DARPA issues BAA for advanced robotic translator
C-3PO. Image credit: Wikipedia.
(PhysOrg.com) — the Defense Advanced Research Projects Agency (DARPA) has issued a Broad Agency Announcement (BAA) for the procurement of research leading to an advanced robotic device capable of performing as a translator between humans who speak different languages, while also responding to a set number of visual cues, and is able to "listen" to communications, whether they be auditory or written and to respond with some degree of artificial intelligence. In short, DARPA wants a robot that would be some sort of combination between C-3P0, the robot from Lost in Space, and maybe the computer from Star Trek, (though it's not quite asking for the sentient android DATA, just yet.)
It's no secret that the DoD wants better translation capabilities than they currently have; soldiers carrying small devices that can do little more than spit out canned phrases hasn't gone very far in helping to understand what is going on during battlefield conditions. and as they have in the past, the top brass turn to DARPA, the folks who helped create the Internet structure that we all know and love today, to lead the way.
Called the Broad Operational Language Translation, (BOLT) the BAA specifies that DARPA is looking for a new Human Language Technology (HLT) research and development program to create a hardware/software combination that can translate multiple (foreign) languages "in all genres" (voice, recorded voice, email, text, etc.) It should also be able to retrieve information from translated material and be able to serve as both a machine-machine and human-machine translator and finally, the trickiest part, to take both visual and tactile cues from the environment and then to perform some sort of reasoning on that information, possibly combined with other inputs, to come up with an appropriate response, either orally, or via text.
As with most requests by DARPA, this request for proposals doesn't necessarily require those who win one or more parts of the contract to actually construct the end result. what DARPA is looking for, is research. They want breakthrough technology, not just an addition to what is already out there.
The problem of course in paying for research instead of a promised product, is that if a contractor doesn't come up with anything new, or at least something that looks good on benchmark tests, the whole program can go away just as quickly as it came; and at a proposed budget of $15 million the first year, that's a lot of money to throw at something that could very easily wind up as nothing more than a simple wish list.
More information: via Wired