HolyBro Pixhawk 4 mini NED velocity bug report in VX and VY


We’ve discovered some unusual behavior in the Pixhawk 4 mini’s horizontal velocity estimates. To call the problem a bug is generous. It’s a full failure with no clear indication of the problem. We are investigating.

Has anyone else experienced this? The velocity in VX and VY in the Earth-fixed NED frame (North-East-Down) is the problem.

To make sure this was not an error in our code, we took the Pixhawk4 mini on a drive in a car with known speed (40mph or 17.9m/s) on a road with almost perfectly aligned drives in the North-South and East-West directoins.

These slides were reported to HolyBro and we are awaiting a response.

The error seems to be in the LOCAL_POSITION_NED message. The GPS_RAW_INT message is good.

These VX and VY estimates are pretty much wrong. How can this be fixed? Is there a firmware update? Can we just change Kalman filter fusion settings?

The test code is available,
Marc Compere

Slide 2 of 9:

Slide 3 of 9:

Slide 4 of 9:

Slide 6 of 9:

Slide 7 of 9:

here are two more images from this test report.

testing scenario on a car sunroof. smooth drive in a smooth car on a smooth road:

time variation from Pi’s time.time() and 4 times from the Pixhawk: