AbsoluteIMU-ACG - a problem with a compass calibration
 Software and Programming Related   Started by Marekk S   2015-09-11 14:54:32 -04:00   Comments: 3    Viewed: 571

  1. Marekk S
    Marekk S Member

    Postby marek.kaminski.1112 » Wed Aug 05, 2015 5:36 pm

    Dear All,

    I have a problem with calibrating a compass of my AbsoluteIMU-ACG, so I have several questions and would like to ask you for advice.
    ( that's unfortunately continuation of my previous story: http://www.mindsensors.com/forum/topic/51_problem-with-absoluteimu-acg-compass )

    My attempts of calibrations result with a compass for which 'north' is almost anywhere :-( (depending on direction the compass is pointed).

    Sometimes 'north' seems even to follow my movement of the compass. There are even small angle areas where the compass 'north' jumps suddenly tens of degrees.

    The best result I achieved resulted in a range of 'north' spanning on the area of about 90 degrees.

    Well, I'm unhappy as I'm afraid I had better results before doing any calibration :-(
    Is there a way of restoring a factory calibration?

    I'm not sure if I do the calibration properly.

    Am I right that the procedure for calibrating a compass should involve pointing all three axes (x, y, z and rotating clockwise and anticlockwise) in both directions: first vertical and then horizontal?

    That's unclear for me as it looks like one user of the IMU ( http://www.mindsensors.com/forum/topic/45_absolute-imu-compass-gyro-and-accelerometer ) skipped a second half of calibration at all.

    For me the sequence of the calibration steps in guide is as it follows:

    - X vertical (i.e. X arrowhead points up), rotate my lego tank/excavator-like platform (standing on a table) cw, than ccw (middle of rotation is on the X axis)
    - analogical: but Y vertical, rotate platform cw, than ccw
    - analogical: but Z vertical (printed surface of IMU was faced down), rotate platform cw, than ccw

    For the next half I rotated the whole platform (so that its turntable was placed vertically, not horizontally as for the previous half of the calibration):

    - X horizontal (printed surface of IMU was faced up), rotate along X cw (while looking on X arrowhead from right), than ccw
    - analogical: Y horizontal, rotate cw, than ccw
    - analogical: Z horizontal, rotate cw, than ccw

    I used gearing resulting in one rotation lasting about one minute. Is that too long / too short? And if so, what rotation speed is the best?
    In each step I've performed a little bit over the full rotation. I've done also attempt with two full rotations per each step.

    And I'm also wondering about how the device excludes from sampling the moments while I'm rearranging the device to a new position.
    Should I introduce some delays? (and again: how long should they last then) Or: what moves shouldn't I do to not disturb the collected data?

    I was using IMU-x-cmps-calib.nxc. Looking into the code tells me that such a thing like notifying the device that I've changed the axis/direction during the calibration is not implemented.
    That collecting all the data is done as a one step, so that (that's my conclusion) the device should recognize itself that axis/direction was changed.

    I calibrated the device inside my flat (inside a block of flats). The platform was high, the device was on its top, so I hoped it was far enough from relevant magnetic sources.
    Possibly I was wrong. Should I calibrate it outdoors?

    Also one thing wonders me (as I calibrated also accelerometer and gyro):

    - there's written in the guide that tilt should read 0 (on all axes? - that's unclear) while the device is lying on the table.

    I have zero on two axes, but on the third one I have 126/127. My intuition / understanding of the things tells me that's correct. Just wanted to be sure.
    As far as I remember, the tilt readings with a factory calibration was for the tilt (more or less) the same.

    And the last question:

    - is it possible that something has been damaged inside the device during the air transport?

    Thank you very much for all the answers,
    Best regards,
    Marekk S, 2015-09-11 14:55:00 -04:00
  2. Mindsensors Support
    Mindsensors Support Member Richmond

    Postby support » Wed Aug 05, 2015 6:35 pm

    here I've uploaded a short video of calibration process : http://youtu.be/YjRLHsP09nY
    Ideally you should mount the AbsoluteIMU on your robot and run this process.
    (so that it is calibrated in it's environment), but for simplicity of video it just shows calibration of the IMU alone.
    Mindsensors Support, 2015-09-11 14:56:02 -04:00
  3. Marekk S
    Marekk S Member
    Postby marke.kaminski.1112 » Thu Aug 06, 2015 8:00 pm

    That's great!

    Thank you very much for a prompt reply. The video was very helpful and explanatory.

    After my first and manual test, following your video, the reading of the compass was much better :-)

    Best regards,
    Marekk S, 2015-09-11 14:57:44 -04:00
  4. (You must log in or sign up to post here)

Log in | Sign up
Tems & Conditions!