Moving baseline, modules and reliability

Hello all,

I have some questions with regards to GPS for Yaw functionality to understand it better.

  1. I see on ardupilot website recommended modules to use but is there any reason why other commercially available modules would not be good for the use if they are based on F9P? For example Sirius F9P from drotek?

  2. Is somebody using this functionality without RTK data from base? Is it visible difference in terms of heading?

  3. Is it possible to use it as a backup to magnetometer or is it just one or another? If it is possible what is primary? magnetometer or GPS Yaw?

  4. What about reliability of this solution? any tests to links?

Sorry for number of questions i would love to test it myself but first need a little more information.

Thanks in advance!


I’m sure you’ve seen the GPS-for-yaw page but just in case and also for others reading this post.

I haven’t actually used GPS-for-yaw myself but here are some answers based on what I’ve heard:

  • I think that almost any pair of F9P GPSs will work.
  • There is configuration to enable falling back to compass if it loses the heading from the GPS. One thing to remember though is that because the EKF always has the IMU’s gyro it knows the vehicle’s rotation rate so even with a yaw sensor failure (compass, GPS-for-yaw, etc) the heading will still be very close to the actual heading for quite a long time (at least a minute)
  • reports have been good with 4.1. There are a number of users over in the Rover forum including Yuri Rage.

Thank You @rmackay9 for the answer.

So i understand that in this configuration that You mentioned the GPS for yaw is primary to Compass?
Is there a way to use GPS for YAW only in case of Compass Error ?

Yes I have read page in the documentation prior to asking :slight_smile: Topic of heading from GPS seems like amazing option but it is kinda new, it all started with amazing performance from new uBlox modules, and still hard to understand if it is better or more reliable than compass or it should be just a diversification option - especially in multicopters.


There’s no option to use GPS-for-yaw as a backup in case a compass fails although you could implement it using a Lua script similar to how transitions between GPS and Non-GPS sensor sources are handled.

I think if your vehicle is flying fine with a compass then you might as well stick with that. GPS-for-yaw gets good reviews so far at least from some Rover users.

I think you’d rather have GPS heading be primary with a compass backup, since as long as GPS heading is reported, it should be highly accurate. The GSF algorithm with compass input as a backup should fill any gaps in GPS reporting.

On a large multi rotor, I should think that a moving baseline configuration would perform very well.

If you are intending to do aerobatics and/or high pitch/roll angle maneuvers very frequently, perhaps it’s not a great use case, since the antenna distance will frequently collapse with respect to the satellite constellation.