PlayThis episode covers android science and human-robot interactions with expert Hiroshi Ishiguro from Osaka. After the interview we feature the last installment of Jack Graham’s Selkies story, as well as a poll on the future of Androids.
Hiroshi Ishiguro
Hiroshi Ishiguro is Professor of the Department of Adaptive Machine Systems, and the group leader of the Intelligent Robotics Laboratory at Osaka University.
If you’ve ever seen an Android robot in the news, it probably came out of the Intelligent Robotics Laboratory. Starting from the robotic replicate of his then 5 year old daughter, Prof. Ishiguro then went on to model a female android after Ayako Fujii, the NHK news announcer. Finally, his latest robot Geminoid, which is a close copy of himself, is able to replace its creator in lectures and interviews.
His robots are highly actuated to give them human like facial expressions and reflexes. However, because the AI needed to interact in a human-like manner is not always advanced enough, Prof. Ishiguro has been looking to partially teleoperate his robots. Another approach investigated is to make his robots autonomous by having them perceive and react to their world thanks to networks of cameras and microphones.
Using these robots as a tool, Prof. Ishiguro has been exploring the field of Android Science, which looks at both the appearance and behavior of humanoid robots and their impact on human robot interactions. In particular, he is looking to verify the existence of the uncanny valley and to explore how to make androids which sufficiently resemble humans to be likable. On the more philosophical side, his androids open the door to understanding what human presence really means.
Survey
What is your take on androids? Will androids ever become indistinguishable from humans, both in looks and behaviour? Will they become unrecognizable like the Cylons from Battlestar Galactica, or remain awkward like Data from Star Trek? Take the poll on the Robots forum!
Selkies
In this last installment of the Jack Graham’s Selkies story Mangan sets his Selkies free with a new schooling algorithm in a sea full of sharks. Will this new generation of robotic swimmers escape the jaws of the sharks? Tune in to find out…
Links:
- Download mp3 (22MB)
- Subscribe to Robots using iTunes
- Subscribe to Robots using RSS
- Hiroshi Ishiguro’s website
- Jack Graham’s website
Latest News:
As always, you can find more info on the robots mentioned in this episode on the robot forum, including what may be your future robot
housemaid, Australia’s anti IED robot SPIKER as well as videos and a slideshow of 25 years of CMU’s Field Robotics Center.
View and post comments on this episode in the forum
It may be entertaining to speculate on this, but I believe the question leads down a dead end road. It’s a bit like asking "What would you do if you had a computer as powerful as your brain?".
I had a discussion with a very seasoned roboticist yesterday and he told me how he was very frustrated with people’s take on human-robot interaction. "All people ever think about are robot servants that you can give orders to." He thought that this attitude, thinking of robots as servants taking orders, is a major obstacle in creating meaningful interactions.
We’ve now had PCs for 30 years. And the best mainstream achievement I can think of right now is the miserable MS Office assistant. I think in many ways our communication with machines is still at the level of a TV remote control – 1 button for each action. And in my view the interaction problem is even harder in robotics.
Will androids ever become indistinguishable from humans, both in looks and behaviour? Yes. Will they then take orders from us? No. Will we get there by trying to make an exact copy of a human? No.
Beyond the crazy-scientist-builds-Frankenstein appeal, I think Ishiguro’s work is cargo cult science.
Hey Adam,
I guess you can see, I’ve been thinking about this issue of late. I do believe there will come a time when androids will be physically indistinguishable from humans. The only issues are time, research, and money. There’s a lot that needs to be done in terms of basic research. I don’t think anyone will be confusing an Asimo for ‘Six’ anytime soon. And even once the technical hurdles are overcome, I can see it would probably be cheaper and simpler to build caricatures or simpler stylized replicas of humans for routine tasks. I mean look at art. We have the ability to make very photorealistic animations, but I sure like the look of anime characters, and when I need a laugh, it doesn’t bother me that Peter Griffin on the cartoon "Family Guy" doesn’t look exactly like an actual human. The facsimile is close enough to do the job, and as robotics and android research progresses, there will be these kind of functional plateaus.
Now..
There will likely be a strong financial motivation to make them more realistic for a variety of, er uh, ‘entertainment’ purposes. No doubt that is gonna advance the science- just like how it advanced the VCR. But I can’t see any technical reason why we could not, in the near future, cross the uncanny valley.
The best example I can think of is computer graphics, now true, the computer generated movies Beowulf and Final Fantasy may not seem to be the best demonstrations of replicating humans as progress. But think back to character animation on ‘The Last Starfighter’ and ‘Amazing Stories’ just 25 years of effort has gotten us to a liquid Golden Angelina Jolie. Give it a couple years..
Mechanics are harder to implement that computer graphic simulations, but the motions (or emotions??) are being modeled now. If we can create actuators that behave more like muscle that will help a lot- Ishiguro and Festo tried pneumatics, and Hanson is doing interesting work with textured skin. It’s gonna get interesting..
And once again our choice of show topics seems synergistic-ly ‘uncanny’!
-MoI Carl
but, we’re not there yet,
http://www.trendhunter.com/photos/18485/2
=shudder=