Normalization of Euler angles

Sensors IMU & Motion
1

Hi there,

I’ve found a bug connected to compute euler angles:

sometimes I’m getting some of Yaw/Pitch/Roll angles which is greater than 359, for example 655. That’s well known problem associated with radian-degrees conversion (angle can be -655 or 1028).

I’d offer you to make following changes in SparkFunMPU9250-DMP lib:

Instead of:

  if (degrees)
  {
    pitch *= (180.0 / PI);
    roll *= (180.0 / PI);
    yaw *= (180.0 / PI);
    if (pitch < 0) pitch = 360.0 + pitch;
    if (roll < 0) roll = 360.0 + roll;
    if (yaw < 0) yaw = 360.0 + yaw;
  }

portion of code in computeEulerAngles method, to use:

  if (degrees)
  {
    yaw = NormalizeAngle(yaw * RAD_TO_DEGREE);
    pitch = NormalizeAngle(pitch * RAD_TO_DEGREE);
    roll = NormalizeAngle(roll * RAD_TO_DEGREE);
  }

Where recursive NormalizeAngle method which normalizes angle to 0-359 range :

float NormalizeAngle(float angle)
{
  return angle > 360 ?
         NormalizeAngle(angle - 360) :
         angle < 0 ? NormalizeAngle(angle + 360) : angle;
}

and pre-calculated RAD_TO_DEGREE (why to calculate division of constant values for this coefficient everytime triply in computeEulerAngles?)

static float RAD_TO_DEGREE = 180 / PI;
2

Seems described problem (pitch can be 655 degrees) strongly connected to [to this bug in library code, computeEulerAngles method. And this bug is still no fixed since 2017 year!.

Just replace this string:

pitch = asin(t2) * 2;

by this:

pitch = asin(t2);
``` to solve problem with incorrect pitch. Somebody will think about drift of pitch, somebody about angles normalization, waste time to find solution... But Sparkfun keeps this bug unfixed :) Bravo!](https://forum.sparkfun.com/viewtopic.php?f=14&t=45516)
3

randyrover, thanks for the suggestion! Will you consider authoring a pull request on the GitHub repository?

https://github.com/sparkfun/SparkFun_MP … no_Library

How to make a pull request:

https://help.github.com/en/github/colla … ll-request