Above shows the performance of a Lockheed Martin civilian landing system developed in the mid-90s. It works by using the weather radar in SAR mode to update the INS, and the final 100 ft are flown by inertial guidance only.
There was interest from American and European carriers, and certification was underway, the rest is unknown. (Rumor has it, an export-embargo was placed due to the sensitivity of the SAR technology.)
Why can't the inertial guidance on its own fly an automatic landing the last 200 ft, thereby saving the CAT II/III installment?
Flying the final 200 ft in 20-30 seconds should have a maximum worst case scenario error of 3-4 m on a 45-60 m wide runway. It should even be better with today's GPS-coupled INS.
The final update before lock-on can happen either via an ILS CAT I or GNSS/SBAS. With a cross-check against stored runway track heading and glide-path angle. Thus eliminating the need for the weather radar and the runway specific SAR data.
- LM, as shown, has done it in 1996 for the last 100 ft. It was for civilian application.
- Initialization/alignment is not required, just a position update/check.