Cat III autoland is not zero/zero, never will be since there is still a requirement to taxi off the active and get out of the way of the next aircraft without getting lost on the way. Typically the minimums for Cat IIIb is zero DH and 75m RVR. The way a fixed wing gets down is to fly the ILS G/S and LOC to a certain height depending on the specific type, but somewhere around 100' the autopilot switches from G/S - LOC to Rad Alt - LOC to execute the final stages of the flare and landing. The LOC is very good at lateral guidance at the touchdown point since there is a good distance from the LOC antenna to the aircraft and the LOC signal continues to guide the aircraft down the centerline of the runway via rudder and eventually nosewheel steering. (we have to remember to disconnect the autopilot at the end of roll out otherwise the thing tries to go back to the centerline and a small fight develops). Flare is done by Rad Alt since this is entirely a function of rate of descent and forward speed. The G/S antenna goes past at 120-140 knots and is pretty useless in the last stages of the landing.
Helicopters are an entirely different thing, I seem to recollect the problems that Bristows had with the LN450 on the Bell 212 with the approach to hover. I understand that the manoevre was done manual collective and cyclic inputs following a flight director, while the TR pedals were maintained by the AP. I gather there were particular problems with directional stability when there were light winds and may be a clue why fundamentally helicopters will be such a hard proposition to autoland.
Fixed wing are fundamentally directionally stable especially at typical approach speeds, even an engine failure at 10' agl will not unduly worry a good digital autoland system, however the crosswind limits for an autoland are typically 50-60% of a manual landing limit. Helicopters are not directionally stable at low speeds and the interaction of downwash over the airframe and TR combined with changes in the wind velocity and direction are too difficult to model with any great accuracy.
I would however see no problem in reducing the approach minimums from Cat I to Cat II for helicopters with relative ease, though justifying the cost would be hard when measured by the number of times it was required.
Interested in Nick's thoughts as he would be closer to any attempts in that direction.