When a person is walking, whether lost or not, with each step thereís a
change of direction. If the left and right side direction changes donít
cancel each other, the overall direction of travel (DOT) will change with
With outside cues, these direction changes are
constantly being corrected for, so we can walk a "straight" line to a
But, when there are no direction cues, these regular
direction changes can result in a person walking at 90 deg to the original
DOT in much less than 50 yds (cf. the football field experiment). It's the
reason that many subjects, when lost, circle the start point, or return to
it, even though they started off by heading directly away.
tendency to wander from the straight path is called path deviation (PD). The
main question has always been, "What causes the direction changes leading to
Previously, it was thought the most important factor was step
length, because it was believed that a difference in the step lengths for
the left and right feet caused turns. Since limb dominance affects, to a
large extent, the step length, it was thought that limb dominance controlled
PD by altering step lengths.
THAT DOES NOT HAPPEN. DOMINANCE MAY
AFFECT STEP LENGTH, BUT DIRECTION CAN NOT BE CHANGED BY CHANGING ONLY STEP
LENGTHS. There must be other reasons for the direction changes leading to
The search for these reasons resulted in this new gait
measurement system. It organizes, measures and can track many new parameters
which show not only the direction, but also the distance changes over a
single, isolated step, or an entire path.
Overhead snapshots capture
the positions of the 4 minimum points of gait (start-heel-point, rear-pelvic
joint, step-pelvic joint, step-heel-point) and the foot-line, projected onto
a 2D plane (usually the plane of the floor). The relationships between these
points and line define the 8 fundamental parameters, which are the basic
measurements of the system. These parameters separate distance and direction
changes, within a single step, as contributions from specific joints and
This method provides accurate definitions of such
basic values as step, stride, and carry lines, walking-straddle, etc., as
well as the accurate, separate determination of direction changes, within a
step, via the four parameters related to how people change direction: foot
angle, foot offset, push-off angle and aberrations.
change per step is the sum of these four parameters, and the exact
characteristics of the path depends on how each is being used in overall
direction control, besides how the other 4 linear parameters; rear-leg-line,
pelvic stretch, straddle-line, and step-out-line, are being manipulated.
And, there could be a great deal more information available within the
system. It currently shows a few standard reference points and lines, like
the pelvis direction and reference-foot model, but how these relate to
forces, momentum, the paths of other points and lines, etc. is currently
unknown. They may represent limits of ranges, or have other standard
relationships. Also, time analysis of the various parameters may show
periodicity with respect to one or more of the others, or with some other
These are points to be discovered. This area is in its
This version removes all the simplifications used in the
The applications to formal gait research are far too
numerous to be outlined in a single scenario, but to illustrate a
significant potential application in Search and Rescue (SAR), please
consider the following:
Application to SAR - One Plausible
10 yr old girl, healthy. On a walk, she passed through
several fields and wooded areas, on a faint trail, and then became lost.
Tracker team was called in, but with the inclusion of a path deviation (PD)
team of two or three people.
Both teams set to work. The tracker team
follows the tracks, analyzes them, interprets sign, and follows the trail as
far as possible. They determine that she went down a small hill to look at
flowers (probably), went out to about the middle of a small field and then
set off in the wrong direction. The rocky terrain didn't reveal anything
While the tracker team does its work, the PD team quickly finds
a series of footprints from the girl which appear to be a normal walking
pattern without unusual stresses, this was supported by the tracker team's
previous observation when they spotted them on the way through. They set a
reference point, and measure the distance to the heel-point of the first
footprint. The line from the reference point to the first heel-point is set
as 0 deg, and is the reference line for all angular measurements.
two measurements required for each footprint are: 1) heel-point to reference
point distance, and 2) angle from reference line.
At least three
consecutive strides of each foot are included, if possible, but we'll take
whatever we can get. Using laser sighting equipment, this would only take
A standard cut-out the size and shape of the girls
shoe print (or footprint, if not wearing shoes) is chosen, which shows the
heel-point and the foot line. This cut-out is placed over each print,
accurately as possible, and fastened.
I believe a practiced unit
could set this up in under 5 min.
Now heel-point measurements are
taken and recorded on a laptop.
For 3 strides each, that's fifteen
measurements taken by two or three people. Start to finish, no more than 15
min. And, we're not taking away from the tracker part of the search, they're
still on the trail, but will lose it soon in this scenario. (Another type of
measure would be the distances from perpendicular horizontal and vertical
lines. Either measurement could be used by the computer.)
was a computer program to do the plots and calculations, these are the only
field measurements required. Let's assume we have the program already. I'd
like to leave the discussion of the measures the computer would use for the
section on the development of the computer program. It's actually a bit
involved, and would only confuse this issue. But, it's ultimately only
simple plotting and standard measurements. Getting a good, graphical user
interface will be the hardest part of that program.
Now, we race to
the tracker unit, to be ready when needed. They're in the middle of the
field, and have lost the trail. The last tracks suggested the girl had
turned around in the field several times, then headed off in the wrong
But, then nothing.
We come in. The trackers tell
us (or we observe) that the girls last tracks didn't show unusual signs of
stress, injury, etc. so we go with whatever standard input is needed for a
girl of that age and size, etc. Stress factors, etc. The previous
measurements and any physical data were used by the computer program to
determine ranges of possible values for straddle length, step length, foot
angle and foot offset for each step, and uses iteration to further narrow
the potential error.
It produces a step model, using the real
measurements, and displays it to help us visualize (but, we don't have to
see it). The step model is kept in a log, and the step models from each new
data input are kept and compared to help see injuries, or predict larger
scale or periodic patterns of deviation.
The program predicts that
the girl will have an average 2.2 deg angular deviation per stride, to the
left, under the input conditions. With 2.6 deg left from the left foot and
0.4 deg right from the right.
A topo map of the area has been input
to the computer, and our current position marked. We input the start point,
and the initial DOT of the potential wander path start. The computer takes
into account standard effects of grade on wander paths (from tables
determined in the lab, and by continuous field observation and measurement.)
and puts a black line on the topo map for the most probable path, with areas
of different shades of gray around the line to indicate different levels of
potential error. The tracker team fans out across the most probable wander
path, spanning a little farther than the areas of potential error, and look
for sign to the entry point of the woods, then fan out along the fringe.
The black line suggests the girl would leave the field, and enter the
wooded section 65 deg to the left. The tracker team searches the fringe with
the suggested point of entry as the center point. It's 10 yds to the right
side, but they find sign, but it was quicker than without the calculated
start point. I believe the method doesn't have to be 100% accurate in order
to be very useful. The important thing is to be as accurate as possible,
while still having a useful, useable field system.
The girl's entry
point into the woods is marked on the topo map. The computer keeps track of
all points were sign is found, and the path of any tracks are marked on the
topo map. (Later, as many of the tracks as possible are examined and
measured for post-rescue path analysis.)
The saga could continue.
There are a series of prints, indicating the girl crossed the wooded
section, starting in a certain direction. Trackers follow the trail about 30
yds, shows her gait is a bit erratic. But, then nothing again. The path
deviation team comes in, determines new input, with the tracker team's
observations of the tracks and estimates of potential stresses, etc. This
time, the tracker thinks the girl is getting tired, so we increase the
fatigue value to 8.
Also, the obstruction factor is changed from 2 to
7 for the density of the wooded area, and the ground moisture factor from 2
We predict the path and exit point from the woods, and the
search for sign is started again. Etc., etc.
Whenever tracks are
found, the PD team sets up a grid and takes measurements. The position of
the tracks is marked on the topo map, so the step model log can be related
Trackers find another footprint. The position is marked
on the topo map. Again, it's to the right of the predicted path, so we add a
"field deviation" factor, to try to account for currently unknown factors.
When new potential paths are laid out, the computer shows both the
calculated path, with error estimate, as well as a blue path taking into
account the observed field deviation factor.
At one point, the girl
slides down a hill and loses her shoe. We get an idea of start point and
direction from the trackers, input the new "no right shoe" factor, and get
another potential wander path.
At another, we hear what sounds like a
highway, though is only wind in the leaves. One or two trackers go toward
that sound to check that possibility, while the rest stay on the calculated
You can see how it should be used as an integrated part of the
greater search effort, rather than a stand alone method for finding lost
people. (It is a stand-alone method for understanding general walking
asymmetries, with respect to distance and direction, though.) All parts of
tracking must work in concert.
And, the factors that are being input
to the program were determined in the lab, but with constant adjustment for
real world conditions, when new information is available. This method would
always be subject to refinement until we finally get something that works
Once the child is found, we start to go over what happened
during the search. Since all the data is plotted on the topo map in the
computer, we start to go over it to try to understand the reason for the
deviation from the calculated path, so we can better understand the effects
of various factors and refine error estimates. Maybe we go back into the
field as well, and try to correlate deviations with larger features, like
scenery or smells.
Maybe a new factor could be discovered, since now
it can be specifically investigated.
This new gait measurement system
can be used to set up a database of walking characteristics, and will
greatly aid in the evaluation of potential path deviation patterns.
Finally, this system should have been created and adopted long ago,
before 3D was possible, and before any detailed knowledge of muscular
control, forces, momentum, etc., well before video. It may be more difficult
to introduce out of phase like this, but once it's adopted it will allow,
among many other things, the creation of a consistent and directly
comparable world database of gait characteristics, since the technical
requirements are well defined, and relatively simple.
Part I Part II