Mathematics and Physics of Walking

Authors:
André Heck (AMSTEL Institute, University of Amsterdam)
André Holleman (Bonhoeffer college, “teacher in research” at the AMSTEL Institute)

A practical investigation task for pupils at upper secondary school level who have no
experience with COACH. Estimated working hours: 8.

© 2002 AMSTEL Institute. Further usage is only permitted with quotation of sources.

Preface
Coach is a computer learning environment that has been developed originally for natural
sciences (physics, chemistry, and biology). In this program you can easily collect data from
different sources, display data in graphical or tabular format, process data mathematically, and
so on.
Coach offers the possibility of collecting data from video clips. You can record yourself a
video clip or you can use an existing video clip for the purpose of carrying out practical
investigation tasks. You may think of studying the orbit, velocity, and acceleration of a thrown
basketball, investigating the motion of a car on a roller coaster, exploring the collision of two
cars, and so on.
Actually, we are going to use the data video tool for measuring human movements. You will
use existing video clips of walking, jogging, race walking, and running. Maybe you can record
your own video clips with a webcam, e.g., for the purpose of studying your own locomotion or
the movement of a fellow pupil. In short, you will play the role of a movement scientist.
The program Coach works with projects, each consisting of several activities. This practical
investigation task relates to the project Walking.

Reasons for Human Movement Research

Humans walk almost all in the same manner: gait is a periodic movement of each foot from
one position of support to the next position of support in the direction of progression. The
process of walking involves moving the center of gravity up and down and swinging of arms.
Yet, one can often recognize an acquaintance from a distance by his or her walk. Apparently
some personal characteristics are present in gait. This offers crime fighters the possibility to
identify thieves and hooligans running away from a crime scene by means of video recordings
of security cameras. Gait recognition could be a useful addition to individual identification
methods such as finger printing and iris recognition.
Researchers in biomechanics try to develop techniques for measuring individual gait
characteristics for authentication and recognition of persons from the way they walk One of
the issues is whether the movements of the human limbs can be described well by
trigonometric functions. Not every periodic movement can be described well by a sine graph,
but for (certain phases of) walking and running it often works fine. There are reasons for this.
Gait recognition is based on the mathematical model of curves that are built up from various
sinusoids. Here, the more scientific term is gait signature, which is based on the Fourier
representation of periodic motion. In this practical investigation task you will apply simple
mathematical and physical models and compare theoretical results with measurements.
Quite another motive for movement science comes from motion-picture industry. One can
often recognize a person’s state of mind from his or her gait and posture. The gait of a cheer-
ful person differs from the one of a depressed person. The emotional aspects of walking can
also be described experimentally in mathematical terms. Creators of animated cartoons and
science fiction movies have an interest in this: they often do not want to have robot-like
figures in their movies, but instead creatures that have emotion-like motions. Mathematics
gives them a helping hand.
Movement scientists also support athletes with their gait analyses. By measuring and
analyzing human motion and the forces involved they can give good advice to sportsmen in
order to improve the sporting achievements. For example, a small improvement in the
technique of a hurdler can make the difference between championship and loss.
Finally, gait analysis is popular because of medical importance of results: orthopedic
surgeons, kinesiologists, therapists, rehabilitation engineers, prosthetists and many others are
all involved in the treatment of movement disorders and are all interested in the modeling of
the gait cycle. The models can be used to assist diagnosis in disorders, to determine the effect
of prosthesis, to determine the success of particular rehabilitation techniques, and to determine
the effect of certain shoes and specific surfaces. Mathematical models are useful in the design
of such products.

The Learning Material

The explanations and the tasks can be read almost completely at the computer screen. We
have also provided them in printed form, so that you can choose the format that you find most
convenient. The computer tasks in the first activities are written down in so much detail that it
may give you the impression of drill and practice work. We have done this with the purpose
that you can familiarize yourself with video measurement as quickly and as independently as
possible. The mathematics and physics is mainly present in the biomechanical topic itself. By
the way, after this project you have learnt so much about Coach that you can continue to use it
in other work without much difficulty.

Report
Part of this practical investigation task is writing a report with a text processor about the last
three Coach activities. Herein we expect the answers to the exercises that ask for a reply,
sometimes short, sometimes a bit longer, but in any case clear and to the point. We also would
like to know which parts of the work you found easy or difficult. When you close an activity,
then COACH asks “Save changed result?”. Click on 'Yes'. In this way, you also get a digital
part of your report!
You can copy and paste tables, diagrams, and text from Coach into a Word document. You
are advised to start Word, in conjunction with Coach. Then you can copy intermediate results
like graphs, tables, and whatsoever into a Word document. You can decide later what you will
use of this in the final report.
Furthermore, we expect a summary of the time that you spent on various parts of the
investigation task (a sort of lab journal).

Table of Contents
Video measurement of walking ……………………………………………….. 3
Gait analysis……..………………………………………………………….….. 5
Arm movements during normal walking...……………………….………….… 8
Swing phase of the leg in low speed walking .………………….………….….. 13
The complete gait cycle of low speed walking ...………………….……….….. 15
More practical work on human gait …………………………………...………. 18

2
The ‘Walking’ Project
This project consists of five activities:
1. Video measurement of walking.
2. Gait analysis.
3. Arm movements during normal walking.
4. Swing phase of the leg in low speed walking.
5. The complete gait cycle of low speed walking.
The main topic of this project is the (mathematical) shape of human movements during
walking. We have included some extra practical work for enthusiasts at the end of this project.
In the first three activities the main purpose is to get acquainted with the possibilities of Coach
for investigating human movements in walking by measurements of video clips. You will also
use simple models to look at the collected data from a more theoretical perspective. In the last
few activities the self-reliance of investigating human gait is increased.

Activity 1: Video measurement of walking

In this activity you will get acquainted with some facilities of

Coach. Part of the work has already been done, as you will
see. In the next activity you will have to apply what you have
learnt here.

Start Coach, (log in as student; no password needed) and

choose the activity 1. Video measurement of walking.

Introduction
You see an activity menu bar, an activity toolbar, and four windows:
Upper-left: a video measurement of a magnified and sharpened video clip.
Upper-right: a window with the original video clip.
Down-left: an empty window.
Down-right: a text window.
In the upper-right window you see a short movie in which Mr.
Holleman walks along a building. This video clip has been created
outside with a webcam, which was attached to a laptop computer.
Every second 25 frames were recorded.
Play the movie to get an impression of the movement.

The movie clip has been magnified and sharpened with video editing
software to make measuring on the computer easier. In the upper-left
data video window you see a coordinate system, a light-blue indica-
tion of the scale labeled 1 meter, and data points (red and lilac
dots) tracing the locations of the right shoulder and hand over
time. By the way, one of the data points on the hand has been
deliberately misplaced.
The video clip consists of 32 frames, which are represented on the
video control bar, at the bottom of the data video window, by small
strokes.

In this activity you will use the tools of Coach to collect data
from digital video clips. Place results of your work only in your
report if there are good reasons for doing so.
We shall mostly pay attention to
- the graphical display of collected data;
- the linking between graphs and the video clip.

3
This introduction is in a text window. In addition there are other
texts with exercises and tasks. To see them, left-click on the
yellow 'Display Text' button in the activity toolbar. First choose
'2. Linking tables and graphs with a video measurement'. The shape
of the mouse cursor changes into a text symbol. Move the cursor to
an empty window and click in this window. In the same way you can
obtain the other texts windows. Actually, you don’t need these
windows because all texts are on paper as well. Choose what you find
most convenient.

Linking tables and graphs with a video measurement

1. Coach automatically places the coordinates of the measured points
into a table and it prepares a diagram (a data plot). To make the
table visible, press the 'Display Table' button in the activity
toolbar. Choose 'Holleman walks' and click in the upper-right
window (we do not use the original movie anymore).
If you right-click in the table window, you get a menu. Choose
'Display as a Diagram'. Because we do not need the introductory
text anymore, you can click in this text window.

2. You see four graphs in the diagram. Explain which graph belongs to
what quantity. Why is one red graph almost constant and the other
one increasing? Of the two increasing graphs, the red graph is in
turns above and below the lilac graph. Express what this means for
the gait.

3. Right-click in the diagram window or press the tool button of this

window (iconified by a hammer symbol). Select 'Scan', place the
cursor somewhere near the time-axis. Notice that the corresponding
frame appears in the data video window, that the corresponding row
is highlighted in the table window, and that a legend appears in
the diagram window from which it is easy to read off values of
displayed quantities. The legend can be dragged to any location in
the diagram window, for example to a location where it does not
hide parts of the graphs.
Drag the cursor in the diagram window and notice that the frame
displayed in the video clip en and the highlighted row in the
table change, too. By pressing ← and → on the keyboard (whilst
the data video window or the diagram window has the focus) you can
step through the frames of the video clip and the corresponding
points of the graph. In this way you can easily make the
connection between interesting points of the graphs and the frames
in the video clip.

4. Drag the cursor in the diagram window until it is located at the

time for which the data collection went wrong and the wrong
location of the right hand was measured. Move the video point to
the correct location and watch the result in the diagram and table
window.
In this way you can relocate more video points. (There are more
ways of accomplishing this as you will see later or find out
yourself.)
Stop scanning in the diagram window.

Collecting data from the video clip

1. In this activity we stopped measuring prematurely: on the last
four frames of the video clip, the location of shoulder and hand
have not been measured. You have to finish this yourself.
Start measuring by clicking the green 'Start' button in the
activity toolbar. The video clip jumps automatically to the first

4
 frame that has not been measured yet. Move the cursor across the
data video window towards the middle of the sticker attached to
the right shoulder and left-click this point. Next, move the
cursor to the middle of the piece of paper that Mr. Holleman holds
in his hand and click this point. Now you have collected the
coordinates of these two locations and the video clip moves all by
itself to the next frame for measurement. Repeat this until you
are done (the 'Start' button changes color from red to green).

Zooming
1. The x- and y-coordinates of the right shoulder and the right hand
during walking have been plotted against time in the diagram
window. Time t=0 has been chosen to correspond with the first
frame of the video clip.
Right-click in the diagram window. Choose 'Zoom to fit'. Notice
the red cross through the magnifying glass on the leftmost button
in the button bar of the diagram window. Press this button, which
we will refer to as the zoom-button. So there are at least two
ways to zoom in and out.

2. Another way of zooming is the following: drag an area to zoom in.

This area will be enlarged to fill the whole diagram. Zoom in a
couple of times on the graph of the vertical position of the
shoulder. By clicking the zoom-button in the diagram window you
return to the previous zoom state.

Leave the activity.

Answer 'Yes' to the question 'Save changed result?'.

Activity 2: Gait analysis

In this activity you will get acquainted with some

facilities that Coach offers. The central theme is
the determination of a curve that fits the data.

If necessary, start Coach and choose the activity

2. Gait analysis.

Introduction
In the movie clip, Mr. Holleman walks along a building. You see a
coordinate system, a light-blue indication of the scale labeled 1
meter, and data points (red and lilac dots) tracing the locations of
the right shoulder and the right hand over time. The video clip
consists of 32 frames, which are represented on the video control
bar, at the bottom of the data video window, by small strokes.

The graphs of the coordinates of the right shoulder and the right
hand have already been prepared.

In this activity and others you will learn that normal gait can be
described well with trigonometric functions. Here we concentrate
mostly on the movements of shoulder and hand during normal walking.

Some observations about normal walking

1. By watching the video clip and the corresponding graphs closely
you can already find out a lot of things about normal walking.
Some questions as a result of measuring the video clip:
What periodic movements do the legs make during walking and what
phases can be distinguished in the gait cycle?
5
 What can you say about the coordination of arm and leg movements?
How are shoulder height and phase of leg movement correlated?
How many meters per second does Mr. Holleman walk in this movie?
What is your estimation of the duration of the gait cycle, i.e.,
the time that it takes to go through a complete period of walking?
Use this to compute the step frequency, i.e. the number of steps
per minute. (Do not forget to explain what you mean by one step)

2. You can easily measure in the data video window distances and
angles with a computerized ruler and protractor. Right-click in
the data video window or use its tool button. Select 'Ruler'. Drag
the endpoints of the ruler so that they lay on the points between
which you want to know the distance. Measure the length of a
single step in this way.

3. From the measured stride length and the step frequency you can
compute the average walking speed. How many meters per second does
Mr. Holleman walk in this video clip? Is this in agreement with
the answer that you found before? What is this walking speed
expressed in kilometers per hour?

Practicing the creation of new diagrams

1. Right-click again in the diagram. Choose 'Create/edit diagram…'.
In the dialog window you see among other things the name of the
window, the option to display a grid, eight columns to choose
from, and at the bottom facilities to change the display of the
graph. Click C1. The 'Clock' is connected with the time of the
video clip. Look at columns C2, C3, C4, and C5. You will probably
recognize the coordinates of the video points. Make the vertical
coordinates of the video points invisible.

2. Click the yellow button 'Display Diagram' in the activity toolbar.

Instead of choosing an existing diagram, you will now create a new
diagram that will only display the coordinates of the right hand:
Press the 'New diagram' and give this window a sensible name (e.g.
'right hand'). Choose as first column the connection 'Clock' and
introduce as 2nd and 3rd columns the x- and y-coordinate of the
right hand, respectively. Click OK and select the diagram just
made: you have now two graphs in one diagram. Find out how you can
plot the vertical position of the hand against the horizontal
position. Do you recognize this graph?

3. Experiment in the diagram just made with the Coach facilities to

display data points with various markers and colors.

Determination of a curve that fits the data

1. Plot the horizontal position of the right shoulder in a new
diagram against time. Only draw the measured points and do not
connect them with straight lines.
Right-click in the diagram (or press the tool button, second
button from the left in the button bar of the diagram window).
Choose 'Analyse' and the sub-item 'Function-fit'. At 'Function
Type' you see that the graph of f(x)=a.x+b is drawn. The value of
a(=1,38) and b(=-9E-3) of the straight line are next to it. Press
the 'Auto fit' button. The straight line that fits the data best
is computed. What does a change of the value of 'a' mean in terms
of the line? What effect has a change of the value of 'b' for the
line?
If you press 'OK', then the straight line is added to the diagram
window. If you choose 'Stop', then the diagram window remains the
same. Write down the formula of the line and press 'OK'.

6
2. Plot the horizontal position of the right shoulder in a new
diagram against time. Only draw the measured points and do not
connect them with lines. Zoom in so that the graph is displayed as
large as possible. It is not a sine graph because the graph has a
tendency to increase in the right direction. What could be the
reason for this phenomenon? (Hint: measure in the video clip with
the 'Ruler' the difference in height between window and pavement
on the left- and right-hand side of the window).

3. We are going to do our best to find a reasonable formula for the

vertical position of the shoulder as a function of time.
Let us first determine the trend in the graph: do a straight-line
function fit of the data points. Add this line to the diagram.
Plot the graph of the difference of the measured vertical position
of the shoulder and the straight line, giving it its own vertical
axis. You can do this as follows:
- Right-click in the diagram window and select the menu item
'Create/edit diagram…'
- Choose an empty column and change the connection into 'formula'
- Choose 'Axis: second vertical'
- Click in the formula text region after 'Formula:'
- Press the 'Wizard' button (the button with the magic hat)
- Click on known quantities and mathematical operations to make
the correct formula.
- Fill out the name of the quantity introduced and (optionally)
specify the unit.
- Press 'OK' and zoom to fit by pressing repeatedly the zoom-
button until you have reached your goal of having the graph of
the difference displayed as large as possible.
If everything went right you now have a set of points that lie
pretty much on a sine graph. Use function-fit to find the best
sinusoid. The sum of this sinusoid and the previously found
straight line describes the original data pretty well. Let Coach
plot the graph of this sum to check this.

4. If you know the position of the right shoulder and right hand at a
certain time, then you can compute the angle that a stretched arm
at that time makes with the imaginary vertical axes passing
through the shoulder. Let us compute a concrete example before the
general case: in the diagram you can read off that at time t=1.2s
the position of the shoulder (P1) equals (1.66, 1.48) and that the
position of the right hand (P2) equals (1.93, 0.82). Positions are
in meters with respect to the coordinate system.
Use the opposite sketch to compute the
indicated angle θ .
Hint: First, compute the tangent of
this angle. Use this result to find
out which angle (in radians) corres-
ponds with this. Check your answer in
the video clip by measuring the angle
in the correct frame with the
computerized protractor (if desired,
maximize the data video window and the
clip for better reading).

5. Let us now try the general case: we denote the positions of P1 and
P2 as (x1 , y1 ) and (x 2 , y 2 ) , respectively. Verify that the following
formula holds:

7
 x 2 − x1
tan (θ ) =
y1 − y 2
In other words: you can compute the angle with the formula
 x − x1 
θ = tan −1  2  .
 y1 − y 2 
In the Coach activity is already a diagram window present,
entitled 'arm angle', which uses this formula (with arctan as
synonym of tan-1). Make this diagram visible. Measure the arm
angle in the first frame of the video clip with the computerized
protractor and check your answer in the graph of the arm angle.

Leave the activity.

Answer 'Yes' to the question 'Save changed result?'.

Activity 3: Arm movements during normal walking

In this activity you analyze the movements of arms

during walking at normal speed. You measure in the
data video window the angle of the line from the right
shoulder to the right hand with the imaginary
horizontal line passing through the shoulder. From this
you derive the arm angle, by us defined as the angle of
the stretched arm from the shoulder to the hand with
the imaginary vertical axis passing through the
shoulder. The arm angle is a function of time.

In this example you learn how to carry out measurements on digital video clips with Coach.
The instructions are very detailed and guide you through the example.

If necessary, start Coach and choose the activity 3. Arm movements during normal walking.

Open and play the video clip

1. Right-click in the upper-left window called 'Data Video' and
choose 'Open Video'. Select the video clip 'Walking'.

2. The video clip, which has been recorded outside with a webcam and
a laptop computer, shows a person who is walking at normal speed
along a building. Play the video by pressing the Play button in
the video control bar, at the bottom of the data video window.

3. The video clip consists of 42 frames, which are represented on the

video control bar by small strokes. By pressing ← and → on the
keyboard (whilst the data video window has the focus) you can step
through the frames of the video clip. Do this and notice that
there is something strange about frame 22: it is identical to the
previous frame. This means that, while recording the video clip,
at this point a frame has been missed (“frame dropped”). You must
take this into account, later when you collect data from the video
clip.

Calibrate the video clip

Before you can collect meaningful data from the video clip you must
calibrate the distances in the video clip. Quite often the creators
of a video clip have added to the first frame an indication of

8
dimensions. In this clip you must manage with the single remark that
the square paving stones all have a width of 50 cm.

1. Right-click in the data video window and choose 'Coordinate

settings…'.
ƒ Scale: because the vertical and horizontal scales are the same,
select 'Same scale in all directions'.
ƒ Origin: Because we want to study the arm movements with respect
to the shoulder select the origin of the coordinate system for
all frames separately, via 'First point clicked at each frame'.
ƒ Time Calibration: the video clip has been recorded with a frame
rate of 30 frames per second. This information is used to
connect the frame number with the time t (in seconds), as soon
as you have decided which frame corresponds with t=0. Leave the
current choice at 't=0 at first frame'. Press 'OK'. Note that a
new window, viz., The Scale Settings window, pops up.

2. Now you can start with the actual setting of the length scale.
Drag and resize the end points of the scale-ruler (red by default)
to match two neighboring paving stones along the imaginary
horizontal line where the footsteps are placed (why should you
actually do this?). Specify the scale length of 100 cm in the
window 'Scale Settings'.

3. Press 'OK' in 'Scale Settings'. Calibration is finished now.

Optional settings
Selection of the number of points per frame and their display
1. Right-click in the data video window and choose 'Points…'.

2. We want to measure 1 point per frame, viz., the middle of the

right hand. So, confirm the current setting by pressing 'OK'.

3. Go to 'Markers and Colours…' and choose your favorite colors and

markers for the video points (a blue big dot works fine).

Selection of frames
You fix which frames of the video clip you want to use for data
collection. There are four possible ways of selecting frames.

1. Right-click in the data video window and choose 'Frames'

2. If you want to collect as much data as possible, choose 'Use All

frames'. This is the default choice of Coach. But as you have
already noticed, frame 22 cannot be used. Also is the automatic
selection of a number of evenly spaced frames less suitable
because the right hand is out of sight after the first 37 frames
(you better check this beforehand).

3. In this example we want to collect data points on equidistant

frames from 1 till 36, omitting frame 22. For example, we can
choose frames from 1 till 36 with intermediate steps 2, i.e.
omitting each time a single frame. In other words, we choose frame
1, 3, 5, .., 33, 35. To do this, choose the option 'Select these…'
and enter the formula 1-36$2. With this formula you express your
choice of ‘frames from 1 till 36 with step size 2’. All selected
frames are black in the frame bar; all deselected frames are gray.

By the way, after your choice of frames to collect data from, you can always use the Insert key
to add more frames. With the Delete key, you can deselect frames for measurement.

9
Collect data
1. Start measuring by pressing the green 'Start' button. The color of
this button changes into red to indicate that you can stop measu-
ring at any time.

2. Move the cursor across the data video window and click on the
right shoulder. Recall that the coordinate setting have been
chosen such that the first point clicked in each frame is the
origin of the coordinate system for that frame.

3. Next, click on the right hand.

4. The video clip jumps to the next frame for measuring. Repeat step
2 and 3 for all selected frames.

5. If you want to stop earlier because you are of opinion that you
have collected sufficient data or because you want to make changes
to the settings, press the red 'Stop' button. But for a good
result of your analysis of the arm movements you better leave the
settings as they are now and collect data on all selected frames.

6. If you want to keep track of the measured points in the video

clip, select 'Trace' under the tool button (or by right-clicking
in the data video window). The marker of the video point will
appear.

Display of collected data

1. Right-click in the video-window and select 'Display as a Diagram…'
to plot the data points. The shape of the cursor changes into the
form of a small diagram. Click in the window where you want to
place the diagram.

2. In the diagram, the horizontal (P1X) and vertical (P1Y) Cartesian

coordinates of the video points are plotted against time. Recall
that that we are more interested in the arm angle, i.e., the angle
that the stretched arm makes with the imaginary
vertical line that passes through the shoulder.
The arm angle is a function of time. Luckily
you collect by clicking more than the position
of a video point. The distance from the video
point P1 to the origin O of the coordinate
system and the P1-Angle φ of the line OP1 with
the horizontal axis are also collected. See the
opposite figure. The arm angle is denoted here
with θ.
Change the diagram: plot 'P1-Angle' against
'time'. If desired, you can also change the
name 'P1-Angle' into 'phi'.
Note that the angle is measured in degrees and not in radians.
This is an activity option that has been set by the author of this
activity. Just have a look at the activity option via the menu
item 'activity options'. Trigonometric functions take this choice
of degree as unit into account: now, for instance, sin(90)=1.

3. Determine the arm angle θ as a formula in 'P1-Angle', add the

graph of the arm angle to the diagram, and make 'P1-Angle'
invisible.

10
Update video points
If you are not satisfied with a data point, you can go back to the
frame in the video clip that corresponds with it and update the
measurement. For better results you are advised to zoom out the
video screen.

1. Click in the data video window to put the focus on this window and
press once the arrow key ↑. The video clip is enlarged so that it
fits in the data video window. If this is not large enough, press
the 'maximize' button in the window (the 2nd button the right).

2. With the arrow keys ← and → you can step through the selected
frames.

3. Update a video point by dragging it to the desired location. The

coordinate system cannot be relocated: if it is misplaced, go
through the following steps:
- remove a measurement in a frame by pressing the Delete key;
- insert the frame again for measurement via the Insert key;
- update the measurement by pressing the green 'Start' button.

4. When you are finished, it is a good idea to minimize the data

video window again to its default size.

If desired, add extra data points

It is possible to add extra video points after the measurement has
been finished and in case not all frames have been selected yet.

1. Select a frame with a gray marker in the video control bar (hint:
you can go through all frames in the video control bar by the
'Page Up' and 'Page Down' keys). Press the Insert key. The
selected image becomes black on the video control bar

2. Press the green 'Start' button to collect data points for the
additionally selected frames. The data are inserted at the right
spot in the diagram and the table.

3. To delete a video point, select the corresponding frame in the

video control bar and press the Delete key. Try this.

Analysis of the arm movements

Analyze the arm movement. In mathematical terms: find a formula that has a graph that fits
well with the measured arm angles, i.e., that stays close to the measured angles.

1. The graph of the arm angle has the shape of a sine graph. Right-
click in the diagram window or click on the tool button of this
window. Select 'Analyse' > 'Function-fit' and try to find the best
sinusoid, i.e., a function of the form

angle(t ) = a sin (b t + c) + d .

2. What is in this diagram the duration of a complete period? Compute

the period with the formula from function fit? Use the foot
movement to estimate the period of the gait cycle. Are these
results in agreement? Actually, this should be the case. If not,
check your work.

3. The arm rotation is the difference in angle between the arm when
it is as far as possible in the forward direction and the arm when
it is as far as possible backward. How large is the arm rotation
11
 according to the graph? How large is the arm rotation according to
the function fit?

4. In which phase of walking is the person in the video clip when his
right arm is directed forward as far as possible? And when the arm
is as far as possible backward?

5. What is the state of equilibrium of the right arm for the walker
and at which gait phase is this state of equilibrium passed
through? What is the amplitude of the arm swing?

6. A standard topic in school physics is the harmonic motion

of a pendulum. The pendulum consists of a heavy weight that
is attached to a light rope and that swings to and fro. For
the period of a pendulum of length l, i.e., the time needed
for one complete swing, the following formula holds:
l
period = 2π ,
g
where g is the gravitational constant (g =9.8 m/s2).
The arm swing around the state of equilibrium resembles the motion
of a pendulum. Measure the length of the arm (including the hand),
and compute the period with the above formula. Is the answer in
agreement with the period of the arm swing that you determined in
a previous exercise? If not, can you come up with a reason why
this formula cannot be applied in this case?

7. In scientific research, the arm movement is sometimes modeled as a

harmonic motion that is driven by an external force. For this
model, there exists a ‘natural period’ that is given by the
following formula:
I
period = 2π ,
2m g d
where I is the moment of inertia of the arm (including the hand),
m represents the mass of the arm, g is the gravitational con-
stant, and d is the distance of the center of gravity of the arm
to the rotational axis, i.e., the shoulder. You can use this
formula, even when you do not know what the moment of inertia
actually means. In measurements with humans it has been found that
the center of gravity is located away from the shoulder at around
53% of the total arm length (hand included). The weight of the arm
is for men about 5% of the body weight. The moment of inertia I of
the arm with the shoulder as rotation axis can be calculated with
the following formula:
I = mk2 ,
where k is the so-called radius of gyration of the arm with the
shoulder as rotational axis; it is equal to about 64.5% of the arm
length. Compute the natural period for the arm according to this
model and compare the answer with the period of arm movement that
you have measured before.

8. Physicists agree about one thing of the periodic arm movement of

normal walking: the formula for the natural period can be written
as
cl
period = 2π ,
g

12
 where l is the arm length (including the hand), g is the gravita-
tional constant, and c is a constant that depends on the stature
of the walker. Determine the constant c for the walker in the
video clip. Compare this with the number that you expect to find
according to the model in the previous exercise.

9. In the first part of this analysis you have approximated the data
with a sinusoid. Determine the graph of the difference between the
graph of the measured data and the sinusoid. Find out whether this
graph of the difference has again the shape of a sinusoid. If this
is the case, then you can describe the arm movement pretty well by
the sum of two sinusoids. This would actually mean that the model
of a periodically driven harmonic motion is applicable. Draw the
sum of these graphs to see with your own eyes how well the formula
describes the arm movement.

Leave the activity.

Answer 'Yes' to the question 'Save changed result?'.

Activity 4: Swing phase of the leg in low speed walking

Gait is a periodic movement of each foot from one

position of support to the next position of support in the
direction of progression. One of the issues is whether
these movements can be described well with trigono-
metric functions. In this activity you focus on the leg
movement: you measure the angles of the thigh and
lower leg have with respect to the (moving) coordinate
system that has the knee as origin. You try to find
experimentally mathematical relations between the
quantities. We distinguish the swing phase of gait and
the complete gait cycle. This activity is about the swing
phase. If necessary, start Coach and choose the activity
4. Swing phase of the leg in low speed walking

Introduction
In the data video window you see a short movie in which a person
walks slowly in a study along two cupboards. The width of one
cupboard is 1 meter. Take perspective into account when calibrating
the video clip: place the scale-ruler of 1 meter on the imaginary
horizontal line that connects the places where the right foot is
flat on the floor (hint: use the parquet strips for a good
estimation). This video clip has been recorded with a webcam and
with a frame rate of 30 frames per second.

In this activity we shall see that the movements of the lower limbs
can be described well by trigonometric functions. We shall look
mostly at the right thigh and lower leg of the walker.

1. Play the movie to get an impression of the motion.

Measurement of angles during the swing phase

The hip angle (θ) and the knee angle (φ) are defined in the figure
below.

13
So, the knee angle is 0 degrees if the leg is completely stretched.
You can easily measure the angles α and β as functions of time in
the video clip. Just take the knee as the origin of the (moving)
coordinate system and click in each frame on the hip and right foot
(not only the positions but also the angles α and β are obtained).

1. Prove the following formulas for the hip angle and knee angle in
terms of the angles α and β: θ = α − 90 and φ = α − β + 180.

2. Select the frames in the video clip in which the right leg
completes a full swing, i.e., start with the frame in which the
toes get off the ground and end with the frame in which the heel
strikes the floor again. Measure the angles α and β for this time
interval and draw the graphs of the angle β, the hip angle, and
the knee angle (as functions of time) in separate diagram windows.

Analysis of the swing phase

1. Verify that the angle β and the knee angle can be described well
by a sinusoid. In case of the angle β, also check whether the
difference between the measured angle and the sinusoid can be
approximated once more with a sinusoid.

2. In the sinusoidal approximation of the angle β, what is the period

of the swing phase? Like in the case of the arm movements during
walking you can compare this period with the theoretical, `natural
period’ of a harmonic motion that is periodically driven by an
external force. The formula is now
k2
period = 2π ,
2gd
where k is the radius of gyration of the lower leg (including the
foot) with the knee as axis of rotation, g is the gravitational
constant (=9.8m/s2), and d is the distance from the center of
gravity of the lower leg to the rotational axis, i.e., the knee.
In measurements with humans it is found that k is equal to 73.5%
of the length of the lower leg (including the foot) and that d is
equal to 60.6% of the lower leg length. Use this formula to
calculate the natural period of the lower leg of the walker in the
movie and compare the answer with the previously measured value.

3. Verify that the hip angle θ during the swing phase can be
approximated reasonably well by the sum of a parabola and a
sinusoid. For those who wonder why we do not take a sum of two
sinusoids: look what happens when you do a sinusoidal function fit
for the hip angle.

14
4. Suppose that the thigh and the lower leg of the walker are of
equal length, i.e., OP1 = OP2 in the above figure. Hen you can
compute the angle of the line P1P2 with the imaginary vertical
line through P1. Let us call this angle the leg angle. Prove the
following formula: leg angle = (α + β) / 2 − 180 .

5. Let Coach plot the graph of the leg angle against time. Verify
that the leg angle can be approximated well by a sinusoid.

Leave the activity.

Answer 'Yes' to the question 'Save changed result?'.

Activity 5: The complete gait cycle of low speed walking

This activity is a continuation of the previous investiga-

tion in which you have measured the angles of the thigh
and lower leg with respect to the (moving) coordinate
system that has the knee as origin and in which you have
found experimentally mathematical descriptions of the
angles during the swing phase. This activity is not merely
about the swing phase, but about the complete gait cycle,
including the stance phase when the leg supports the
upper body. If necessary, start Coach and choose the
activity 5: The complete gait cycle of low speed walking.

Introduction
In the data video window you see a short movie in which a person
walks slowly in a study along two cupboards. The width of one
cupboard is 1 meter. The calibration of the video clip takes
perspective into account. This video clip has been recorded with a
webcam and with a frame rate of 30 frames per second.

In this activity we shall see that the movements of the lower limbs
can be described well by trigonometric functions. We shall look
mostly at the right thigh and lower leg of the walker.

Measurements of angles and distances

Again we define the hip angle (θ) and the knee angle (φ) as in the
figure below. The angles α (P1-Angle) and β (P2-Angle) have already
been measured as a function of time by taking in each frame the knee
as origin O of a (moving) coordinate system. We have selected the
frames in the video clip for measurements in which the right leg
completes at least one gait cycle. We start with the first frame in
which the right foot is flat on the floor and the lower leg is
perpendicular to the floor, and we end with the frame when this has
happened once more. The time calibration has been chosen such that
t=0 coincides with the beginning of our segment of frames. The
measured angles have already been plotted against time in the
diagram window. The lower leg angle (ψ) is defined as ψ =270–β.

15
1. Draw the graphs of the lower leg angle, the hip angle, and the
negative value of the knee angle (as functions of time) in
separate diagram windows. The reason that we consider the lower
leg angle and –knee angle is because we want to compare our
measurement results with results from the literature in human
movement science.

2. The following picture describes the various phases in human

walking and their relative duration. This picture has been taken
from Rose & Gamble, “Human Walking”, Williams and Wilkins, 1994.

Use the video clip and your graphs to measure the duration of the
phases in the walk of the video clip. Figure out the duration of
the gait cycle, the duration of double support, and anything else
that you find noteworthy to measure.

3. Measure the stride length and the length of the right leg of the
walker in the video clip. What is the ratio between stride length
and leg size of the walker for the motion shown? If you have not
done this before, measure the duration of the gait cycle. Calcu-
late from the measured data the average speed of the walker.

Analysis of the leg movement

1. Compare your graphs of the hip angle and lower leg angle (ψ) with
the following graphs that have been taken from Yam et al, Gait
Recognition By Walking and Running: A Model-Based Approach. ACVV
2002: the 5th Asian Conference on Computer Vision, Melbourne,
Australia.

16
 What similarities and difference do you notice?

2. The mathematical model that the authors of the afore-mentioned

conference paper use is the same as we have applied before: a sum
of two sinusoids. Apply this model to the hip angle and the lower
leg angle that you have measured.

Function fit in practice: some hints

If you select in function fit the formula of a sinusoid, then it
could happen in this investigation task that you do not find the
best fit by pressing 'Auto-fit'. The reason is that Coach takes
the currently drawn sinusoid as a starting point in a numerical
process to a new fit and it may end up in an unwanted curve. The
software is not to blame here, but the mathematical algorithm
fails. The only remedy is to choose manually a good starting
point.
The screen dump shows a manual
fit in action. The formula is
a sinusoid; the measured knee
angle has been plotted against
time. First we have dragged
the original sine graph such
that its maximum coincides
with the maximum of the measu-
red data of the knee angle. By
left-clicking we have fixed
the pin. Now the sinusoid can
be deformed into another
sinusoid with the same maximum
by dragging some point of the
sinusoid. As soon as the sinusoid is close enough to the measured
data the 'Auto-fit' button can be pressed to set Coach to work for
finding a good sinusoidal function fit.
When Coach stops the process initiated by 'Auto-fit' this does not
automatically mean that the best fit has been found. Coach only is
of opinion that the process of numerical approximation has already
leaded to a reasonable result. A measure of the quality of the
solution found is the standard deviation (St. Dev.). By pressing
'Auto-fit' once more it may happen that the numerical
approximation process continues and that a better approximation,
i.e., with smaller standard deviation, is found

3. Another graphical representation of motion that is much used in

studies of human walking is the so-called hip-knee cyclogram. In
this diagram the knee angle is plotted against the hip angle for a
17
 complete gait cycle. Mathematically, it is nothing more than a
parameter curve (time is the parameter in this case).
Create the following cyclogram: plot the negative value of the
knee angle against the hip angle. Compare your diagram with the
following diagram (taken from A. Goswami, A new gait parameteriza-
tion technique by means of cyclogram moments: Application to human
slope walking. Gait & Posture, 8 (1998), pp. 15-36).

Has your cyclogram the same shape? Are the characteristic points
in the movement pattern that are mentioned in the above legend
also present in your graph and at the same place?

4. In which phase of walking is the right knee almost stretched? With

which points or pieces in the hip-knee cyclogram do these phases
correspond?

5. Besides in the swing phase there is another moment during which

the right knee is bent. When? With what point or piece of the hip-
knee cyclogram does this correspond? Can you think of a physical
explanation this bending of the knee?

6. During walking you also rotate your hip along the vertical axis of
your body. Think up a reason for this?

Leave the activity.

Answer 'Yes' to the question 'Save changed result?'.

Extra practical work about human locomotion

In the project Walking more activities have been prepared. They are related investigations
that you could do.

Activity A: Backward versus forward human locomotion

In this activity are two video clips of one and the same person walking forward and backward.
What are the differences and the common points in these gait patterns?

The data video window contains a movie in which a person actually walks backward, but the
video clip displays this locomotion in reverse mode.

18
Activity B: Jogging
This activity contains a video clip of the same person
that walks in activity 3. But now he is jogging.
Investigate the gait pattern. Two questions that you
could ask: What is the stride length and the step
frequency for this mode of running? Can the mathe-
matical model of a sum of two sinusoids for the leg
angles again be applied?

Activity C: Sprinting
In the video clip, Haile Gebrselassie from
Ethiopia and Paul Tergat from Kenia
compete for the first price in the men’s 10
km during the Sidney Olympic Games of
2000. Like four years before in the
Atlanta Olympic Games, the Ethiopian
wins. Use this video clip to investigate
the sprint movement. Some topics that
can be investigated:
- leg movements;
- step frequency and stride length;
- factors that contribute to the victory of
Gebrselassie.

Activity D: Race walking

This activity contains the following five video clips
about the women’s 20 km road walk during the
Sidney Olympic Games of 2000:
1. the last few meters of the Olympic champion,
Wang Liping (China);
2. the finish of the silver medallist, Kjersti Tysse-
Plätzer (Norway);
3. the finish of the bronze medallist, Maria Vasco
(Spain);
4. Jane Saville (Australia), who leads the pack of
athletes approaching the Olympic stadium;
5. The disqualification of Jane Saville a few hundred meters before the finish line.

19
The data video window contains the arrival of the gold medallist in the Olympic stadium. In
this video clip you have a side view of the gait of a race walker. The fourth movie of Jane
Saville leading the pack of athletes give you a front view. These movies give you a good
impression of the to typical gait.

Use these video clips to investigate the race walking gait. What are the official rules of race
walking and how do they influence the walking technique? Use Internet to find data about race
walking that are relevant for a description of this gait.

Activity E: own video clips of human locomotion

Of course you can record yourself video clips with a digital camera or with a webcam: record
and analyze the human movements of yourself and fellow pupils. You can think of
movements like normal walking, jogging, running, but you can also have a look at hopping,
crutch walking, fitness movements, dance movements, and silly walks like goose walking.
You can investigate whether the gait patterns change when instead of normal walking you add
heavy weight to wrists and ankles and whether walking with high heel shoes differs from
walking with regular shoes. In short, you may use your imagination to study an interesting
motion. But before you do this, make a work plan and discuss this with your teacher.

Activity F: Investigating the walking speed-frequency relation

In human science it is reported that a person who is constrained to walk at a given speed v on a
treadmill chooses a particular step frequency f and step length d = v / f. It is commonly
observed that humans use a fairly well defined set of v and f combinations that one might term
the natural walking gait. The empirical relation is f = C vb , where C is a constant and b=-.58
seems to be a good fit for humans over normal walking speeds. Go to a fitness center and
carry out treadmill experiments to investigate the v - f curves of you and fellow pupils. Also
record some movements with a digital camera or webcam for gait analysis purposes.

20