If it's for offseason and outreach, you're in good shape!
I've been using the Leap with robots for a while now, time allowing me...eheh. There are a few different ways you can go about doing this, as long as using it during competitive play isn't a concern.
The most straightforward way is to have an off-board computer (basically, a secondary driverstation -- or your actual driverstation) that forwards tracking data from the Leap to your cRio for interpretation. This would involve some work with sockets and networking, but the FRC API does a good job of mitigating the troubles involved with that. In essence, this method would involve programming your cRio to poll the off-board computer for tracking data from the Leap so that it can move the robot accordingly.
As far as the off-board computer client goes (the part that gets data from the Leap), you can hit up one of my early Leap projects at this page: http://www.mechakana.com/blog/leap-motion-developer-kit-arrives-robotics-gets-100-cooler/; this project does almost exactly what you were asking for in your first post, mapping hand tracking data to two virtual joysticks using some Java -- perhaps this helps a little bit? You would need to add some code for sockets, but I think I have some of that lying around as well...maybe it's time for me to write a new tutorial. ;D
On a semi related note, I'm assuming that your robot is non-holonomic given that you've only got two motors installed; this means for movement, you'll only need to use two of the four possible "joystick" axes to move, right? ;D In practice, if you can, I would suggest mapping your control system to the pitch and roll of the hand, since physical X/Y/Z axis coordinates of the palm can be extremely finicky and non-intuitive for this kind of application.