Communications of the ACM
Popeye, the Robot with Brains Not Brawn
European researchers have developed a new approach to artificial intelligence that could empower computers to respond intelligently to human behaviour as well as commands.
The dramatic rise in raw computing power due to parallel computing, super-clusters and the like, has meant that computers are now extremely adept at performing complex computational tasks that once bogged down mainframes and server farms for days.
Pattern recognition, in particular, has reaped the benefits; faces from surveillance cameras can be identified and compared in real time; robotic dogs can recognise and track a "football" in the Robotic World Cup; and our voices can be interpreted with an unprecedented degree of accuracy, enabling robots and computers to respond to our every command.
The pattern-recognition problems underlying these achievements have been fundamentally solved and put to good use. But before we can engineer an army of house-bots and robo-slaves, they must first be able to intelligently respond to the subtle nuances of human communication: the meaning inferred from our behaviour and conversations. It's no trivial feat – many humans struggle with the challenge on a day-to-day basis.
But by combining multiple sensory inputs and enabling them to work together, the Perception-on-Purpose (POP) project has taken an important step towards that goal.
"The originality of our project was our attempt to integrate two different sensory modalities, namely sound and vision," project coordinator Radu Horaud explains to Science Daily. Their robot, named Popeye, was built to work out which voices are "relevant" amongst a cacophony of noise by combining video input and image recognition technology with sound analysis.
"It is not that easy to decide what is foreground and what is background using sound alone, but by combining the two modalities – sound and vision – it becomes much easier," Horaud continues. "If you are able to locate ten sound sources in ten different directions, but if in one of these directions you see a face, then you can much more easily concentrate on that sound and throw out the other ones."
Hardly the most aesthetic robot to grace our screens, Popeye is not much more than a mounted bust. But what it lacks in looks it makes up for with smarts: using two microphones and cameras mounted on its head, Popeye is able to identify a speaker with a fair degree of reliability, which the team is working to improve even further.
From Wired
View Full Article
No entries found