Improve magnetometer measurements


I’m working on the magnetometer of my Zed2i.

I have calibrated it, in my garden so not much magnetic interference due to other devices.

Then, I recorded small SVO heading north, south, east and west. The camera was steady and did not move at all.

groundtruth average angle found
north (0°) 6.489484222730001
east (90°) 98.43067403940054
west (-90°) -99.79151725769043
south (180°) 152.7984662737165

Two questions :

  • why is the south so bad when the north was correct ?
  • how to improve the measurements ?

I suggest you try to orientate the camera to fill the holes in top left of the XY (Red) circle, and the center of the YZ (Green) circle.

I will make this test.

In the meantime, what is the expected accuracy of the Zed2i 's magnetometer ?

I mean, is it accurate with a 2° error ? A 5° error ?

This is a very good question.
Please do not compare the output of the ZED SDK with a digital compass application running for example on your smartphone.
The magnetometer measures the magnetic heading, not the true heading, the same heading measured by a real compass.
The smartphone applications instead output the true heading, calculated by fixing the magnetic heading with GPS information.
For this reason, it’s not possible to reply to your question with a numeric value.

We are well aware that the magnetic and geographic poles are two distinct things.

But when compared to the magnetic pole (the one we are supposed to find with Zed’s magnometer), how accurate are we ?

1 Like

This information is not currently available. We can try to export it.

I would really love it, if you don’t mind.

By the way, I assume if we try to use the magnetometer in a car (with the EM field from tires, battery, …) the accuracy will decrease. Do you think we might be able to calibrate the magnetometer ?

Hi @Francois
Yes, you can calibrate the magnetometer also inside the car.