Pihawk 2.1 GPS precision claimed to be centimetric, is it true?

I would like to have mor information on how Pixhawk 2 uses GPS to get an RTK like precision on flight. Do we need more GPS units? Can we use USRT gps (old connections) and just change the connector?

RTK is gained form having a reference source on the ground sending corrections to the copter.

So you will need a GNSS unit on the ground and a link to the copter.
To get centimetre accuracy you will also need a multi frequency GNSS.
There are some claims that single frequency GNSS can get this accuracy but I have learnt not to mention that near GIS experts else I am in for an hours tirade on just why it cannot be done.

1 Like

And what about DGPS. Fixing distances between Several GNSS units fixed in a mobile plattform (DRONE) and using an algorithim to determine the correct position of the drone? By the way Pixhawk 2 claims to have centimeter level precision. Without GNSS how does it do that?

Damn good question.

I am waiting to see this in reality.

We are currently getting +/- 6mm using RTK with a multifrequency Septentrio AsteRx-m UAS GPS.

Hope some generous soul answer our questions.

Hi All
Lots of cool stuff happening in the world of RTK

No one is trying to say that a $300 GPS will meet the accuracy of the likes of Trimbal triple frequency systems etc… However, they are many thousands of $$$$$

Without as Mike mentioned, getting into a detailed debate on how this works on L1, lets just say, it does. (even Trimble sell them… http://www.trimble.com/gnss-inertial/bd910.aspx?dtID=feat ) I’m not going to spruke it… People will be getting their hardware soon, and once they do, we will see real data flowing.

the key is to determine, exactly what are you wanting from your system, and what are you getting…

DGPS, and RTK are related but different… DGPS just sends assumed position error based on a fixed ground station… RTK corrects for timing and position errors in the signals themselves.

1 Like

Ok, I believe the technollogy is there. Assuming it will work better than regular GPS, still there are 2 countermeasures needed, that are not so expensive, that would enable a “better” performance: a second unit and the second power module. Tell me where to buy this set and you have a sale. At least I will try.

any of the resellers listed @ www.proficnc.com/stores will be able to help you

Hi All,

Great conversations!
Just wanted to drop in my 2-cents.
I am a Land Surveyor in the U.S. and have been using Trimble, Topcon, etc. GPS/GNSS systems since the late 90’s.
I can assure you that ALL of the L1 ONLY systems I have been using over the past year can achieve the exact same accuracy as the Trimble, Topcon, etc. $10k+ units. But you must know the limitations of using L1 only. There are situations where the multi-frequency receivers are superior to the L1 only systems. In the air, on a drone, L1 works great!

Keep in mind, there is also a LOT of new development going on in the L1/L2 GNSS world as well. For one, check out Tersus… I know Piksi has a similar L1/L2 system too…

What I do know is that I will NOT be buying any Trimble, Topcon, etc. $10k+ GNSS units in the foreseeable future…



I have another inquiry. Even if we use super precision RTK, GNSS, etc on the drone, it is necessary to mark the terrain using a regular RTK. The image geotag will enhance the final result, but is not enough to create precision mapping, right?


At least a couple to verify geotagged information.

At present it takes a lot more GCP’s, more than a dozen, to get real precision.

I am compiling information at the moment from my GIS sources that Mission Planner and Arducopter are not yet ready for true centimetric RTK.
Will be posting when I have definitive evidence of what needs fixing.

Hi folks,

Does anyone know which is the GPS chipset used in Pixhawk 2.1? If I am not wrong, the one in Pixhawk 2 is the newest Ublox M8N.


@Fabio What do you mean by “regular RTK”? You can set GCP’s with the same L1 ONLY if you want. Also, you can get within around ~4cm accuracy without any GCP’s. Add a few GCP’s and it will increase the accuracy.

@mboland Have you done real-world testing? We have and also compared it to others that have got similar accuracies.


1 Like

By regular RTK I meano not on board, with higher precision.

Hi Miquel, it can´t be an M8N alone, should be sometihing else. (I am not claiming to be 100% right on this), but near perfect accuracy is achieved through military grade positioning systems and would make no sense to open to the general public for safety reasons. Better accuracy can be achieves through several other tecniques. Altough M8N is good, alone it is not able to achieve centimeter level accuracy claimed here.

Some interesting points here…

The GPS will come in 2 varieties

RTK (M8P) and standard (M8N)

Rather than compiling your concerns for a post… why not list them in the issues list, and work with Michael Oborne and Tridge to make the improvements.

You may be shocked at how fast features are added when the issues are clearly marked out… that’s the benifit of open source!


You might be interested in our recently launched PPK GNSS L1/L2 precise positioning solution for photogrammetry without Gound control. Please have a look and if you have further questions I’d be more than pleased to answer your questions. www.drobit.es

@proficnc Good Idea, I will include this in the issues list.

I have added the issue of limited decimal places for the position accuracy as 1e7 can only give you +/-1.5cm at the very best. To get sub cm we need at least 1e9.

But it is probably a good topic for debate as it will mean some internal changes in the arithmetic in Ardupilot code, and it’s overall worth.
The copters movements are never going to be this accurate, so why go for more accuracy than can be practically used in controlling the copter, other than for use of the logged position.
In most ‘high end’ mapping situations, the RTK board is going to have it’s own logging facilities, and these would be used instead of the Ardupilots logs.

I have had it confirmed that the injected RTK corrections are not truncated at all and are passed as received, so this is probably the best solution, for us anyway.

hi, i want to know what situation where the multi-frequency receivers are superior to the L1 only systems. now i use the oem615 of Novatel L1/L2 GPS Glonass. I have a doubt of the accuacy of L1 only.