URDF compositional complexity
I'm in the process of creating a URDF description for my quadruped robot. I have a bunch of bioloids frames and motor models but if I define the leg chains using them I end up with a bunch of fixed joints that do nothing except allow me to define geometries and visuals. Is that to be expected? Should I try to 'bake' the models into single shapes? How do the fixed joints affect KDL chains generated from the description? Does the performance suffer?
Are there any best practices in this regards?