ASIMO learns to understand three people at once

ASIMO has already proven itself to be pretty multi-talented, but it looks like it's now added yet another trick to its repertoire, with a pair of researchers giving it the ability to understand three people speaking at once. That was done with the aid of eight microphones and a specially-designed software program dubbed HARK, which works out where each voice is coming from and isolates it from other sounds. Right now, however, that's only being used to referee games of rock-paper-scissors, with each individual shouting out their choice at once, but the researchers eventually hope to get ASIMO up to the level of humans' ability to listen at a cocktail party, although they admit that is still a "long way" off.


















so now it can be a foursome?
Orgy*
Damn... I didn't get your joke... and now I wish I could delete or edit my comment.
ENGADGET! WHY DO YOU TAUNT ME SO!
This would be awesome for spying on stuff. They should make a military version in mass quantities, but with friggin lasers and leet doomsday devices. Oh yeah, then ship them over to Iraq.
Balls to spying. Arbitrating Paper-Scissors-Rock contests at cocktail parties is where the fun's at.
What? So it can rock paper scissors the insurgents and leave the innocent without shelter in the midst of a civil war?
Ah crap, this is where skynet starts.
I hear that
Sentient computer system beats rock, paper and scissors all at the same time.
So, if 4 people start to talk, does it yell "HARK!" and fall over?
:)
Will it smash its face again like that stairs stunt?
maybe when a fourth person starts talking it will get nervous and pass out.
What if I just ramble on and on and on and on until it starts falling backward and foaming at the mouth like this...
Looks like Asimo has been committed to a sanatorium.
add a core 2 quad and gigs of ram, will ASIMO understand 5000 people talking at once ?
"You had 'AIDS' so you lost."
the game
We ALL wish you could delete or edit your comments as well.
Great, now ASIMO can pwn at Brain Age 2 as well.
Video PLEASE!!!!!!
Problem is it would be more like
"You got scissors so you won, but you got rock and rock beats scissors so you won but you got paper and rock covers paper and so you won, but you got scissors and scissors cut paper so you wo..."
Starts to Smoke and Flames appear from the bots eyes while head is spinning and arms move in stereotypical fashion...
"overload... overload... why why why was i programmed to feel pain?"
kaboom..
I'm pretty sure it'd just declare a tie. Sorry to burst your not-funny-and-totally-overused-scenario bubble.
HAHAHA nicely done sir; that was a great episode of the Simpsons
Maybe it has "paradox absorbing crumple zones"... Although, those usually only work for paradoxes that are spoken do it, not generated by it...
(Bad Futurama ref.)
Oh well. :)
Can it outrun an ant yet?
I can never hear crap at cocktail parties, so it's already doing as good as me.
Just a matter of time before it starts ordering up it's own parts and building some new ASIMO buddies and so on until we're arbitrating their RPS matches
Now, if we can only teach ASIMO to put the comma in the correct place.
All obey the robot overlords.... but if you get scared of them just run up a flight of stairs
I would say that ASIMO in this case is peripheral to the microphones and software package. The robot is nothing more than the interface to signify the winner and to kick off a game.
Stupid, because it’s try to recognize each person from the voice source in 3D world, that mean it useless when people start to move around, which what happen in real live, it should recognize each person voice from its voice variables like vocal difference, pitch & tone, volume,….etc like we human do, sure the down side of this as when 2 people sound the same, but this is also confusing for us human, but if you add sound source and match that with visual image of the mouth moving, you can get more accuracy.
So when are *you* going to get off your a-double and create that software?
Quit complaining that work that someone else has done isn't good enough, and create some value of your own!
Gasp! This means that we're one step closer to making Data!
If anybody remembers Nexi http://www.engadget.com/2008/04/03/mits-nexi-bot-wants-to-be-your-friend We are already close.
@ taurus owner
No no no no. Clearly, you didn't get the reference. In one episode of TNG, Data listens to a bunch of different classical pieces, all at once. And then a human teaches him that there's more to beauty than rapid processing, or some such meaningful, human thing.
You know I think I saw that episode.
You would have, if you'd seen all of them, as you should have.