Measuring running speed (average every 10m) during a 5-10 km distance run

[Pieter Fiers]

Dear colleagues

We are searching a method to measure the running speed during a 5-10 km distance run. We would like to register the average running speed every 10 m. Total running distances will be between 5 and 10 km (1 or 2 laps). All subjects will run on the same track where certain distances can be indicated (ex. adding poles, marks on the ground, ...).
Since the subjects will run regularly, the system must work without involvement of an experimenter at the time of running (postprocessing is not a problem) and must be easy to handle by the subjects. Ideal, the system is light to carry and doesn't need a lot of power (ideal, the system is powered by a smart-phone).

We tried a sonar sound system (responding to objects on the track) but it doesn't work in outdoor conditions.

Maybe:

• a GPS-system will do but the precision (short distances) might not be sufficient, especially in bad weather conditions.
• a distance-laser would be a good device but we don't know about the power-issues.

• filming the run and indicating certain objects on the track (placed every x metre) is another way but this would be quit timeconsuming.

Any thoughts?

Thanks

Pieter
Belgium

[Daniel Feeney]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

You could consider a garmin running watch and analyze the GPX file. I believe you can alter their sampling rate, but default is 2 HZ for the models I've used. The newest models have the best sampling rate.

[Edmund Cramp]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Have you thought about a cell phone app like RunKeeper on Android? I doubt that any commercial applications will do exactly what you want but they might get fairly close to it, and while the spot accuracy might be lacking, the overall results could be quite reasonable.

[Tim Lindsay]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Hello Pieter,

I think what you are asking is the "dream" of researchers who are interested in pacing strategy during running. I would love to see you be successful in collecting this data. When you say "track", you don't mean a 400 m athletics track, but a measured course on the road or trail, right?

In addition to GPS technology, I would also try inertial sensor/accelerometer technology such as those products mentioned here:

http://www.topendsports.com/testing/equipment-footpod.htm

I work with accelerometers, but I haven't used them to determine speed, although it is theoretically possible. I have not yet seen any pacing strategy studies with data points as close as 10 m, but I hope it will be possible soon. If foot-pod devices don't turn out to have the resolution or reliability you desire, I suggest revisiting the GPS option, but with a lower resolution like 20-50 meters. Even measurements every 20 meters over the course of 10 km will generate 500 data points.

But, then again, I would encourage you to think carefully about your need for high resolution measurement. If you are going to apply non-linear analysis methods to the fluctuations, then I can understand the need for many data points, but if you wish to describe pacing strategy on a more global level, the number of data points may not be crucial.

Of course, I am trying to anticipate your research question at this point, and I apologize if you have already given these items much consideration and aren't able to be more forthright with your plans.

[Scott McMillan]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

As Timothy Lindsay mentioned, you are looking for the "dream" to track people accurately.
I have looked into this field for some time (honestly, to bring technology forward to researchers, athletes and coaches with our company X Co).
As you say common GPS probably won't work, it is too inaccurate.

Your best bet is to search for GNSS and differential based systems but those may only be prevalent in research and not commercially available at the moment.
Catapult Sports provides commercial units with GPS and intertial sensors (like accelerometers), but I don't know how accurate they are and if that meets your requirements.

We specialize in RTLS (RF based) precise positioning that would be more than accurate for what you need (centimeter level). However, these and other similar systems have limited range. It is possible to do over long ranges but does get expensive to implement. Whereas, a 400m track or similar area would be much more feasible.

Hope that helps,
Scott McMillan, MSc
Biomechanist
XCo Tech Inc

https://www.xco.io

[Michael Orendurff]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

I recently purchased an OptimEye S5 unit from Catapult and have just begun to test its accuracy. First, I almost fainted when I saw the $5000 USD price tag, but I somehow managed to suppress my outrage and bought it. I believe it has a 10 Hz GPS that locks onto both US and GLONAS GPS systems. It also has 100 Hz 3D accelerometer, 100 Hz 3D magnetometer, and 100 Hz 3D rate gyro...and is Polar heart rate sensor compatible.

Previous published work has been favorable for tracking the individual's speed, and the 10 Hz sample rate a big improvement in "instantaneous" forward velocity. (http://www.ncbi.nlm.nih.gov/pubmed/22122431)

Running in a motion capture lab (sampling at > 100 Hz) will demonstrate a sinusoidal forward velocity pattern of the center of mass, forcing you to reconsider the meaning or utility of a measurement of "forward velocity". The COM forward velocity fluctuates about 10-20% within the stride compared to the average forward velocity across that stride, and this delta V of the forward velocity increases with speed (with stride length). Sometimes looking with too great a sample resolution confuses the point of the variable you seek to measure. So the solution is pay way too much money to oversample, and then work way to hard to decide on the right smoothing characteristics to get a meaningful and useful variable.

I have used the OptimEye S5 sensor to walk the lines painted in a parking lot for each stall a car should park in, and then plotted the XY data over satellite imagery. The position data clearly falls on the painted lines of the image, but occasionally drifts ~0.3 m along a single axis over a 1-3 sec interval until the S5 sensor drops the offending satellite from the matrix of satellites that provide data. The S5 sensor is worn in a tight vest in a small pocket between the shoulder blades, so stays close to the individual's body, but not close to the COM.

Validating the OptimEye S5 is quite a daunting task, as there is no easy to use, large scale system to use at the "gold standard". If you could get access to a football (soccer or Australian Rules) stadium, there are a few possibilities for improved measurement validations. (http://www.ncbi.nlm.nih.gov/pubmed/23221963).

Michael Orendurff, PhD
Seattle

[Tilak Dutta]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Hi Pieter,

Would you be open to setting up a series of GoPro or other video cameras around your track to create an inexpensive 3D motion capture system? If so, you could look at freely available software from the Hendrik lab: http://www.unc.edu/~thedrick/software4.html that can use up to 9 cameras. It was developed for tracking birds and bats outdoors. You'd have to consider the cost of processing tons of video of course, but it may be faster than the manual method of placing objects around the track. To make it easier to track your runner, you may consider constructing a large (say 10cm diameter) glowing orb attached to a hat/helmet.

Good luck!
Tilak

Tilak Dutta PhD
Scientist and Technology Team Leader
Toronto Rehab Institute
tilak.dutta@uhn.ca
416.597.3422x7956
tritechteam.org

Dear colleagues

We are searching a method to measure the running speed during a 5-10 km distance run. We would like to register the average running speed every 10 m. Total running distances will be between 5 and 10 km (1 or 2 laps). All subjects will run on the same track where certain distances can be indicated (ex. adding poles, marks on the ground, ...).
Since the subjects will run regularly, the system must work without involvement of an experimenter at the time of running (postprocessing is not a problem) and must be easy to handle by the subjects. Ideal, the system is light to carry and doesn't need a lot of power (ideal, the system is powered by a smart-phone).

We tried a sonar sound system (responding to objects on the track) but it doesn't work in outdoor conditions.

Maybe:

• a GPS-system will do but the precision (short distances) might not be sufficient, especially in bad weather conditions.
• a distance-laser would be a good device but we don't know about the power-issues.

• filming the run and indicating certain objects on the track (placed every x metre) is another way but this would be quit timeconsuming.

Any thoughts?

Thanks

Pieter
Belgium

[Philippe Malcolm]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Maybe you could also look into devices that are used by biologist for wildlife tracking? Most of these are GPS based but some are probably not.
Good luck!

----------------------------------------------------------------------------------------
Philippe Malcolm, Postdoctoral Research Fellow at Harvard Biodesign lab
and Wyss Institute for Biologically Inspired Engineering

http://biodesign.seas.harvard.edu/soft-exosuits

http://users.ugent.be/~ddclerc/WALL-X/

https://www.researchgate.net/profile/Philippe_Malcolm

----------------------------------------------------------------------------------------

[Rodger Kram]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

re: measuring locomotor speed of a person.

Here is an idea that is only 120 years old.
Etienne Jules Marey built a "portable odograph" for this purpose.

Basically it was a wheel pushed or pulled by the person.
Marey did not have electronics (though he did have electricity), but he did have abundant creativity and ingenuity.

I would suggest that for running, it would be possible to make a pull behind cart that attached via a padded waist belt, with one or two lightweight, small diameter bicycle wheels and instrument them with either an optical encoder or a magnetic reed switch that pulsed every revolution or more frequently if desired. It could input to a data logger etc.

To see Marey's device, google E.J. Marey portable odograph or click on this link.

https://books.google.com/books?id=e5ZBNv-zTlQC&pg=PR9&lpg=PR9&dq=E.J.+Marey,+Portable+Odograph&source=bl&ots=GetTpp500y&sig=Gs0YwVcwamR664FhSosGkvYCx6A&hl=en&sa=X&ei=YA08Vc6ML9TcoAT4yoGwDQ&ved=0CB8Q6AEwAA#v=onepage&q=E.J.%20Marey%2C%20Portable%20Odograph&f=false

At that same location is a drawing of Marey's stationary version in which the runner appears to push a lightweight "gate" every so many meters. That somehow sent an electrical pulse to his recording smoke drum.

Marey is a hero for many of us.

Rodger Kram
University of Colorado

[Ross Miller]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Pilot data (n=2) from my favorite new toy (the Apple Watch) suggests it's about 96% accurate for running distance.

By "accurate" I mean in comparison to what I normally use (hand-drawn routes on runningmap.com when outdoors, or the treadmill odometer when indoors) and who knows how accurate those are.

Ross

[Ton van den Bogert]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Ross, that's interesting.

Any idea about the accuracy for small time intervals such as the 10 m mentioned by the OP?

And I assume that the watch uses GPS to measure running distance, so it would not work on a treadmill. Is that right?

Ton van den Bogert

[Ross Miller]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Ton: I haven't tried it for anything shorter than a few miles or anything where I actually have a good gold standard for what the distance really was, but I plan to compare it to some force plate-derived distances and some treadmill runs with the speed measured by a hand tach this week. I'll post the outcome here.

The watch itself does not have onboard GPS. If you have your phone within Bluetooth distance of the watch, it will use its GPS, but I've just been using the watch by itself. It does work on a treadmill actually. I've only used it once and the watch said the distance I ran was a little over 4% shorter than the distance on the treadmill. Most of that was from the first mile then it drifted a bit each mile as I went. The two times I've used it outdoors the errors were 3% and 0% compared to what I mapped out on runningmaps.com (which I think uses Google Maps data).

I'm not sure what algorithm it's using for any of the workout data (it also gives you pace, calories, and heart rate, both at the end and instantaneously as you're going). You have to "pair" it with a phone to activate it, so I suspect it's using the long history of GPS data on my phone somehow, combined with the accelerometer/inclinometer data from the watch itself.

Ross

[Ton van den Bogert]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

That's remarkable that it works so well without GPS (and therefore also on a treadmill). Very curious if anyone has an idea about the algorithms that are used.

Also curious how reliable this is over a range of speeds.

Ton

[Ross Miller]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

Here's some distance data at four different speeds (a half-mile each at 6.0, 7.0, 8.0, 9.0 mph) on a treadmill with a speed reading I'm pretty confident in, compared to what the Apple Watch reported, at every 0.1-mi increment on the treadmill.

It looks to have been pretty spot on at the two slower speeds then underestimated distance at the faster speeds.

Not that close to the 10-m increments the OP is looking for, but I would expect a similar trend there.

edit: also the treadmill was at a 1% incline.

Ross

[Ross Miller]

Re: Measuring running speed (average every 10m) during a 5-10 km distance run

An update after using this gizmo for a while now (Apple Watch):

The total distance it measures is not as accurate most of the time as what my previous post indicates. It is sometimes off by an error of up to 10% or more (highest I've seen so far is 13%). I stopped keeping detailed records a while ago but I would guess that the average error per mile is on the order of 4%.

By "error" here I mean distances on gym treadmills and Google Maps (not necessarily gold standards) but the watch is definitely reporting different distances, on several different treadmills and a variety of outdoor routes.

Interestingly it seemed more accurate early on before I had used the GPS on my phone to "calibrate" it to my stride length. My understanding of how this works is (i) if you haven't calibrated it then it uses an estimated step length based on your height, and (ii) if you take both the Watch and an iPhone with you on a run while the "Workout" app on the Watch is running in "Outdoor" mode, it will use the Watch accelerometer to get your step count, use the GPS on your phone to get the total distance, then from these data it has your average step length and can then get distance and speed from the Watch on its own by assuming this step length and by getting the step count and step rate from the Watch's accelerometer data.

Apple doesn't publish any documentation on what the algorithms are specifically, but what it sounds like to me from the limited info they do provide is that the Watch stores two values (a step length for walking and another one for running) and uses one or the other depending on what type of workout (walking or running) you tell it you're doing. It seems like it would be pretty easy to develop an algorithm from a more extensive calibration that learns what your step length is as a function of step rate, based on features of the acceleration signal and GPS info.

Ross