A Robot That's Learning to Smile
The UCSD robot watches itself to learn how to pull new facial expressions.
Kristina Grifantini 07/10/2009
- 3 Comments
 |
Researchers at the University of California, San Diego (UCSD), who demoed a realistic-looking robot Einstein at the TED Conference last February, have now gone a step farther, infusing the robot with the ability to improve its own expressions through learning.
Previously, the head of the robot--designed by Hanson Robotics--could only respond to the people around it using a variety of preprogrammed expressions. With 31 motors and a realistic skinlike material called Frubber, the head delighted and surprised TED conference goers last winter.
Inspired by how babies babble to learn words and expressions, the UCSD researchers have now given the Einstein-bot its own learning ability. Instead of being preprogrammed to make certain facial expressions, the UCSD robot experiments in front of a mirror, gradually learning how its motors control its facial expressions. In this way, it learns to re-create particular expressions. The group presented its paper last month at the 2009 IEEE Conference on Development and Learning.
According to a press release from the university,
Once the robot learned the relationship between facial expressions and the muscle movements required to make them, the robot learned to make facial expressions it had never encountered.
Such an expressive robot could be useful as an assistant or teacher, or just as a means of learning more about how humans develop expressions. But a robot that watches itself in a mirror, practicing and improving how it looks, seems like another step into uncanny valley.
On Twitter
Become a Fan on Facebook
Subscribe to the Feed
ms
190 Comments
evaluating expressions
One thing that puzzled me about this story is that no mention was made about how this robot evaluated what it saw in the mirror. Was it preprogrammed with some expression evaluation algorithm? In which case, making expressions it had never encountered (but had been preprogrammed to recognize) wouldn't be particularly surprising.
Ideally, it would learn, without a mirror, from reactions of others to its experimentation, as a human baby might.
Reply
Bruceahz
38 Comments
Re: evaluating expressions
I would guess that the intent is for the robot to see someone else make a face and then know which motors to drive to duplicate that face.
In other words - the feedback from the mirror was only used to learn what each motor did (and also the cross terms in the control matrix that tell the computer the effect of more than one motor being driven). Clearly the robot has eyes to look at a human's face; now it knows how to imitate what it sees.
Reply
naitdawg21
1 Comment
Re: evaluating expressions
I find myself agreeing. While it is interesting to note that it is possible for a machine to learn which motors do what for facial expression, that does not mean that the robot is "learning" new expressions. It isn't really learning to express itself, it's learning to manipulate it's facial structure. Learning to express yourself would need to include realizing which particular muscle/motors to use in precisely which situation based on how the robot is "feeling". In order to do such a thing the machine would have to first know what anger is, or sadness, or joy. And I just do not think that's the kind of thing that can be programmed into a machine to recognize. Certainly you could teach it to "express" what it computes is the appropriate response in a given situation, but in this instance above that is not what is happening. Its not teaching itself to express, its teaching itself to move its face. I find this to be an overzealous definition of "expression" which is not quite correct.
Reply