Sergei Grichine

I am having the same problem on a ground vehicle (gas powered lawnmower) - something that hasn't been happening half a year ago. I have an 9250 IMU which I...

Here's a log with EKF2 heading jump at mark 480 sec  [log11.ulg.zip](https://github.com/PX4/PX4-Autopilot/files/10100697/log11.ulg.zip)

BTW, I am using an RTK GPS, so when the vehicle is moving - GPS heading over ground is very precise.

I tried the same in SIM (I5 desktop, Ubuntu 20.04), no matter how much I turn - the heading stays good. 

My bad, sorry. Here are two logs, log11 is as described above, log12 shows similar heading jump. Let me know if you need different data - what would be the...

I tweaked EKF2_* parameters to disable fusing magnetometer data (as well as Ranger, EV, Barometer). The QGroundControl shows yaw 0 when px4 starts, and stays until the vehicle moves enough...

@ryanjAA : Fair enough - but didn't I actually disable compass by setting "param set EKF2_MAG_TYPE 5"? Before I did that - px4 started and immediately showed compass heading in...

I did more experimenting to expose the bug. I have two platforms - a real gas powered lawnmower and a smaller robot (see the photo). Both run latest PX4 main...

@bresch - Thanks for responding. I am not sure why in the "_Magnetometer-only fusion_" log you concluded that "_the heading gets pulled on one side by the GPS data during...

@bresch - Thanks for explanation, it helps. I will be happy to test whatever needs testing on my rovers, feel free to ask. There's a lot of potential in agricultural...