search

mrbubble62 / GyroCompass

NMEA 2000 style CAN Gyro Sensor
MIT License
9 stars 2 forks source link

Drift issues #4

Open erik-rosen opened 4 years ago

erik-rosen commented 4 years ago

Hi there,

Would love to dig deeper into the drift issues you mentioned here.

Do you have any idea what the drift issues could be caused by and how does it present itself? Changes in bias/scale factors of the sensor itself over time? Changes to the soft/hard iron effects in the environment?

I'd be curious to understand why the SIMRAD Precision 9 compass performs better than the MPU9250. Could it be using a more sophisticated sensor fusion algorithm, perhaps including GPS with IMU data? Superior hardware?

Thanks in advance, Erik

mrbubble62 commented 4 years ago

I would guess it is the proprietary sensor fusion.

erik-rosen commented 4 years ago

My Axiom MFD publishes velocity and position which most likely can be modelled as 0-mean Gaussian noise. If I include that in the update of a kalman filter, I think we could add the magnetometer sensor bias to the state vector to handle changes to the bias over time and remove the need to turn the sensor in all orientations. My guess is that that is what the Precision 9 is doing as well.

What do you think?

Sent from my iPhone

On 15 Aug 2020, at 16:08, mrbubble62 notifications@github.com wrote:

 I would guess it is the proprietary sensor fusion.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.