I’m having a lot of EKF variance/ground mag followed by yaw-realignment errors and I’d love some insight into ideas about why they’re happening. I know they trigger when, among other things, IMU1 and 2 readings don’t align and/or GPS signals disagree with estimated position/heading, but I’m trying to figure out if I’ve got a hardware problem or there’s something else going on.
Logs 38 and 51 are the big crashes; log 37 was a successful flight with a completed mission, but the same EKF variance error/ground mag anomaly. We took off in stabilized mode and switched to auto. We’re flying close to some bleachers and don’t have great satellite numbers, so signal reflection or poor estimation could definitely be a problem, but I’m not clear as to whether that’s the only/primary problem since GPS and HDOP seem reasonable throughout.
That said, the innovation estimates fort the magnetic field spike like crazy when there’s a realignment. After EKF variance errors, there’s almost always an increase in velocity.
We’ll update tomorrow with tests
- done in a wide open field away from any structures
- compass calibration done onsite
But any help in the meantime is greatly appreciated. Should point out that for some reason the there’s no IMU data for 38… not sure why – it’s there for 37 and 51.
My drone is a quadcopter with a the Pixhawk 1; arducopter 3.4; GCS = qgroundcontrol; analyzed in APM Planner 2.0.