I was testing the Oak-D for non-GPS navigation. For initial testing of odometry, I flew the drone entirely in AltHold mode. However, during the flight, it suddenly stopped responding to manual controls from the transmitter and drifted away in the direction of the wind, eventually crashing.
I also switched the flight mode to LAND, but instead of descending, the drone gained altitude and flew away.
Note: The drone was also equipped with an optical flow sensor. There was no GPS onboard. I was using ELRS Mavlink for both radio control and telemetry.
Here is the time from ardupilot web tool from where the issue started :
Hey!! I seem to have lost control over the throttle, elevator, and aileron channels, while the other channels appear to be working normally.
The issue began when EKF errors were observed. Could you please help check this?
For context, I was using ELRS over the MAVLink protocol for the RC connection.
What you are asking me to check doesn’t make sense. The EKF errors have nothing to do with your RC link. Or at least they shouldn’t have. You were having EKF problems all the way through the log. That isn’t unusual with what you are trying to do.
Yes, you’re right — the RC link and the EKF are separate concerns.
I was actually asking about the EKF issue specifically, to understand what might have gone wrong.
The question about the RC link is a separate one. I’ve never experienced any issues with the RC link before, and there were no “RC lost” messages shown.
Additionally, even after switching to Land mode, the vehicle continued to climb in altitude.
I have a separate AUX switch for Land mode on channel 12. Although it was in Land mode for a short time, the drone continued to climb in altitude and eventually crashed. To locate the crash site, I kept switching modes to trigger the buzzer sound and help me find it.
Also in the graph there is continuous data for all the channels. So it must be controllable manually but it didn’t : (
Yes, it was indeed a tough day : (
What can I do to ensure that the same thing doesn’t happen again?
I want to continue my work with the external nav…
I’d recommend setting up a repository to document your setup and a simulation environment with Gazebo to test VINS fusion. This will help you avoid expensive hardware crashes and let others reproduce your work without having hardware.
You can also fly off of EKF2 and feed data into EKF3. look at the alignment between the two solutions, and don’t fly off EKF3 until you know it agrees.
