SparkFun Forums

[stone]

Hello,

I was wondering if I set up a gps and just left it there,
would it eventually find its position to within some
very tiny number, a centimeter or what?

Thanks.

If this is true, how would you do it? Would a simple average
do it?

[jremington]

The output of a GPS receiver wanders over a large range, depending on how many satellites are currently overhead, whether there are line-of-site obstructions such as trees or buildings, or signal reflections from nearby surfaces. The best you can do is average your position for a long time (days) and even then the average has some uncertainty AND inaccuracy, which gets smaller the longer you average. Beware of the difference between the terms "accuracy" and "precision", they mean different things.

In general you can't expect one centimeter accuracy from a simple GPS receiver, no matter how long you average. In some situations, you can't even expect one meter accuracy. A number of web pages describe various tests of GPS accuracy, you can start here: http://www.doylesdartden.com/gis/gpstest.htm

[stone]

Thanks.

I'm not sure I understand accuracy and precision, to begin with. But...

If you're still there, what is going on with something like a trimble GeoXH
type of machine? It claims to have decimeter/centimeter 'accuracy'. What does this
mean for this type of machine? Even with this type of machine, do you still
have to leave it in place for a long time to get to it's best accuracy?

I want to record the lat/lon/altitude of my property, to make a 3d mesh so
that I can enter this into a CAD program. Some of the gps equipment is
coming on to the market as used equipment, and it might become possible to
find equipment available to get my data.

Actually, if anyone knows of another way to get this data, I'd be grateful to hear it.
It doesn't have to be lat/lon, it's really just a relative 3d type of thing.

[jremington]

You can think of precision as the number of significant digits in the "answer" (location), and accuracy as how far off the "answer" is from the true value. A measurement can be precise to 1 cm but inaccurate by many meters.

The Trimble GPS units can receive both the Russian GLONASS and the American satellites and given more satellites, you can expect higher precision. However, my understanding is that to achieve full specified accuracy, the GeoXH and similar units also require access to a Trimble base station, with a very accurately known location, to be within 30-100 km. I believe they use the cellular network for access. You will probably have to dig deeply into the manufacturer's fine print to find out how well they perform when no base station or cell phone service is available.

You can probably hire a surveyor to produce a 3D map of your property, accurate to a couple of centimeters, for less than the cost of a full blown Trimble system. But if you have plenty of time, a clear, unobstructed view of the sky everywhere and are willing to average for at least a day in each location, a good WAAS-enabled commercial GPS receiver ($150-250) should give you a similar map, accurate to roughly 1 meter horizontally, 3 meters vertically.

[eccentric]

Actually, if anyone knows of another way to get this data, I'd be grateful to hear it.
It doesn't have to be lat/lon, it's really just a relative 3d type of thing.

The professional seems like what you want but with some work you could probably use the standard edition. Check out the pdf link to see the end mesh.
http://www.agisoft.ru/products/photoscan/standard/
http://www.agisoft.ru/products/photoscan/professional/
http://downloads.agisoft.ru/photoscan/s ... mple04.pdf
http://www.agisoft.ru/tutorials/photoscan
I hope this helps you in your project

[michelebavaro]

Hello,

I personally have done quite a bit of work with low-cost high-accuracy differential GNSS.
There are modules out there that when properly configured can give you the 3D ENU with centimetre accuracy within tens of minutes.
You can get them as evaluation kits or as bare ICs from distributors like the following
http://www.newark.com/nvs-technologies/ ... =lookahead

It depends a lot on the actual environment and specific application, but in clear sky conditions it is certainly feasible without spending the money of a professional receiver (especially when talking volumes).
You may want to take a look here:
http://michelebavaro.blogspot.it/2012/0 ... nd-up.html

Cheers,
Michele

[jremington]

@michelebavaro:

The NV08c-csm module looks interesting in that it can receive Glonass signals, but I noticed from your blog that you get cm accuracy by having a reference station at the University of Pisa. What sort of accuracy could the OP expect from the module, in the absence of a reference station?

I suspect from the documentation that it would be on the order of +/- 2.5 m, which would be similar to a readily available WAAS-enabled civilian GPS unit.

[michelebavaro]

If you need absolute accuracy then yes of course the coordinates of a base station must be known very accurately and measurements taken at that station must be available for your observation period.
One can use a base station belonging to some public/professional network (e.g. the University of Pisa or ItalPos in my case) or set up his own base station (with any other low cost carrier-phase-enabled receiver).
Measurements will be accurate to centimetre level relative to the base. So one could have a base station surveyed with 1 meter error (in an absolute frame) but rover positions accurate to centimetre level to that base.
I thought this would apply to your case, where you don't particularly care about the absolute LLH (Latitude Longitude Height) but rather ENU (East North Up) relative to some static reference point (e.g. a corner of the field).
In wish for absolute standalone accuracy, one shall look at PPP (Precise Point Positioning) techniques which usually - as you correctly guessed - require fairly long observation times.

Cheers,
Michele