
The first area involves 2 computer programs and a
database. The initial program would be for creating Step Models from
specific input, taking relevant measurements on the Model, using the Model
(with any others) to create footfall plots, taking relevant measurements on
the plot, showing the acquired data via figures and graphs, and tabulating
the data in a standard format.
The stage 2 program would use the
previous for the base, but be enhanced to facilitate SAR in the field (see
Foreword, Application to SAR - One Plausible Scenario). This would require
the input of physical data and it’s use to estimate Step Models, the
plotting of wander paths, the evaluation of terrain if such data is
available (popular areas could have a detailed topographical map), the
calculation and display of potential error path regions, allow input of
field determined factors like rockiness or moisture, create logs of input
and all analyses, etc.
The world database would be to tabulate real
The second main area of application is the study of normal
walking patterns, as well as changes in the parameters due to specific
stresses (such as wearing a heavy backpack) and limb dominance. This would
require the definition of a standard method to determine limb dominance
(which there currently isn’t, I believe), as well as the analysis of
parameter data. This requires technical data input, such as 3D marker
Part I Part II