Large EKF errors with SD Flight and v1.7.0-rc1

Can anyone tell me what is going on here? I’m running PX4 on a sensor to get an EKF solution so it’s just attached to a multirotor. Why is the GPS good but the EKF way off?



  • Do you have the module mounted isolated from vibrations?
  • Could you try a test when you walk around with the multirotor without powering it, except for the px4 sensor, in order to see the performance of the ekf without vibrations and mag interference?

Did you do the mag calibration indoors or on the field?

the vibration is looks good. not too large

Appreciate the help. I’ve been wracking my brain around this for weeks.

I recently browsed the public logs and found this flight so I know it should work.

The mag was calibrated outdoors and the yaw looks pretty clean. Here is a plot of the yaw overlayed with the cog. Is my rotation incorrect?

Here is a link to just walking around with the rotor/sensor outside.

I know the hardware works b/c it runs fine on v1.5.4. Here is a link using v1.5.4.

From v1.5.4 to v1.7.0 is there a new EKF check where the system has to be in a certain arming_state or in a certain nav_state to find the correct solution?

Also are you flying EAGLE’s? What version of BSP and qrlSDK add on are you on?

Another log just walking around outside.

What kind of platform are you running px4 on? The sensor data looks very strange
Here is a hand held test from a pixracer

And here is one of your hand held logfiles

I think I’ve solved the EKF solution problem. Some of the modules were commented out. More specifically the navigator, dataman, land_detector, mc_control, and mc_att, Do you know which modules directly affect the EKF?

Good flight.