Dear Parrot Team,
FYI, the heading field of the GLOBAL_POSITION_INT messages is seemingly improperly implemented in the Malvink interface. It is locked onto 0 in a 120 degrees range, and goes from 0 to 180 in the remaining angular range. Magnetometer calibration was executed multiple times, with no change to this behaviour.
As you can imagine, it is a somewhat of an issue if someone wants to jump between body frame and NED frame, and use. I know that the VFR_HUD has a similar field (which does work, so congrats on that), but it is broadcasted at only 4Hz, which is much less that the 10Hz of the global position message.
That might be something that deserves a fix for a military product which is advertised as being compatible with the Mavlink protocol. At least perhaps documenting it in your post describing supported Mavlink commands could be decent.
I won’t elaborate on the overall quality of the Mavlink interface, but this one bug in particular sounds like a significant problem.
Regards and happy unit testing,