In what little I've read about it, it seems one of the big limitations to accuracy and speed is the amount of number-crunching required. The GPS unit triangulates its position in 3 dimensions by doing a bunch of high-precision floating-point math on the signal strengths and directions from all the satellites.
an example of the amount of number-crunching power it takes:
http://www.dsprelated.com/showmessage/22833/1.php
Quote:
The theoretical limit is down to the integration time for the receiver,
typically 1-10ms. If you want high dynamics, you need to update faster.
For car racing, it's probably closer to 1ms.
Most comercial receivers output 1 position per second because it takes
some number crunching to compute the satellite positions and solve for
the receiver position. This is purely a matter of computing power.
Given a fast enough processor, 1000 positions per second is possible.
I was running an automotive receiver IC (60MHz,32bit MIPS processor)
at about 10 fixes per second, getting close to 100% CPU.
based on a quick googling, it looks like a DSP chip of that speed is something like $25 in production quantities, and I would imagine a GPS system that used it would be a few hundred bucks at least, and for something in the range of 30-50Hz (3-5 times faster) I'd expect the price to rise exponentially, which seems to be the case based on the prices that have been tossed around.
__________________
But don't take it from me! here's a quote from a real, live newbie:
Quote: Originally Posted by Viscouse
I am learning buttloads just by searching on this forum. I've learned 2 big things so far: 1-it's been done before, and 2-if it hasn't, there is a way to do it.
eegeek.net