The accuracy quoted is known as CEP -- circular error probable. A 2.5m CEP means, if you drew an 2.5m circle around where the GPS said it was, The GPS is inside that circle.
So why not a perfect location? why are some better than others? It comes from the error in the system.
GPS is measured via time-of-flight. Each satellite has a super accurate, synchronized clock, and the difference between the real time and the measured time tells the GPS how far away it is from each satellite. since it knows where the satellites are in orbit, it can triangulate where it is on earth using those distances.
So where's the error? Couple of sources:
- round-off error can make a *HEUGE* difference. those satellites are a *LONG* way away.
- where the satellites are in the sky -- are all the visible satellites over on one side or are they evenly spaced around the horizon? The closer they are together, the harder it is to measure the position -- also why it's not the same vertically as horizontally, but rocket's right -- it's worse vertically -- they're quoting the baro accuracy in the vertical
- interference (lets assume that's not a problem -- most OEMs do
)
Why should you care? first a neat trick -- if you've got an 2.5m CEP, and you measure the position over time does the location move quickly or slowly? the answer -- it slowly drifts around that circle --VERY slowly. why is that neat? the difference between one measurement and the next *even while you're moving* will be *EXTREMELY* precise. you might have a huge CEP, but can measure speed very accurately, because you'll keep landing in the same offset in that circle!
OK, DAN, WHY SHOULD I CARE? because you're multirotor will hold it's position my measuring an initial point when you let go of the controls, and it will keep controlling the copter back to that point. Is it really in the lat/long it says it's in? NO! but who cares?!? All it needs to know is it's drifted 3cm left since the last measurement, and that's how far it needs to move back! that 2.5m GPS can *still* measure being 3cm off!
So why do some drift worse than others? processing delay and that nasty round off error. If the processor is having trouble keeping up with the corrections, the GPS doesn't update fast enough, or the control loops aren't tuned well, it'll hunt for that position more.
