update for using multiple visual for collada_urdf

[YoheiKakiuchi]

I added code for using multiple visual for converting urdf to collada.
If urdf link has just one visual, there is no difference with former result.
urdf link has more than two visual, first visual would be added with former manner.
Next visuals would be added as transformed geometry.

[isucan]

Could we remove this?

[YoheiKakiuchi]

I have left some error messages. They were not needed. Please remove lines start with std::cerr .

[isucan]

This is great, thank you. Do you just take the "default" group and ignore the rest?

Can we do the same for the collision tags?

[YoheiKakiuchi]

Do you just take the "default" group and ignore the rest?

Yes. I don't know exact usage visual(collision) group. It seems to have detailed visual(collision) and simple one, and will change depend on situation. I have no idea to do that in collada format.

Can we do the same for the collision tags?

Yes. Now exported collada file has just visual description (same as collision), so we should choose visual tag or collision tag for exporting.

[isucan]

Is it not possible to export both collision and visual in collada? (I am not too familiar with the collada format either). How does one choose visual / collision tags now?

[YoheiKakiuchi]

Now in urdf_to_collada, if visual tag exists, visual tag is used as geometry(collision/visual) in collada, and ignore collision tag.
If visual tag does not exist, collision tag is used. If both visual and collision doesn't exist, there is nothing to do (link has no visual/collision).

Collision geometries extension is described at http://openrave.org/docs/latest_stable/collada_robot_extensions/
Now, urdf_to_collada doesn't support this extension. I think it is future work.