PX4 Multirotor Oscillations on Roll/Pitch/Yaw Immediately After Takeoff – Disarm Not Possible Because Vehicle Still Considered In-Air

PX4 Autopilot
1

We are testing a large custom-built quadcopter with a motor-to-motor diagonal of about 1.5 meters (quad X-configuration, see attached diagram).
We are still in the development stage and currently working on stabilization.

A few flights have already been carried out. In previous flights the drone was able to fly, but we observed noticeable oscillations (although it never flipped). Today, during another test flight, we tried again — the only change compared to the previous flights was adding a vibration damper plate under the autopilot.

Unfortunately, this time we encountered a critical issue:

Right after takeoff, the vehicle started to oscillate heavily on all axes (roll, pitch, yaw).

We immediately attempted to land, but the system did not detect landing properly, and PX4 still considered the vehicle in-air, so the disarm command was not accepted.

In the 3D replay you can see the drone tipped over (rolled to the side and flipped).

Frame size: ~1.5 m motor-to-motor diagonal

CG not perfectly at the geometric center due to payload distribution

Autopilot mounted slightly away from CG and isolated with a vibration damper

GPS: UM982 dual-antenna RTK module with heading

Flight controller: [CUAV V6X / Holybro 6X etc.]

PX4 firmware version: [e.g. v1.15.x]

Flight mode: [Position]

Attached log file from QGroundControl

Questions:

What could be the root cause of these severe oscillations? Could it be related to PID tuning, vibration, or sensor bias?

Could you please check our logs and share your interpretation?

Any specific suggestions for tuning or sensor setup for large multirotors with off-center CG?
First flight LOG without crash
Last flight LOG

2

Most likely PID tuning. Check the PX4 docs.

And presumably the way you mounted the flight controller. If you are using the CUAV 6X or Holybro 6X, these have internal vibration isolation, so don’t add more when you mount it.

3

Yes PID tuning is likely to help. Set K-gain to 0.25 orso, wait for a day with (very) little wind and see what happens. This is under the assumption the frame is stiff and IMU is properly mounted as Julian suggests.

What you describe is why I like to use a three position switch for arming, when I use a RC.
Kill-Disarmed-Arm
This switch can kill the drone any time, after some flights I add a restriction that Kill only works for low stick. It takes some fiddling in OpenTX to get it to work this way, but it is worth it!

4

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.