EKF Lane Switch & Yaw Reset During Acceleration

ArduCopter Copter 4.5
1

Hi everyone,

We’re experiencing recurring EKF lane switch messages during flight on a 25 kg MTOW drone.

The drone uses a Cube Orange with one Here3 and one Here4 in GPS blending mode. The EKF lane switching consistently occurs during acceleration, and it coincides with yaw reset messages.

Despite this, the flights remain nominal — no compass, mag field, or vibration alerts are present.

We’ve run compass calibrations twice, but the issue persists. The drone flies well, but we’d appreciate any insights in case we’re overlooking something.
Attaching log file: https://drive.google.com/file/d/1MNWF_uXTLe9aa7oC9A5YzLIrbjndEopK/view?usp=drive_link

Thanks in advance!

2

IMU3 (The un-damped IMU in the Cube) is showing some really high levels.

image
image1036×551 90.6 KB

There are also some motors that are hitting their min value during the high speed runs and that might be causing issues.

3

Hi Allister,

Thank you for your reply.

Following your observation, I’ve installed a vibration damping module beneath the Cube Orange to address potential vibration issues.

Regarding the motors hitting their minimum values during high-speed runs — could this be related to the motors being slightly tilted or misaligned? If that’s the case, I’ve taken the time to re-align them properly.

Additionally, I recently updated the firmware to version 4.6. During today’s flight test, I noticed there was slightly less lane switching compared to previous flights, which seems promising. Attaching log: https://drive.google.com/file/d/1hx-bVf8GV4hOA8UcjvUyidHh4xBkL4dY/view?usp=sharing

Thanks again for your support, and I’d appreciate any further insights you might have.

Best regards

4

Just ran the log through mag fit. Try these:

MAGFit.param (1.0 KB)

5

Thank you for the suggested MagFit parameters. I updated them today and conducted another flight. Unfortunately, the EKF lane switch messages are still appearing, but only during acceleration phases.

I’m attaching the log from today’s flight for your review.

6

set

INS_HNTCH_REF,1 
INS_LOG_BAT_MASK,0
INS_RAW_LOG_OPT,9

The vibration levels are still higher than ideal, but not the worst. I’m also thinking that the filtering isn’t effective, so try the setting I mentioned above and see if that makes any changes. I’m not expecting it to be a revolutionary change, but it might help.

7

Re-evaluate the Vertical Acceleration Controller gains relative to hover throttle also. Not really to help with Lane Switching but before you collect data for Notch filter configuration.

And to @Allister point re-check MOT_SPIN_MIN.

8

Your RPM filtering is static without INS_HNTCH_REF set to 1 and fundamental frequency much lower than 80hz you have set (however because there is no noise on 1st harmonic it’s OK). I would setup your filters like this

image
image627×781 15.2 KB

Expected noise levels:
image
image483×465 20.7 KB

I’m still ardupilot noob, so if gurus can confirm my advice it would be nice.

You still have a bit of vibration at Z axis at 20hz but I afraid nothing can be done here, only mechanically find the source of vibration which IMHO are very strong.