PDA

View Full Version : summary - delays with a Polhemus



Andrew Karduna
01-09-2001, 12:08 AM
I recently asked for help with the delays between sensors with a Polhemus
Fastrak magnetic tracking device.

Two approaches were suggested to address this problem:

Shannon's Theorem
Hamill, Caldwell and Derrick, "Reconstructing Digital Signals Using Shannon's
Sampling Theorem," Journal of Applied Biomechanics, v13, 226-238, 1997

SCAAT (single-constraint-at-a-time)
www.cs.unc.edu/~welch/scaat.html

According to Stephen Cheetham at Skill Technologies, Shannon's Theorem has been
incorporated into their software.

I am including the actual replies below. Thanks to all who replied.

- Andy Karduna




__________________________________________________ _

Jason Harrison:

Have a look at the SCAAT website -- the problem of multiple sensors
with individual lags is something that lots of people would like to
ignore! I know of at least one kinesiologist who claims that the
Optotrack system doesn't have any intrasensor lag -- which is
ridiculous because the infrared diodes are strobbed.

http://www.cs.unc.edu/~welch/scaat.html

__________________________________________________ _

Paul Treffner:

Have you contacted Skill Tech who do this in software?

Try:
"Bryan St. Laurent"

__________________________________________________ _

Stephen Cheetham:

It is true that the Fastrak reads its sensors sequentially. The time skew
that is created by sampling sequentially can be corrected by using Shannon's
reconstruction algorithm to reconstruct up to 120Hz and then simply match
samples at the new rate. This provides a much better more accurate method
than simply doing a linear or spline interpolation. This allows for the
FasTrak to be used for capturing motion accurately at high speeds. Skill's
6D-RESEARCH system software V3.0 has this algorithm built into it for user
access when required.

Please note however, that the Polhemus StarTrak reads all sensors AND COILS
(i.e. x,y,z coils in each sensor) simultaneously at 120Hz, for 1 thru 16
sensors. The Ascension MotionStar, on the other hand multiplexes between
coils in each sensor. It samples these three individual coils (x,y,z)
sequentially and therefore it has time skewing problems in each sensor.

__________________________________________________ _


Peter Condie:

I have been forwarded an email in which you discuss acquiring data
from a Fastrak machine in real time.We have been attempting to
collect data from Fastrak machines in real time, though we have
had only limited success.

We have been using a LabVIEW program to monitor the contents
of the serial buffer of the COM port connected to the Fastrak
machine. Although this does not tell us when the Fastrak machine
has collected the position data we are after, it does tell us to within
a millisecond when the PC we are using has received a full string of
position data from the Fastrak. We then use this information to help
interpolate the Fastrak data.

We are not satisfied that this method is satisfactory for the
experiment we are currently attempting as the activity we are
analysing is only 500 ms in duration, but I think it may useful in a
slower activity.

I have been sent some demonstration software from a company in
the US called Skill Technologies, that claims to be able acquire
Fastrak data in real time. I am sceptical of these claims though I
haven't tried the program as yet.

---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------