The new LEGO PF Servo Motor will be out as a single-item set in March this year.
That means this simple solution will be available without buying too many 9398 Crawler sets:
Just a 9V supply and a 1-way switched supply is enough to drive the LEGO PF Servo Motor as it returns to the centre position when C1/C2 power is removed.
I drive the servo with a standard PF battery box and pole reverser switch (one way and centre only, the other way disabled with plates blocking the lever).
I tested it successfully with several lead added so I know it will work 2 metres from the control panel.
The PF pole reverser switches make a control panel quite easily and the extra little switch allows the user to set the polarity for pairs of facing points (which is why I petitioned for it to be added).
In the picture the mast foot clamps the point slider casing to the motor mechanism, keeping the end of the rack in the lever shoe under high torque.
The lever raises the red piece when the point is set curved.
The mechanism is designed to minimise the profile close to the track, to accommodate large train overhangs that would knock a yellow lever out of place.
This solution is designed for best reliability in places you can't see or reach over a 2-day show, such as a fiddle yard under the station (hence the stanchion).
The NXT would not provide more than C1/C2 with the standard NXT-to-PF converter lead but the PF lead could be hacked to add a 9V supply for several servo motors if necessary, or a bridge rectifier could be added to each lead if single motor sub-system autonomy was required.
I dare say an easier PF Servo Drive solution for the NXT may present itself in the future. I can see NXTs driving PF Servos being popular for robotics.
For the moment the easiest no-mod NXT solution is to use a PF IR Receiver and for the NXT to drive it (or several of them) with an IR Link sensor (which are on January sale at LS@H).
More mechanism pics in this folder
This is simpler than my previous solution, which used either PF, monkey or 71427/43362 gearmotors but more gears:
This previous solution fell foul of a sticking switch point. It stuck just a bit more than the torque of the white clutch gear. The clutch gear would eventually wear out, which is why I went for the servo motor solution instead - more torque, no wear-out mechanism and open-loop deterministic position - it ticks all the boxes!