have an single source of truth abstraction around kinematic trees and individual frames in space (be it cartesion or joints) here, which is used by all downstream tasks
    -> have documented clear conventions about coordinate systems
