I'm working on an importer for Collada v1.5. I've got the impression that there is redundancy in the definition of the kinematics_model, namely in the visual_scene node tree and the kinematic_model. When I'm defining a kinematic do I specify all transformations between the axis and geometries in the node hierarchy of the visual scene. Do I have to repeat the same transformations in the kinematics_model (in between the attachment_full)?
That seems like unnecessary redundancy to me, or am I getting it wrong?