Hi @drtritm,
I say the set up you are describing will work since one of my projects was exactly that. However, since you are only using the raspberry pi as a the Kinect "driver" I would recommend to distribute the different tasks you want perform between machines. Tasks like mapping will perform better and faster with more computing power, besides take into account possible communication delays between your PC and the Raspberry Pi.
If you want some info on the know-how you can check the kinect drivers here or here, the ros navigation stack for path planning, gmapping for static map generation (You will able to find more tools like that apart from gmapping) and localization tools like robot localization and, since you will be using a static map, AMCL. In your set up it will make more sense to use AMCL with your generated map instead of robot localization.