Hey guys, we wanted to share a visual demonstration of  RTK GPS.  The demo was made with NAVIO board and Raspberry Pi running RTKLIB. Plots display how RTK GPS solution compares to a single GPS solution from the same receiver.

RTK is a type of differential GPS where two receivers are used - one is stationary and acts as reference station by calculating corrections and sending them to another receiver - rover, that can apply those corrections and increase the accuracy of coordinates.

Corrections in RTK are calculated based on carrier phase measurements and resulting coordinates can be accurate to millimeters. But unlike code cycles, phase cycles are indistinguishable and that creates a problem of integer ambiguity.

In RTK terminology there are different types of solutions:

  • Fix (green on plots) - Solution with fixed integer ambiguity (centimeter accurate).

  • Float (yellow) - Integer ambiguity is not fixed (accuracy floats, may be decimeters).

  • Single (red) - No differential corrections applied (meters accuracy).

RTK is a well-known technology in the field of geodetics, but commercial systems could cost a fortune.

NAVIO is equipped with a special version of u-blox NEO6 receiver - NEO6-T, which outputs raw GPS measurements. And as RTK is differential GPS - two receivers should be used, but one of them could be replaced by corrections from the internet.

We have conducted several tests with NAVIO boards. The antenna was placed on the roof of a vehicle and kept static. 

After that it was driven in a circular path.

RTK is an advanced feature and multiple factors influence quality of the solution - antenna, RF interference, satellite visibility. To keep track of the latter, we usually check ourselves with SatPredictor tool before going to the field. It creates a nice plot with quantity of visible satellites based on your area, date and elevation mask (common rule is to set it to 15deg).

We have different antennas, but usually use Tallysman dual-feed patch as it has very low noise figure and high gain. A rule of a thumb for good RTK performance is to have several satellites with SNR (signal/noise ratio) around 50. A good idea is to have the base permanently installed on the roof of your home or on the roof of club building at the field and broadcast corrections over 3G. This way you can share the base station with your friends and colleagues. Permanent installation would result in a more reliable and predictable satellite visibility. In some areas RTK corrections are already available from universities or other organizations, search for NTRIP servers nearby.

P.S. By the way, float solution does not mean bad solution.

We’ll post a tutorial how to set up RTKLIB soon, you can subscribe to our blog emlid.com or follow us on twitter to stay informed.


Views: 11486

Comment by Meez on April 15, 2014 at 10:56pm


Comment by Andrew Rabbitt on April 16, 2014 at 12:01am


Comment by Martin on April 16, 2014 at 12:18am

Very nice. I know that it's just the beginning but It would be good to show the performance with respect to a reference GNSS receiver (e.g dual freq, surv. grade) and also demonstrate its performance under some dynamics. I am afraid that the on a drone you will have problems to be fixed during the flight due to the flight dynamics and lower SNR due to the local disturbances from UAV's electronic.

Comment by Christiaan van Vollenstee on April 16, 2014 at 12:55am


Comment by Emlid on April 16, 2014 at 4:34am

@Meez @ Andrew Rabbitt @Christian van Vollenstee

Thank you guys!

Martin, actually survey receivers are usually are expensive because they are L1/L2, but on short baselines <10km there should be almost no difference between L1 and L1/L2 solutions, if distance to the base station is larger L1/L2 will outperform single frequency system.

We did some tests on a flying platform with GPS postprocessing, but with forward filter only performance should be almost the same as RTK. And the solution was fixed during the flight except for the takeoff stage. The trickiest part is right antenna choice and placement.

Monroe, it would be awesome to see. We have thought multiple times that NAVIO could be used for simple precision agriculture and things like that.

Comment by scampos on April 16, 2014 at 4:40am

In professional usage, and remote areas, we cannot rely on 3G. Do you have NAVIO RTK samples with 2 GPS?

Comment by Greg Dronsky on April 16, 2014 at 4:44am

Awesome! Just when are you going to sell these? 

Comment by Emlid on April 16, 2014 at 4:58am

Stefhan, in this demo two GPS units were used, connected over 3G network but you can use any other kind of radio link - for example Xbee.

Greg, thank you! We have to finalize some organization issues, but we are trying hard to start preorders in a couple of weeks.

Comment by Charles K Taylor on April 16, 2014 at 8:20pm

I am interested to see how this system works compared to something like Novatel, which I use everyday for RTK solution.  How much do you expect this system will cost?

Comment by Emlid on April 17, 2014 at 4:39am

Charles, we are not trying to compete with survey grade receivers, what we want to do is come close to their performance at a fraction of a cost and in the same package with autopilot. The NAVIO board will cost 165$ and you will also need a Raspberry Pi (25-35$). And assembled it will look like this. 


You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2017   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service