* Officials plan to deploy underwater robot
* Search now in sixth week, well beyond pinger battery life
* Chances of finding floating debris now remote (Recasts throughout)
By Matt Siegel and Byron Kaye
SYDNEY/PERTH, Australia April 14 (Reuters) - A U.S. Navy underwater drone will be deployed to scour the floor of the Indian Ocean for a missing Malaysia Airlines plane, search officials said on Monday, launching a new phase of the operation after nearly six weeks of fruitless searching.
The hunt for flight MH370 will head deep underwater as the batteries in the flight's black box recorders had probably died and there was little chance of finding floating debris, said Australian search chief Angus Houston.
The search is now relying on the U.S. Navy's sophisticated Blue-fin 21 autonomous underwater vehicle, which is set to search the ocean floor for wreckage some 4.5 kms (2.8 miles) beneath the surface.
The aircraft disappeared soon after taking off on March 8 from Kuala Lumpur bound for Beijing with 227 passengers and 12 crew on board, triggering a multinational search that is now focused on the Indian Ocean.
Searchers are confident they know the approximate position of wreckage of the Boeing 777, some 1,550 km (963 miles) northwest of Perth, and are moving ahead on the basis of four acoustic signals they believe are from its black box recorders.
"Despite the lack of further detections, the four signals previously acquired taken together constitute the most promising lead we have in the search for MH370," Houston told reporters in Perth.
"The experts have therefore determined that the Australian Ocean Shield will cease searching with a towed pinger locator later today and deploy the autonomous underwater vehicle, 'Bluefin-21', as soon as possible," he said, referring to the U.S. Navy device designed to detect the tell-tale "pings".
The batteries in the black boxes are now two weeks past their 30-day expected life and searchers will be relying on sonar and cameras on the Bluefin-21 drone.
An aircraft's black box records data from the cockpit and conversations among flight crew and may provide answers about what happened to the missing plane.
The Blue-fin robot will build up a detailed acoustic image of the area using sophisticated 'sidescan' sonar, hoping to repeat its success in finding a F-15 fighter jet which crashed off Japan last year.
If it detects possible wreckage, it will be sent back to photograph it in underwater conditions with extremely low light.
Building up the necessary mosaic of thousands of high-definition photos in the undersea gloom can be a long and frustrating task, a point Houston reiterated on Monday, citing the extremely large, remote and deep search area.
Officials are currently focusing their acoustic search on an area about the size of a medium city - 600 sq km (230 sq miles) - and say it could take the underwater robot months to scan and map the whole search zone.
"I would just say to everybody, don't be over optimistic, be realistic and let's hope, let's hope that that very strong signal that we were receiving is actually coming from the black box," Houston said.
The mystery of Malaysian Airlines MH370 has sparked what is on track to be the most difficult and expensive search and recovery operation in aviation history. Up to a dozen planes and 15 ships will be searching in three separate areas on Monday.
Malaysian authorities have still not ruled out mechanical problems as causing the plane's disappearance, but say evidence suggests it was deliberately diverted from its scheduled route from Kuala Lumpur to Beijing.
Houston added that although an oil slick was located in the search area on Sunday evening, he was pessimistic about the likelihood of finding anything floating on the ocean surface after this amount of time.
"The chances of any floating material being recovered have greatly diminished and it will be appropriate to confer with Australia's partners to decide the way ahead later this week," he said.
(Reporting by Matt Siegel; Editing by Michael Perry)
Our Standards: The Thomson Reuters Trust Principles.