problems with headings

[Nocentinia]

Good morning,
I'm using the MKR IMU mounted on a boat, with the aim of studying various motion behaviors.
I noticed that heading read by Euler Angles is very wrong, even 180° so it is extremely unusable as it is.
Is there any way to fix the situation? maybe changing how to use the sensor?
I'm desperate, please help me :( .
thanks in advance for your support

[AnalyticEquestrian]

I'm pretty late to this question, but I came here with a similar issue, which is not an imperfection in all fairness. When your Euler angles are off by 180 degrees and/or erratically jumping around at a certain orientation, you're probably experiencing gimbal lock. As noted here, quaternions provide the way out. I ran into this myself and created a fork of the MKRIMU lib with quaternion support (#29) based on the Adafruit lib, which did the trick for me.

Note that the BNO055 gives you a right-handed quaternion, converting this to your application's coordinate system can be every bit as maddening as gimbal lock. I'm sure smarter people are able to reason out how to map your axes, but eventually I gave up and just brute-force tried all options for mapping the xyz components and their signs, until one mapping worked halfway through.