Fix errors in rotation function in OSMPDummySensor

[ClemensLinnhoff]

Reference to a related issue in the repository

Fixes #94

Add a description

• Added a transpose function for the rotation matrix, as it is a transformation from a global coordinate system to a local system and therefore, the inverse needs to be used
• Changed using the objects orientation to using the ego orientation for calculating the relative position of an object with respect to the ego vehicle.

Check the checklist

• I have performed a self-review of my own code.
• I have made corresponding changes to the documentation for osi-sensor-model-packaging.
• My changes generate no new warnings.
• I have added tests that prove my fix is effective or that my feature works: Failed unit test, Passing unit test
• New and existing unit tests / Github Actions pass locally with my changes.

[PhRosenberger]

As this occured during unit testing and is now solved, it should be merged asap.

[thomassedlmayer]

Thanks for noticing/fixing the bug! The fix works with our tools.

Though, I thought we would maybe want to use the actual roll-pitch-yaw rotation matrix instead of using the yaw-pitch-roll matrix and then transpose? I made a proposal in my commit. What do you think?

I also added the transformation to vehicle coordinate system (center rear axle) and virtual sensor coordinate system as this was missing anyways. And I removed the rotatePoint function from the DummySource as this probably was a copy/paste mistake.

@ClemensLinnhoff Could you check if this works for you?

[ClemensLinnhoff]

Apart from the individual comments in the code, I would acutally prefer to use the inverse of the rotation matrix instead of reversing the rotation order and the angles, because it would then be alligned with the mathematical definitions in the documentation.

[ClemensLinnhoff]

As I said above, for me it is easier to understand the coordinate transformtion by having only one rotation matrix and using the inverse depending on the transformation direction. But I can understand, that people might also find it easier doing it by reversing the rotation direction and the angle directions. So it would be fine for me to use your approach @thomassedlmayer

I tested the current implementation with some unit tests and could not find an error, so the transformation (incl. translation and rotation) to vehicle and to the sensor seem to work.

[ClemensLinnhoff]

After removing the transpose function as noted above, the PR can be merged from my point of view.

[pmai]

CCB 2023-05-22: Merge as-is.