Tuesday, July 19th, 2011 « Untold Arsenal: Arsenal News. Supporting the Lord Wenger; coach of the decade
Untold Arsenal on Twitter @UntoldArsenal
By Gordon Haverland, P.Eng.
Some of us Arsenal fans have noticed that Arsenal is using GPS vests in the friendlies this summer (and at least once – at Barnet – last
year). We may have personal GPS, or GPS built into our cars. How does this help Arsenal? Can this help me understand what GPS is doing for me now?
I am an engineer who has always been involved with lots of mathematics and statistics. For about 6 years (1996 – 2002), I spent my life using GPS and geostatistics to help farmers in Alberta, Canada.
When I started working with GPS, a feature called Selective Availability was in effect, which scrambled the system such that positioning errors of 100m were relatively common.
At the same time, it was fairly common for GPS solutions to suffer “blunders”.
In GPS, a person is interested in the distance between the satellite and the receiver, and this distance is divided into two parts: the integer number of wavelengths of the radio signal, and the fraction that is left over (usually called phase).
A blunder happens when the phase is correctly calculated, but an error in
the integer number of wavelengths is made.
Precisions of 100m weren’t of much use back then in farming (and wouldn’t be useful in football), and so we actually employed something called Differential Global Positioning System (DGPS).
Over a precisely known location that was reasonably close to where we were working, we would set up a GPS antenna at some known distance above the ground. Typically we used formal survey markers. And we set up a little computer at this location which would be constantly comparing the location via GPS with the known location (to a couple of centimeters), and calculating an error vector. This error vector was transmitted over a radiomodem to
our “roving” GPS antenna.
By this means, we were able to obtain precisions on the order of 10cm (in general). But we still observed the occasional blunder.
With Selective Availability turned off now, most people should see positioning errors of 5-10m. Back in the late 90′s you could subscribe to satellite services, which would produce DGPS estimates on the order of 50-100cm. These kinds of corrections won’t get much better over time.
I would imagine that Arsenal is setting up their own GPS base station at the field, and calculating corrections. This would also be a good place to receive data from each player and store it on the computer.
Since the early/mid 90′s, the maths has improved, reducing the chance of blunders. Even back then, GPS was commonly solved via methods involving the Kalman transform. Do you remember FM radio receivers talking about phased lock loops? That’s an example of a Kalman transform.
In any event, in a Kalman transform you can tie together related information to stabilize the solution. Handheld GPS receivers have had atmospheric pressure sensors (nominally measuring changes in elevation), and magnetic direction indicators (compass) for quite a while, having gyroscopic information is also useful.
Our (farming) DGPS setup was measuring positions once per second. I am going to guess that the GPS setup for Arsenal is probably running 50 to 1000 times that fast. If two players collide, or a player collides with some other object (for example a goal post), these higher sampling rates would allow someone to get a fairly good idea of the collision forces. Having video evidence will be very useful in that circumstance.
The Arsenal.com article mentions that heart rate information is gathered. For me, I would want the gyroscope information, accelerometers and possibly body temperature. The article doesn’t mention whether the data is stored on the athlete, or broadcast over radio to a central receiver/computer.
Storage like flashdrives are quite small, but I would think that an important criteria is to minimize just how much extra mass is present in that vest, and whatever mass is there should not be a hazard to the player in the course of the game.
If the information is broadcast back to a receiving station, I would guess it is encrypted. I don’t think Arsenal would want anybody getting copies of the information. How much equipment is in that vest might be reduced if the GPS equipment on the player is sending back nominally raw information, which would have to be post-processed in order to convert it to positions (and orientations in space).
If a person just wants to know how far a player runs during the course of the game, trivial analysis of the GPS information will probably be pretty close. If you are trying to find out how much force any of the muscles working are exerting, things get more complicated. You need to model how the player’s body is moving, not just assume the GPS antenna is some constant distance from the ground in some known relationship to the athlete’s two feet.
Untold Media: Launch of new media section, and we need your help