Archive for July, 2009

Baby Robots?

July 12, 2009

Can scientists build a “baby robot” that, like a child, learns as it goes and plays well with others? July’s Smithsonian magazine profiles Project One, a “wildly ambitious effort to crack the secrets of human intelligence. It involves, their grant proposal says, ‘an integrated system … whose sensors and actuators approximate the levels of complexity of human infants.'”

As the article points out, we have been able to develop robots that effectively carry out a wide variety of tasks, from building cars to playing the violin. And through a process called “supervised learning” (the “laborious analysis of spoon-fed data” such as a smile-detection system that “learns” by being shown tens of thousands of photographs of smiling and non-smiling faces) we can train robots to respond to humans. However, supervised learning isn’t how babies discover their world.

“If you want to build an intelligent system, you have to build a system that becomes intelligent,” says “Giulio Sandini, a bioengineer specializing in social robots at the Italian Institute of Technology in Genoa. “Intelligence is not only what you know but how you learn more from what you know. Intelligence is acquiring information, a dynamic process.”

This is the goal of Project One:

The robot baby will be able to touch, grab and shake objects, and the researchers hope that it will be able to “discover” as many as 100 different objects that infants might encounter … and figure out how to manipulate them … The subtleties are numerous; it will need to figure out that, say, a red rattle and a red bottle are different things and that a red rattle and a blue rattle are essentially the same.

Javier Movellan, a psychologist at UC San Diego and the director of the school’s Machine Perception Laboratory, was also involved with the development of RUBI, a robot designed to test interactive computing applications in the classroom. Interestingly enough, we first learned of RUBI at a talk given by Dr. Paula Tallal, one of the creators of the Fast ForWord programs, in which she emphasized the criticality of timing in language. RUBI, she said, also showed how important timing was in developing human interactions. The initial RUBI responded fractions of a second too slowly to the toddlers it encountered, and the children disengaged. However, when the researchers sped up the interaction time to approximate how quickly humans respond to sensory input, the children developed a relationship with RUBI.

In RUBI, the Smithsonian article says, Movellan developed a robot that humans can love. The goal of Project One is to develop a robot that can love humans.

The project is not without its skeptics, who doubt that a “baby robot” will tell us much about how humans learn. For example. Ron Chrisley, a cognitive scientist at the University of Sussex in England cites the fact that infants develop physically as well as mentally as they grow:

“To mimic infant development, robots are going to have to change their morphology in ways that the technology isn’t up to.”

And besides, says Chrisley, Project One is going at things all wrong by focusing on things like realistic human features. After all, “Human beings learned to fly when we mastered aerodynamics, not when we fashioned realistic-looking birds.”

Still, any new insight into how children learn to observe and reflect their environment will be welcome in the field of learning. And Project One’s robot won’t need diaper changes and could probably be programmed to make its bed and clean up its room without nagging…

%d bloggers like this: