obj-Mesh orientations in rock-roboviz

[hwiedPro]

This is a link to this issue: rock-simulation/mars#129

Because it also applies here.

ToDo: Review the mesh import orientations for obj

Explanation
Obj mesh orientations are interpreted unintuitively, I guess this comes from the old default how they were exported from blender by default and thus were handled in MARS.

I had a look in the behaviour of pyBullet, Gazebo, rock-roboviz and MARS using the doc/a test model.

Test procedure for pyBullet and MARS:
I created a simple urdf with a single obj-mesh which represents a coordinate frame, and this was exported with z is up x is forward.

The results:

• Gazebo uses a right-hand coordinate system where +Z is up (vertical), +X is forward (into the screen), and +Y is to the left. (see GAZEBO Docs) (complies with stl)
• pyBullet uses a right-hand coordinate system where +Z is up (vertical), +X is forward (into the screen), and +Y is to the left. (tested) (complies with stl)
• MARS messes with the mesh rotation: Importing the mentioned test-urdf, the obj is placed with the x axis facing in -z, the z axis facing in x and y facing y
• Rock-roboviz does the same as mars

This issue is somehow related to #98

I suggest for MARS adding the orientation info how the obj-meshes should be parsed in SMURF.

[hwiedPro]

Will close this for further research on this topic and reopen later.