Biomch-L Readers

I have included the responses to my Kistler Bioware CoM calculation
question below. Thanks to all those who replied. From all of the
feedback I am expecting to be able to further understand and resolve my
original issue.

In appreciation,


Dr Peter Clothier
Lecturer - Sport & Exercise Science
Building 24 Level 4 Room 7

School of Biomedical & Health Sciences
University of Western Sydney
Campbelltown Campus
Locked Bag 1797
Penrith South DC NSW 1797

Ph. +61 02 46203743
Fax. +61 02 46203792
Hi Peter,

I played around with this a while ago and put the spreadsheet here: (third row from the bottom of the

Integration is always a fiddle because you don't know the integration
coefficient - in this case the initial vertical velocity. Luckily, the
vertical direction is unique, in that you know that it must average zero
(otherwise the runner would start flying or sink into the track), so the
trick is:

1. Run the integration with initial velocity set to zero
2. Calculate the mean vertical velocity, Vm
3. Re-run the integration with initial velocity set to -Vm

Note that the subject's body mass is also quite critical, so needs to be
measured accurately (or fiddled!). Also, you can't know the absolute
height of the CoM so you have to estimate it (the NASA rule of thumb is
54.3% of height from the ground).

Have fun!

Dr. Chris Kirtley MB ChB, PhD
608 Dockside
44 Ferry St.
Kangaroo Point
Queensland 4169
Australia (GMT+10)
Tel. 61+7-3891 6644 x 1608
China mobile: 13418732724
Fax 3891 6900
You would need to ensure you have included the initial conditions
(velocity and position) before you perform the double integration
otherwise your calculations will be inaccurate. These initial conditions
have to be determined at the point of impact on the platform. I'm not
sure whether Bioware v4 allows you to include these values as I don't
use this.
Dr Drew Harrison
Senior Lecturer and
Head of Dept of Physical Education and Sports Science University of
Tel: +353 61 202809
Dear Peter,
One step that you have not described is that of subtracting the
bodyweight of the subject from the vertical force component before
integrating it. This may have been done, but not written in the previous
Vertical force - subtract bodyweight to get acceleration of COM -
integrate to get velocity of COM - integrate to get displacement of COM.
Hi Peter,

I have been working on measuring COM displacement in my research. I am
not familiar with the Bioware software, but I am familiar with kinematic
and kinetic approaches for measuring COM displacement. Double
integration of the force plate signal will result in vertical COM
displacement, but only if you take into consideration the mass of the
subject and the integration constants that emerge when you perform the
integration. The following article gives a good overview of different
approaches for measuring COM displacement.

Gard SA, Miff SC, and Kuo AD. Comparison of kinematic and kinetic
methods for computing the vertical motion of the body center of mass
during walking. Human Movement Science; 2004; 22(6): 597-610.

Good luck!

Ben Roewer
MSME/MBA student
University of Delaware
That's true, double integration of vertical force should be related to
the vertical displacement. The problem are the numerical errors that
grow like time^2. However, since running is a cyclical motion then
perhaps a Fourier method would work? FFT the force data, replace the
frequency coefficient a(f) with -a(f)/(2 \pi f)^2 [corresponds to double
integration; f = frequency] and then take the inverse FFT.
Regards Frank Borg
Hi Peter,

I have never used the kistler software to perform the double
integration. So, no idea what's going on with the software. However, one
thing to keep in mind is that the constants required to obtain exact
solutions from the integration steps, being the initial velocity and
position respectively, are critical to obtain realistic CM traces.
Attached is a paper that could point you in the right direction, if this
indeed is part of the problem.

I don't know if it is possible for you to plot the raw data and look at
the pattern first.
The double integration is very sensitive to noise in the signal,
therefore the signal is usually conditioned/filtered first. The nature
of the forceplate instrument is that it typically have some inherent
signal noise, in our case, it's about 0.25%. So, that might have to be
resolved before you can compute for accurate CoM. I hope this helps.