Magnetometer calibration goes to 0 (BSXLite application on BMF055) while walking in a building

[janakiram]

We have BMF055 shuttle board and extension board.    We use BSXLite Integration application.     Host is raspberry pi.     We calibrate BMF055 until the valibration value is 3.    After that we carry the board (shuttle board+extension board+ raspberry pi) around in our building open area.    But, the calibration value moves 0.    There is nothing nearby for atleast  one meter.     Anything that we should be aware off?

[janakiram]

Hello,  Any help would be appreciated.    We have a mobile robot which goes through passages in a building.    We wanted to get absolute orientation at any given moment.     We are using BMF055 to get absolute orientation.    We use BSXLite application.    After calibration, we are getting accurate heading information.    But, when we are going through some passages within a building, heading accuracy quickly degrades and sometimes the calibration accuracy goes to zero.    Any help would be appreciated.

[Vincent · Accepted solution]

The magnetic sensor are highly affected by the environment. 

After calibration done, the calibration status for magnetic sensor will go to 3.  but when any distortion happened afterwards, the calibration status will drop to 0.  then recalibration is needed to achieve 3 again.  the calibration process for magnetic sensor are doing always in background in BSX lite lib.

In your case, the robot cross the building, then the magnetic environment are obviously changed during moving.  So the calibration status will be drop to 0 as expected.   When the magnetic environment changed, the heading accuracy will also be affected if calibration can not achieve to 3.

[janakiram]

Thanks Vincent.    Since the magnetometer would be built into the BOT, i am wondering how to calibrate  when the magnetometer calibration is gone.     We cannot make complex movements with the BOT to calibrate.     If we write the old calibration profile, i guess the heading will not be accurate.    Is that right?   Any suggestions to easily recalibrate?

[Vincent]

we provide FMC mode which is fast and simple calibration method for magnetic sensor.  so if you use this mode (actually default NDoF mode),  you don't need complex calibration movement. 

But the device still need to move certain angle in the air.   if the magnetic sensor is in the body of the robot, then it will be a problem for you i think. 

Even you use the old calibration file, the heading accuracy is not guranteed unless the magnetic environment is same as the saved profile you loaded.  this is difficult to identify if there is no calibration performed.