Alice, I think it is quite possible that this could be part of the picture. I know nothing of Tu154 avionics or autopilot systems but in a western standard Flight Management System (FMS) there are likely to be LNAV final approach tracks to individual runways. These can be part of the data base or they can be loaded into the system manually. The FMS position may be constantly updated by GPS and hence may be considered to be extremely accurate. Of course the coordinates of individual waypoints must be correct otherwise errors will occur.
The autopilot can be coupled to the LNAV track and in theory will take the aircraft to the runway position. However this facility of itself does not constitute an approved approach method, certainly not for low visibility.
At Smolensk my understanding is that the only promulgated approach procedures are NDB and PAR. Pilots will want to use the lowest available approved minima for their approach in poor visibility and PAR will give them that. It is quite possible the flight was actually navigating the final approach course by LNAV and because it was very accurate there was no need for interventions by ATC with heading changes, hence the apparent lack of input from the PAR controller.
Then at the end of the approved PAR procedure, they were theoretically visual but in practice in 200m visibility they could see nothing useful ahead but by leaving the autopilot coupled to the LNAV they believed it would take them to the runway. And, I hasten to add.this is more speculation, because I don't know if the Tu154 was so equipped.