Not criticizing, just wondering.
I'm following the flag rush tut to try to make a vehicle with multiple turrets, each with multiple guns. I can see where using Node to process movement of your model is useful, but I am getting confused about the dual role of controllers and nodes.
Aren't controllers meant to be used to control the movement of an object? Would it make sense to add the .model to the scenegraph and let the Vehicle be scenegraph independent? Would this work better in a multiplayer client/server setup? Then you could have controllers on the client side that were just moved by the server rather than doing much thinking. The'd just be responsible for taking server commands and doing a little prediciting if they didn't get server data on that update.
Well, I should say that the flag rush tutorials show one way of doing things, but that doesn't mean it's the only way.
Making it a full blown controller is a bit of overkill for what is going on in Flag Rush, but as you said that's a legitimate way of doing it.
Don't take what is done in those tutorials as gospel, it's just giving a good example and explaination of how to do things. In fact, I sometimes do things in an inefficient way so I can cover something in the next optimization section.