Android Copy of Young Woman Unveiled In Japan 264
An anonymous reader writes "According to IEEE Spectrum, Japanese roboticist Hiroshi Ishiguro, who had previously built a robot copy of himself, has now created a new android — and it's a 'she.' Geminoid F, a copy of a woman in her 20s with long dark hair, exhibits facial expressions more naturally than Ishiguro's previous android. 'Whereas the Geminoid HI-1 has some 50 actuators, the new Geminoid F has just 12. What's more, the HI-1 robot requires a large external box filled with compressors and valves. With Geminoid F, the researchers embedded air servo valves and an air servo control system into its body, so the android requires only a small external compressor.' It's also much better looking. Has the Japanese android master finally overcome the uncanny valley?"
Re:Dammit Japan. (Score:3, Interesting)
Re:George Lucas did it first (Score:3, Interesting)
Cherry 2010 and if it follows Moore's law they will be posting on slashdot by Cherry 2020.
I would guess that this would be a life extension method at its completion, which allows space for the brain and spinal cord along with an interface. I think that is another of those uncanny valleys.
Though it does seem more like a science fiction movie such as reanimator [wikipedia.org].
He needs to work on the skin texture (Score:5, Interesting)
If we can't get CGI characters to act human... (Score:5, Interesting)
CGI animators, in some sense, have a much easier task then the roboticist. Its much easier to program a full musculature into an animated character than to physically build a robotic one.
The difficulty of all this is exemplified by Robert Zemeckis' dismal "Polar Express" and "A Christmas Carol". Even when capturing hundreds of control points on the faces of the actors, you're still left teetering on the edge of the Uncanny Valley.
"The Curious Case of Benjamin Button" [ted.com] and "Avatar" were more successful because they did complete surface capture of the actors faces rather than point-capture.
Which gets back to the difficulty of making robots appear human. Its the same problem, magnified 1000 times by the fact that, in essence, you have to pack the equivalent of a millions of "control-points" into the robots face.
Not an easy task.
-Sean
Pygmalion (Score:4, Interesting)
I certainly hope this story doesn't turn into a modern day retelling of Pygmalion [wikipedia.org] .
Re:If we can't get CGI characters to act human... (Score:3, Interesting)
Which gets back to the difficulty of making robots appear human. Its the same problem, magnified 1000 times by the fact that, in essence, you have to pack the equivalent of a millions of "control-points" into the robots face.
You're thinking about this like a software guy, not a hardware guy. The human face doesn't have that many control points. It has a relatively small set of muscles and a bunch of connective tissue with known (albeit nonlinear) properties.
The problem is that much of what humans do with facial expression is "non-functional" in the sense that it doesn't involve any practical intent or communication. We blink, we chew our lips, we move our muscles simply because stillness is fatiguing. I'm betting that with a relatively small amount of attention to such "non-functional" movements we'll see adequately canny robot faces in fairly short order, particularly given the huge range of perfectly acceptable facial behaviour that humans actually have.
I get the sense that no one complaining here has ever dealt with a person whose face is partially paralyzed, or burned, or what-have-you. Anyone who has knows that you get past those aspects very quickly, and simply start seeing the person. But I'm sure the same people complaining here will still be complaining about the uncanny aspect of robot faces long after a wealth of empirical data has shown that the average person can't tell the difference without already knowing the object they are dealing with is a robot.