[diracs_stache]

Sounds like a good project to me (we did this for a couple of labs in an undergraduate [sophomore maybe junior year] control systems class). The "inverted pendulum" problem and its subsequent derivations should be a good model for this. When the object is nearly upright (or at small angles from the vertical axis) a nice linear control loop should suffice but if you leave that region the control solution becomes more difficult (would make a good target for improvement after initially getting it going).