[dbeckwith]

Not to discourage you or anything, but has anyone actually tried this and gotten it to work? My team tried using it at first, but we realized that automatic PID control could never be as good as human prediction. I think the problem with auto-balancing is that PID is completely reactive. In order to be able to balance successfully, a human driver needs to be able to pretty accurately predict how the bridge will react to the robot moving and adjust accordingly. If you can get this to work, great, but I feel the problem may be more complicated than just using PID and human control will always be better.