You might want to do a SLOC count as a rough metric of whether the processor has enough memory to handle the size of the executable. I do not have the time to do this and if I were to build a system like this I would port it to a processor that has enough power to do near real time updates.
sbright33:
I believe the only limitation is the processing power of the Uno. This is why I said it will not work with frequent updates. Can we agree on that? If not, why not?
As I said before, to know the answer you need to understand what the code is required to do. This will enable you to work out the amount of program space, memory and processing speed that would be required to do it. Are you seriously asking somebody to undertake all that research for you? If you're interested in the possibility, then go see what would be required. It's just as easy for you as it is for anyone else, and you have the advantage that you're the person who wants to know the answer.
Or to put it another way: No, it has no chance of working; the Arduino does not have sufficient program space, RAM or processing power to do this. Now go prove me wrong. 
I guess the only way to know for sure is to try to do it! You've raised some good points. The Speed is only relevant if you want to do it in real time. The program space is sufficient. 2K RAM is the biggest problem. Luckily you can add RAM or use UNO32 instead.
sbright33:
The Speed is only relevant if you want to do it in real time.
Are you sure? I thought that timing in GPS was crucial, given that you're measuring the distances to a set of objects all moving. Given that you're trying to measure cycle-to-cycle variations in the phase if a signal that has a very short wave length, with the signal passing between objects that are moving at thousands of miles per hour, it seems to me that you have a lot of data to process and that the data is inherently real time.
I could be wrong - I have no idea how this clever algorithm works other than what you've told us - but to me this sounds like a maths-heavy number crunching problem. This is why I thought you might want to get it working on a decent spec PC (if that's possible) and then see what resources it needs and whether you can finesse it down to run on something that's about a thousands times slower.
They already got it working on a PC. It seems like the parts where you have to measure time are doable on Uno or Uno32.
sbright33 any updates about your RTK project on arduino ?
sbright33:
Is there anyone who would wish to have this capability using Arduino?
Can you think of any applications?
Am I the only one?
Surveying devices with this accueacy already exist, they do not work in moving vehicles however.
They cost loads.
Base station idea with tx error signal has also been done.
I think it was commercial but dont have a link.
PeterH:
sbright33:
The Speed is only relevant if you want to do it in real time.Are you sure? I thought that timing in GPS was crucial, given that you're measuring the distances to a set of objects all moving. Given that you're trying to measure cycle-to-cycle variations in the phase if a signal that has a very short wave length, with the signal passing between objects that are moving at thousands of miles per hour, it seems to me that you have a lot of data to process and that the data is inherently real time.
I could be wrong - I have no idea how this clever algorithm works other than what you've told us - but to me this sounds like a maths-heavy number crunching problem. This is why I thought you might want to get it working on a decent spec PC (if that's possible) and then see what resources it needs and whether you can finesse it down to run on something that's about a thousands times slower.
I do not think the number crunching would be a problem for an arduino.
Phase and timing measurment though would require secific (fast) hardware.
wwbrown:
This stuff works I have seen it in practice and it is pretty neat for those who have the need for it. I wonder whether SA would be turned back on if there were cheaply available RTK systems. The combination of RTK and some of the emerging and established technologies is a scary mix for some people.
How cheap is cheap, for that scenario. Receivers capable of 1 inch accuracy, using RTK or GLONASS are readily available to the agricultural folks for well under $10,000.
I have this nagging memory of reading about a relative position sensing system that had been developed quite recently (perhaps in the last three months) that worked indoors and also gave a positional accuracy down to tens of cm.
You might be thinking of the Piksi, Peter. 4cm resolution at a price of $900 (per the pricing of the Kickstarter "rewards").
They appear to have completed their first production run last month.
Yes, that is probably it. I hadn't realised it was going to be that expensive, but compared to the alternatives it's a bargain.
It is only $400 / receiver + XBee costs. Soon to be $200!
Did this thread die?
I have a project where I need accurate location within a 100'x100' area.
This looks like it might work for me.
Where did the $400 soon to be $200 figures come from?
I'm going to try dead reckoning first.
How accurate? Will it have a clear view of satellites?
Should get you within ~8 feet (2.5 meter)
AFAIK there's two options: the Piksi and the Navspark. At their heart both are just GPS modules with "raw" GPS output. One GPS unit is "roving" while the other is stationary; both transmit their readings to a computer running the RTKLib software which resolves the rover's GPS readings to a more precise value based on the changes in the GPS readings seen from the stationary unit. Piksi claims a 2cm accuracy while the examples from the Navspark blog look more like 8cm accuracy. Mind you, at those levels things like antenna quality and nearby electromagnetic interference (motors, etc.) are an important consideration.
Not at all a plug and play solution. Someday we might see RTKlib's functionality written into the firmware of these modules to simplify things a bit but that's just not here yet.
I could live with a +/- 5cm but would rather have a +/- 2cm. It will actually be following several predefined patterns.
The Piksi would work. The wheels will have a quadrature encoder that will have 1 tick = 1mm resolution. I know that I'll have cumulative errors to deal with if I use dead reckoning. $900 is out of my price range.
Anybody still playing with this? sbright33?
I just need a positioning system 100x100m to within 5-10cm. Can someone help me?
pedrodafloresta:
I just need a positioning system 100x100m to within 5-10cm. Can someone help me?
You could walk the same area few times and take out an average...
I'm interested in accuracy of this little guys myself - as an cartographer I may run some accuracy calculations - hit me with your txt or csv files guys.