View Full Version : Summary of Responses: Foot Contact in Running

06-18-2007, 01:25 AM
To All,

Thank you for your responses and suggestions. Below is a summary of
responses to my question regarding foot contact in running.

Evie Burnet
************************************************** *************************
Original Question:

I am trying to identify the timing of foot contact during treadmill running
in the absence of an instrumented treadmill. The foot switches I am using
are prone to breakdown, and can be unreliable. Therefore, I am looking
for an alternative indicator. I also have kinematic shank data, and would
appreciate any references or suggestions.
************************************************** *************************

If you are collecting kinematic data (I assume with a motion capture
system) there are ways to extract heel marker positions and derive the
accelerations of the heel. From this, reliable heel strike timing can be
computed with MatLab. We have a bit of code that does this fairly
reliably. Just a suggestion.

Best of luck,

Jason A Schoen
VA Center of Excellence for Limb Loss Prevention and Prosthetic Engineering
Seattle WA

Use high-speed video (>60 hz), sagittal plane, and count frames.

Kathleen E. Costa, PhD
MEA Forensic Engineers & Scientists, Inc.
(949) 855-4632


Here are some references you may find useful:

Hreljac, A., Stergiou, N. (2000) Phase determination during normal running
using kinematic data. Med. Biol. Eng. Comput., 38, 503-506.

Hreljac, A., Marshall, R. N. (2000) Algorithms to determine event timing
during normal walking using kinematic data. J. of Biomech., 33, 783-786
W. Brent Edwards, M.S.
Ph.D. Student
Dept. of Health and Human Performance
Iowa State University

Hi Evie,
One solution is to place a leg of the treadmill on top of a force platform
and use the vertical signal as a foot switch. Alternatively, here is a
paper on a method to measure heel-strike and toe-off using the time series
of heel and toe markers:

Hreljac A and Marshall RN (2000). Algorithms to determine event timing
during normal walking using kinematic data. Journal of Biomechanics, 33,

Best regards,
Med Biol Eng Comput. 2000 Sep;38(5):503-6. Links
Phase determination during normal running using kinematic data.
Hreljac A , Stergiou N .

Hope this helps,
Toran MacLeod
We make footswitches to detect foot/floor contact. Price: $195.00.


Best Regards,

Lee Barnes
B&L Engineering
3002 Dow Ave., Suite 416
Tustin, CA 92780
TEL: (714) 505-9492
FAX: (714) 505-9493


A number of years ago, a PhD student working with Dave Winter at the
University of Waterloo studied treadmill gait and used changes in the
current in the treadmill motor to calculate mechanical power. His name is
Bertrand Arseneault and I believe they did publish a paper from this study.
It may be possible to monitor the treadmill motor current and use the
fluctuations to correlate foot-treadmill contacts.


Drew Smith, PhD
Senior Lecturer, Biomechanics
Department of Sport & Exercise Science
University of Auckland, Tamaki Campus East
Private Bag 92019, Auckland Mail Centre Auckland
New Zealand 1142
Office: 09 373-7599 x86849 [Int'l: +64 9 373-7599 x86849]
Fax: 09 373-7043 [Int'l: +64 9 373-7043]
Mobile: 021 264-9790 [Int'l: +64 21 264-9790]

Inshoe pressure systems are your best bet, we've used the Pedar mobile
and X systems successfully.

Rami Abboud
Professor R J Abboud
Deputy Head of Division, Surgery & Oncology
Master of Orthopaedic Surgery (MCh Orth) Course Director
Director, Institute of Motion Analysis & Research (IMAR


Hi Evie,
I have used a modified method from Hreljac & Marshall (JOB (2000),
33:783-786) adjusted for treadmill locomotion. I am working on a technical
note for the method right now. Do you have a heel marker with kinematic

John De Witt, Ph.D., C.S.C.S.
NASA - Johnson Space Center
Exercise Physiology Laboratory
Houston, TX 77058
281-483-8939 / 281-483-4181 (fax)

Dear Evie
I used a method base don the cross-correlation of the trajectory of a foot
marker with a reference of the same foot marker signal measured on the
force-plate (in fact, this reference was an average of different trials
with different subjects about 100 ms around the foot contacts). It has been
described (briefly) in a paper: Forner Cordero, A.; Koopman H.F.J.M; van
der Helm F.C.T. (2003). Multiple-step strategies to recover from stumbling
perturbations. Gait&Posture, 18(1): 47-59: "The gait events, heel contact
and toe-off, were calculated from the heel and toe markers data by
comparing their temporal sequences with a reference pattern of heel contact
and toe-off previously measured on the force-plate. This automatic
detection was checked manually and compared with the data from the
footswitch (right heel contacts). The differences between both methods were
below 20 ms, which was the video sampling period and was considered
acceptable". I hope this helps. Arturo

Hi Evie,

have a look at page 13/14 of the attached paper. They describe a method
based on local maxima and minima of toe and heel acceleration (rather
counterintuitive at first sight). It's about walking, not running, though.
Besides, I heard that some people use a correlation method: select a
typical moment of foot contact (e.g. heel-on) in your data, cut out some
interval around it and correlate it (shifted along the time-axis) with the
rest of the data. Times (shifts) of maximum correlation are likely to be
similar events. I have no reference for this one and haven't tried it
myself, but I heard that VICON use it in their kinematic analysis software.
Good luck, and best regards, Julius

Hi Evie,

If you have not checked out this article then it is probably worth a look
as it seems to address your issue.

Hreljac, A. and Stergiou, N. (2000) Phase determination during normal
running using kinematic data. Medical and Biological Engineering and
Computing. 38 (5) 503-506.

There are also quite a few other papers which look at using kinematics to
determine foot contact in walking (one of the most recent is: Automatic
detection of gait events using kinematic data Gait & Posture, Volume 25,
Issue 3, March 2007, Pages 469-474 Ciara M. O'Connor, Susannah K. Thorpe,
Mark J. O'Malley and Christopher L. Vaughan)

I have also heard of a system which uses the kinematics of 2 markers placed
on the treadmill which show rapid vertical acceleration at the instant of

I would be interested to hear what method you selected.


Hi Evie,

I do have a few suggestions. To check them out you will need an
oscilloscope with a high impedance probe. I've noticed that the DC drive
motor on my treadmill generates extra transients on my video line at each
foot strike. Maybe the start and end of the transient bursts could be used
to capture the TD and TO times. Or, you could pick up the signal from the
speed controller circuit. That's always toggling on and off. Or, you might
look at the frequency of the speed pick-up. That will shift abruptly with
each TD and TO. I'd suggest that you poke around with the probe while
someone is running on the machine. You might find the signal you need on
the controller board.

Another approach might work. I built a primitive circuit last summer that
worked pretty well. It had two rheostats with 10-cm arms that rested on the
edge of the treadmill. Together they formed one side of a Wheatstone
bridge, so I could get a voltage curve that represented the tread board
deflection. I have some grounding problems so the signal had a lot of 60 Hz
noise and it had to be filtered. I was able to extract the TD and TO times
from the leading and trailing edges of the voltage curve. Unfortunately,
post processing the data was time consuming.

I had hoped to use this circuit to measure vGRF, but the tread board
flexure was non-linear. Maybe you will have better luck with your

I'd be interested in reading what others have done to solve this problem.

Ted Andresen
St. Petersburg, FL

Dear Evie,
You could use your kinematic data for indentified heel strike and toe-off
events during running. The paper of Alan Hreljac (California State
University) could help you:
Hreljac A, Stergiou N. Phase determination during normal running using
kinematic data. Med Biol Eng Comput. 2000 Sep;38(5):503-6.

Best Regards,
Reginaldo Kisho Fukuchi
Laboratório de Biofísica -EEFEUSP
Laboratório de Análise do Movimento
Instituto Vita


Hmm- you could look at Hreljac and Stergiou as a starting point (Med. Biol.
Eng. Computing, 2000, 38, 503-506). They use shank data as a method of
determining HS and TO overground. The trick is how to make it work for
treadmill data. During overground gait, the shank is always moving forward
in the inertial frame. However, in treadmill gait, the shank moves forward
always in the treadmill reference frame, but moves forward and backward in
the inertial reference frame.

It seems like a reversal of forward to rearward angular velocity of the
shank should occur near HS, although I am not sure if this is an absolute.
It is possible that the rearward motion of the shank precedes HS by a small
interval. Maybe another way to say this is that if you are standing on the
side of the treadmill and the person is facing the left, I would expect a
clockwise (CW) rotation of the shank as the leg swings, which reverses to a
CCW rotation near HS. TO is tougher, although I would expect a 'jump' in
shank AV magnitude as TO occurs and the leg begins to swing.

Do you have AV info on the shank?

Dear Evie
Currently, I have defined a method which can be refered to as SET (Sagittal
Extremim Technique). the abstract has been accepted for salfod
international biomechanic conference. I am preparing the full paper.
attached is the abstract. if you think this method might help, and you need
more information about it please dont hesitate to contact me.

Behdad Tahayori
Research Officer
SUMS,School of Rehabilitation, Iran
Tel: +98-711-6261081
Fax: +98-711-6272495
mobile: +98-917 7029441
Good Afternoon,

I saw your posting on bmch-l about treadmill stance determination. I know
of one running paper, which developed a method to determine stance using
kinematic shank and foot data. I hope it helps.

Hreljac, A and Stergiou, N. 2000. Phase determination during normal running
using kinematic data. Med Biol Eng Comput. Sept;38(5):503-6.

Good Luck!

Becky Fellin
Research Assistant
University of Delaware
i expect you would get a number of relevant responses about this, but this
may work for you. in the not so distance past (~6 months ago) i received
some info on low cost force sensing strips/pads. as i recall they might
give you a low resolution profile of foot impact if you digitized the
output. if you want i could search out the info.

Dear Evie Burnet,

in our lab at the Department of Movement and Sport Sciences at the Ghent
University (Belgium), we have the same problems. We tried insoles to detect
gait events, but they were also damaged really fast.
It is our purpose to investigate gait also in overground conditions.
we can not alway acquire kinematic data.

I would appreciate if you keep me in touch with this problem.

Kind regards,

Kristof De Smet
Department of Movement and Sports Sciences
Watersportlaan 2
B-9000 Gent
Tel +32 (0)9 264 63 12
GSM +32 (0)486 41 56 20
Fax +32 (0)9 264 64 84
Dear Evie

The switches I am using at present seem very reliable although I have not
used them running they have lasted for many hours without breakdown. They
switches are designed for TV remote controls and the like and are about 4mm
thick the only ID numbers on them are DIP / B but I but them from the local
electronics store for about 80p ($1.40) each. I mount then in 3mm self
adhesive poron and are very comfortable underfoot.

All the best Dave Smith