›

Euler angle outputs have discontinuities

kristen_bkristen_b
in iOS
Hi there,

I'm streaming Euler angle outputs from 2 MetaMotionR sensors (via iphone6) during a simple 1DOF rotational sinusoidal motion. 
The outputs seem to producing discontinuities (I'm assuming from switching quadrants during the rotation motion). The jumps occur in pitch, roll, and yaw - pitch produces jumps to just below 180degrees when fluctuating around 0 degrees, roll (the dominant axis of sinusoidal motion) inverts the portion of the curve that is below -90degrees, and yaw produces jumps around both 7degrees (drops to ~0degrees, then sometimes up to 360degrees, and back up to 7dergrees.

I'm familiar with dealing with discontinuities in IMU data using the unwrap function in matlab, but because all of the jumps are just below pi or 2pi, the function doesn't work. I'm wondering if you have any suggestions or solutions when either collecting (preferred), or processing the data to eliminate these issues. 

Thanks!

Kristen

Comments

  • EricEric
    edited November 2017
    • Can you post some data that shows the discontinuities?
    • What settings are you configuring the sensor fusion algorithm with?

    A quick suggestion is to try streaming quaternions instead then converting those values to Euler angles in your app.
  • kristen_bkristen_b
    Hi Eric,

    Here is a link to some of the data that shows these trends. I was collecting Euler angles at 100Hz from 2 sensors. 


    I have tried streaming quaternions instead and still get these issues even after converting to Euler angles. 

    With that being said, I rotated the sensor 90 degrees (so that the primary axis of rotation was y (not x)) and it eliminated most of the issues. However, I don't see this as being a permanent solution. 

    Thanks for your help! :)

    Kristen
  • EricEric
    Can you point me to specific lines of interest?  I took a quick glance at the file and noticed some pitch values going from ~4 degrees to ~176 then back to ~4.
  • GtracksGtracks
    we see that also. Especially in some orientations depending on how the sensor is placed. Attaching a screen shot here.
    image

    also the file that has the euler angles is here below (please see second Tab)
  • EricEric
    Transitioning to/from 0 and 360 degrees is expected; the yaw data in the second sheet looks fine.  Also, your picture is not visible.

    The OP was experiencing jumps occurring before the pi/2pi mark.
  • krakra
    Hello Kristen,

    I have had the same problem - I'm not sure how to deal with these 180 and 360 degrees transitions... I was just wondering if you managed to solve the problem and how if so?

    Thanks,
    Klaus
This discussion has been closed.