I was wondering how - suppose the pings last long enough for the approximate location of the recorders to be determined. How then will the recorders be actually found and retrieved? It seems that the ocean in the location of the detected pings is very, very deep. Close to three miles in places.
Well, the AF447 recorder(s) were retrieved from a similar depth, 13,000 feet, in a flat area next to very mountainous terrain. The fourth search mission finally succeeded, 2 years after the plane craished. They used three semi-autonomous "Remus 6000" robot submarines, mapping the ocean floor, and were prepared to search 10,000 sq km at the rate of 100 sq km/day. In the end the fourth mission located the wreckage after a week. Many passengers' bodies were still strapped in the seats, in the partly intact fuselage, preserved by the cold and depth. The robot subs located and retrieved the recorder, and then retrieved 104 bodies. 74 bodies remain missing, 50 were previosly retrieved from the surface.
http://boingboing.net/2011/05/06/air-france-447-how-s.html
I'm supposing something like this will be attempted with MH370. The sub operators will have a big advantage over their AF447 predecessors if the recorders' location can be narrowed down in the remaining battery life of the pingers. They will have a big disadvantage if the wreckage is in an unstable or steep location. I am guessing that, since no floating wreckage has been found, the airplane might be more intact than AF447. I would guess that would be helpful.
Here is something I don't understand. Apparently the Remus subs are semi-autonomous, meaning the sub is programmed for the dive, and then operates autonomously during the dive. The sub operator cannot control or receive information from the sub during the dive, except for some limited positional data. I can see how a semi-autonomous sub can search and map an area of ocean, following a programmed path and using its limited intelligence to react to obstructions. I do not see how a semi-autonomous sub can retrieve a recorder - recognize the object as a recorder, undo fasteners or cut material, extract the object from wreckage. I also do not see how the sub can retrieve a body - recognize the object as a body, cut or unfasten seatbelts, pull the body from the seat and manuever it through a damaged cabin. I didn't think robots are that intelligent yet. But obviously it can be done. Anyone able to explain how?