Why there is a terrible drift in my GPS?

🧪 Flight Testing & Log Analysis
1

Hi, All
Something really strange with my quadrotor, I need some help.
My log file shown as bellow:

crash:(arm- > test RC -> switch to AUTO.TAKEOFF -> everything is fine -> fall down -> kill)
http://logs.uaventure.com/view/usPigkFbcBkbUviVG9FGmJ

success:
http://logs.uaventure.com/view/rfZQksKdVfmuKFZ3AxWG6P

What I did is, first arming the throttle, second waiting GPS ready for a few seconds, then I switch to TAKEOFF mode, everything is OK but suddenly my quadrotor fall down so I have to push KILL switch.

My quadrotor did not fly away, but when I check log files, the GPS status looks very strange.

Why there is a terrible drift in my GPS, and is there any wrong in my configuration?

Any advice will be appreciated, thanks…

2

based on the large acceleration jolt it looks like you hit the ground at 16:46, which is when the GPS drift started happening. This is common to see GPS go a bit nuts on a hard landing. Also if it landed under trees then that can also mess with the GPS.

1 Like
3

@clovett Thank you, I will try again later…

4

@Boss_Kwei,

Some observations on your flight:

  • Altitude Estimate between 16:46:43 and 16:46:48 the GPOS.Alt snaps down after a seeming runaway
    • Q: What is the difference between GPOS.Alt and GPS.Alt?
  • Roll Angular Rate: At around that same time, the roll angle begins to oscillate. The oscillations grow, attenuate, then grow until the vehicle crashes.

I would suspect an unexpected measurement caused instability in a x-gyro bias estimate and the closed-loop control finally lead to a crash.

5

@hamid-m
GPOS.Alt is the estimated altitude from the EKF
GPS.Alt is the measured altitude from the GPS sensor

6

Thanks @CarlOlsson. The EKF altitude snapping back is highly suspect and it seems things did not recover after that.