Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
Technology

Meet Lucy, The Orangutan Robot 336

Posted by michael from the watch-out-when-it-picks-up-a-thighbone dept.
Roland Piquepaille writes "Lucy is not an ordinary robot, driven by software. She's a pure product of artificial intelligence (AI). And after a three-year long training, she's now able to make a difference between an apple and a banana, which is quite handy for an orang-utan, even if she doesn't eat them. Her five microcontroller chips wouldn't like this... In "A Grand plan for brainy robots," BBC News Online tells us that Lucy is the brainchild of Steve Grand, an honorary research fellow at Cardiff University's School of Psychology. And why did he choose an orang-utan design? "I made Lucy as an orang-utan because, can you imagine how scary it would be if she looked like a human baby?," said Grand. More details and references are available in this overview which also includes the cover of Grand's last book, 'Growing Up with Lucy: How to Build an Android in Twenty Easy Steps,' which was already reviewed on Slashdot."
This discussion has been archived. No new comments can be posted.

Meet Lucy, The Orangutan Robot More

Meet Lucy, The Orangutan Robot

Comments Filter:

  • Best quote in article... (Score:4, Funny)

    by jhouserizer ( 616566 ) * on Friday March 19, 2004 @04:15PM (#8614621) Homepage

    Best quote in the article: "I like 'intelligent' people. It's the thick ones that worry me."

  • Heh (Score:3, Funny)

    by Ash87 ( 739997 ) on Friday March 19, 2004 @04:15PM (#8614622)
    "Meet Lucy, The Orangutan Robot."

    Contender for best story title? =D

    • Re:Heh (Score:1, Funny)

      by Anonymous Coward
      "Meet Lucy, The Orangutan Robot."

      Contender for best story title? =D

      Contender for worst failed first post attempt?

      • Re:Heh (Score:3, Funny)

        by Penguinisto ( 415985 )
        "Contender for worst failed first post attempt?"

        No, that would be "To hell with the monkey, I want my Linux Fembot with a penchant for evil!"

        • No, that would be "To hell with the monkey, I want my Linux Fembot with a penchant for evil!"

          This is going to be the best. prom. ever.

  • The final test... (Score:4, Funny)

    by ackthpt ( 218170 ) * on Friday March 19, 2004 @04:16PM (#8614629) Homepage Journal
    she's now able to make a difference between an apple and a banana,

    The final test will be if she can pull the football away just before Charlie Brown tries to kick it.

    that or rip his legs off...

    • Re:The final test... (Score:4, Funny)

      by ozbird ( 127571 ) on Friday March 19, 2004 @07:27PM (#8616894)
      The real final test is for the robot to know what the fruit is, but is able to lie about it:

      Lister: OK try again what is it?
      Kryten: It's a banana
      Lister: No it isn't. What is it?
      Kryten: It's a banana
      Lister: No it isn't! What is it?
      Kryten: It's an orrrrrr its an orrrrrrr
      Lister: It's an orange say it. IT IS AN ORANGE.
      Kryten: It's an orrrrrr it's an orrrrrr It's no use sir I can't do it
      Lister: You can. I'm going to teach you. (Puts down banana picks up apple) Ok what is it?
      Kryten: It's an apple
      Lister: No No No. What is it?
      Kryten: Oo it's no use sir I just can't lie I'm programmed to always tell the truth.
      Lister: Kryten it's easy look. (holding an apple) It's an orange (picks up orange) it's a melon (holding a banana) it's a female aardvark.
      Kryten: Oo that is just so superb sir. How d'ya do that, especially calling a banana an aardvark. An aardvark isn't even a fruit. It's total genius.

  • King Louie's head on a toaster oven. Creepy. (Score:5, Funny)

    by The I Shing ( 700142 ) * on Friday March 19, 2004 @04:16PM (#8614638) Journal
    It doesn't need a baby's head... it's creepy enough as a primate.

    I haven't been this creeped out since the first time I saw that Quiznos Subs commercial.

    And what's with that glowing blue Terminator eye? Imagine that thing chasing Linda Hamilton around.
    Your clothes... give them to me... I'll take that banana, too...
    Can't he cover that thing with fur or something? Make it look like a toy instead of like something out of madman's nightmare?

    • If his goal was... (Score:5, Funny)

      by Atario ( 673917 ) on Friday March 19, 2004 @04:19PM (#8614697) Homepage
      ...to avoid being scary, he's failed miserably.

      Can't sleep...orangutan robot'll kill me...can't sleep...orangutan robot'll kill me...can't sleep...orangutan robot'll kill me...

      • The Japanese do it right (Score:5, Informative)

        by macshune ( 628296 ) on Friday March 19, 2004 @04:40PM (#8615063) Journal
        I don't know if this is a trend exhibited by the majority of Japanese android/robotics researchers, but from what I've seen [androidworld.com] they tend to follow [androidworld.com] a no-face design ethic that I'm most pleased with. I think it's safe to say that most people would find anthropomorphic robots that don't look 100% identical to people (there's something off with that one) very creepy.

        And besides, these Japanese robots look way cooler and have this implied subservience about them, at least to me. It's a lot harder to humanize and attach (scary) emotion to something that's faceless and non-human looking, rather than something that looks like a hairy/scary-ass rendition of a planet of the apes extra.
        • Best quote: "Normally, I wear fur, but since we're such close friends I feel comfortable about stripping off."

        • Re:The Japanese do it right (Score:4, Insightful)

          by Yokaze ( 70883 ) on Friday March 19, 2004 @05:07PM (#8615442)
          I don't think it is a particular ethic, as there research insitutes which concentrate on the facial part alone. IRC, the reasoning for the no-face design is, that when a robot has an actual face, people are trying to interprete its expressions. A hard to read face gives people a bad feeling. So, they are developing independently from the robots faces which can express "feelings".

          Robots have a different association in Japanese culture. In Western pop culture the first reference to a robot I can think of is Maria from Metropolis, in Japan Astro Boy. So, I'd say the bad feeling is also partly rooted in culture.

          > It is a lot harder to [...] attach [...] emotion to something that's faceless and non-human looking, [...]

          Um, I'd like to refer to Tamagotchis. It is not uncommon that people attach feelings to things.

          • thanks, i appreciate the insight, it's nice that i now have a reason behind the face-less design theme. i totally agree about hard to read faces...some folks' scariest childhood memories revolve around unblinking eyes or faces missing key components. an example would be if eastman & laird had kept the ninja turtle's pupils out of the cartoon show. they would have been way creepier and probably not done nearly as well.

            fyi, when i mean "emotion" i'm not just talking about positive, so-cute-it-makes
        • This has been noted before by Carl Jung (see his little-known theories of "uncanniness", can't find a link) and by a robot designer in Japan (can't find link to story on /., can't find anything today!) When something vaguely resembles a human it's amusing and cute, but when it reaches a certain threshold of similarity (which is to say it looks too much like a real human) people suddenly and severely dislike it.

          If you want people to like it, you have to keep pushing the similarity until people can't tell t

          • ... Carl Jung (see his little-known theories of "uncanniness", can't find a link) and by a robot designer in Japan ...

            I am not so sure about Jung, but the Japanese roboticist is Mashiro Mori. Insofar as I know, he was the first to put forth the "Uncanny Valley" theory. The idea is that people exhibit a positive reaction to a robot that is somewhat human. This is one side of the valley. However, people respond with disgust if the robot is very realistic but, for whatever reason, not quite right. The arc

        • I think it's safe to say that most people would find anthropomorphic robots that don't look 100% identical to people (there's something off with that one) very creepy.

          OK, what about C-3PO?

          I guess if you're going to build an anthropomorphic robot, you need to give it an irritating voice to balance out the face.

      • Seriously (Score:1, Funny)

        by Anonymous Coward
        Couldn't they have gotten a better face for the poor thing? It looks like it was dragged out of the dumpster by the dog, and then viciously attacked by mauling centipedes, whereupon it flopped on the metal body of the ape and took on a very sinister appearance. I vote they spend their budget on getting a cute orangutan face!

    • Re:King Louie's head on a toaster oven. Creepy. (Score:5, Funny)

      by rjelks ( 635588 ) on Friday March 19, 2004 @04:20PM (#8614718) Homepage
      OMG, I followed the link and yikes. It looks like a combination of the Terminator, Chucky and some scary-assed monkey thing. Very cool idea, but I'm going to have nightmares about this one.

      -
    • Fur?! Good god man!

      A thick blanket is what is needed here!

    • Forget artificial intelligence, they've created the first android butter (but-her) face.

      Yeah, she's got some nice internals, but her face....
    • It looks kind of like the Teddy Ruxpin Borg [nebe-b.org].

    • If you read Star Wars books like any respectable slashdottter, you'll find that C-3P0 was designed to be as neutral as possible to a wide range of cultures, it has eyes, fixed arms, neutral mouth with no teeth, open gesture.
    • I read a piece about him in the tree-unfriendly Mail on Sunday sometime last year and apparently he (or was it his wife?) tried mutilating an existing cuddly orangutan, but it didn't fit.

  • And $20000 dollars to finish my robot (Score:1, Funny)

    by Anonymous Coward
    My girl robot. <nerdy laugh>

  • Lucy's home page (Score:5, Informative)

    by morcheeba ( 260908 ) * on Friday March 19, 2004 @04:18PM (#8614681) Journal
    Lucy's home page [cyberlife-research.com] is an even better place for technical details, including an anatomical overview [cyberlife-research.com] and scrapbook pictures [cyberlife-research.com]

  • SSI? (Score:1, Interesting)

    by Anonymous Coward
    cluster-capable system implementing native SSI (Single System Image) which is something that no other operating system can do today

    umm...unicos/mk?

    dyyghrnmiw

  • scary (Score:4, Funny)

    by AnonymousCowheart ( 646429 ) on Friday March 19, 2004 @04:19PM (#8614695)
    he said: "can you imagine how scary it would be if she looked like a human baby?"
    did you guys look at the picture [bbc.co.uk] of that thing? It looks like my mother-in-law! Thats friggen scary! I guess he spend all the money on research, and not on matching eye's.:)
    • I guess he spend all the money on research, and not on matching eye's.:)

      It actually surprised me a bit to find out he didn't have mismatched eyes as well. My eyes have a slight variation in pigment, and I usually put some asymetry into the eyes of anything I draw just as a kind of signiture.

  • Extinct (Score:2, Interesting)

    by Ethernet_Jedi ( 763592 )
    Maybe he could build and program an entire zoo (extinct species and all)..No feeding, no vets, just an occasional tune-up :0
  • Even if her conversations are above the "why don't you just download me?" level, she won't get too far if she looks [bbc.co.uk] like she was standing in front of, and looking at a microwave oven as it was exploding.

    Maybe she was trying to download the oven?

  • Are there any... (Score:4, Interesting)

    by incom ( 570967 ) on Friday March 19, 2004 @04:22PM (#8614745)
    opensource AI projects? It'd be interesting to play around with something, even very primative. It' would need to be OSS so I could actually modify it though.
    • Depends on what you mean by AI, of course, but OpenCyc [opencyc.org] is a great project that could really use more contributors.

    • Re:Are there any... (Score:3, Informative)

      by devnull17 ( 592326 )

      There's all kinds of great stuff available, even if some of it is very old. You can even get an implementation of SHRDLU [semaphorecorp.com] with the mechanical components replaced by a 3D Java layer.

      SHRDLU, like most AI projects written in the past 40 years, uses LISP, so it's actually not that hard to read. (Incidentally, SHRDLU is more than a bit unstable, but if you can get it to work, it's pretty amazing, especially for something written in the 70's.) Definitely worth a look, if only for the "coolness" factor.

      One of

    • Re:Are there any... (Score:3, Informative)

      by asavage ( 548758 )
      It isn't free, but the MATLAB neural network toolbox is really nice and you can see and modify a lot of code. They have some nice demos like appcr1 which is a neural network that takes images of letters and can tell you what letter is being shown. It tests it with random noise being added to the image and it works quite well.

    • Re:Are there any... (Score:1, Informative)

      by tanksalot ( 762778 )
      You can also check out JOONE [jooneworld.com]

      It is a java based open source (developed on sourceforge) Neural Network Framework.

  • Realistic features to be added to Lucy 2.0: (Score:4, Funny)

    by mystery_bowler ( 472698 ) on Friday March 19, 2004 @04:22PM (#8614751) Homepage
    - Incredible tree-climbing ability
    - Facial-gesture mimicry
    - Pick parasites out of fur (useful!)
    - Poo-flinging

    And I don't know if it's all orangutans, but the ones at my local zoo have an affinity for tire swings. They wear through the rope and then roll the tire into the safety moat.
  • Did you see the picture of the orang in the article? Looks like the crypt keeper from tales of the crypt.

    An AI crypt keeper is the last thing we need.

  • Hi. I'm Troy McClure (Score:3, Funny)

    by Anonymous Coward on Friday March 19, 2004 @04:25PM (#8614810)
    hi. I'm Troy McClure. You might remember me from such robot-apes movies as "Bedtime for Bender" and "Bananabots: Gorilla Rampage"

  • Controversy continued on Monday as surgeons successfully transplanted little Django's brain into a robot monkey body. Scientists now say human-to-robot brain transplants will be possible within ten years. On a sad note, however, Django died late Tuesday, after drinking his own urine. (Sealab 2021, I, Robot [pod-six.net])

  • That Robot violates the #1 rule of robotics (Score:4, Funny)

    by Lord_Pall ( 136066 ) on Friday March 19, 2004 @04:29PM (#8614878)
    That it not be absolutely terrifying looking.

  • Bat-Shit Terrifying (Score:3, Funny)

    by Enigma_Man ( 756516 ) on Friday March 19, 2004 @04:30PM (#8614884) Homepage
    That is one of the most frightening things I have ever set eyes on in my entire life. I can't imagine that a baby would have been even _more_ terrifying. Look at the cover of the book. It resembles the aliens from "Mars Attacks" to me. Also, according to the article, Frankenstein is a robot? I always thought he was a meat-bag like us? And, do we all have to refer to the hour-too-long movie "AI" every time Artificial Intelligence is referenced? Would it be infringement otherwise? Yeesh, I can't stop looking at that train wreck of a face... haunt me all night. -Jesse
    • Well, Frankenstein was built, so even if he was composed of human body parts, wouldn't that make him at least half-robot.

      Quite a bit like fantasy flesh golem, but animated by technology instead of magic.
  • It's a DO-IT-YOURSELF Orangutan Erector set, from Hasbro Toys. Where can I buy one? I wonder if these will sell like Tickle Me Elmo's?

  • Question WRT development language (Score:4, Funny)

    by NecrosisLabs ( 125672 ) on Friday March 19, 2004 @04:31PM (#8614929)
    Is she programmed in Ook? [dangermouse.net]

  • At least they're thinking ahead (Score:4, Insightful)

    by robson ( 60067 ) on Friday March 19, 2004 @04:34PM (#8614970)
    Considering that the real ones are headed for extinction [orangutan.com], at least we'll have some facsimile as a reminder.

    (Not sure whether I should follow that with a winky-face or a sad-face.)
    • I think I would follow this with a scared-and-crapping-my-pants face myself, jesus, if this is what they look like, get rid of them!

      On a serious note, orangutangs going away is sad :(
  • I haven't been able to find any more of the sordid details about it, but I do remember this [delboy-enterprises.co.uk]

  • Whence the 'orang-utan' hifen?

  • Am I the only one drawing the connection between that escaped gorilla [bbc.co.uk] and this AI orang?

    We all know Bubbles was the mastermind behind the recent Jackson family assaults on our children and our public decency.

    And you know that gorilla that uses sign-language [koko.org]? Well, when the researcher tells you Koko loves you, Koko is really outlining the Monkey Master Plan to overthrow humanity.

    No wonder Heston joined the NRA...Those damn dirty apes...

  • I would bet (Score:4, Interesting)

    by nate nice ( 672391 ) on Friday March 19, 2004 @04:48PM (#8615172) Journal
    I would bet that the thought of a baby would be on the order of 1,000,000 times of that of this robot. And probably 100,000 times that of a real Orangoutang. Obviously I have absolutely no backing for those figures. As for AI, the studying I have done has made me conclude it's a failed, crack science at this point for people who really have no concept that a brain doesn't act like a computer, or a computer programmed to act like a brain. In order for this to work we have to be able to quantify a brains element, chemicals etc and we haven't much idea of most of these anyways, and if we do we don't have a clue as to how they function together.

    I'm just sick of recursive "best yet" algorithms that claim to be AI when in fact it's nothing more than deduced logic and we are, thankfully, a but deeper than that.

    So, go ahead and study AI as perhaps one day something may come of it but be realistic in that you're becoming skilled in a clever art of trickery and deterministic patterns. Good luck!

    • Nice rant. But I don't think you're right.

      For one thing, the point of AI research, short-term, is to provide machines that can do reasoning on their own. We aren't talking, ``I think therefore I am.'' We're talking about self-driving vehicles, for example (a very real possibility, even if DARPA's Grand Challenge was a bit of a bust). For that matter, there are plenty of reasonably efficient ways to make entirely self-learning software (expert systems and the like are good at this within a limited context

      • "AI research is, at least, churning out real results."

        But do you really think these results are really AI? Albeit I am all in all very uneducated in the grand scheme of the field but with my limited understanding and intuition I see more or less better ways to resolve patterns in things rather than emotional thought and decision. It's just something we have that is a part of intelligence that I think a computer or machine, perhaps even biological, cannot have if human developed unless we can eventually q
        • I'm reasonably knowledgeable about computer science, but when it comes to AI, I'm a layperson. So I'm not entirely sure what the goals of AI are, exactly. It's certainly a departure from traditional computation, in which the problems are quite clearly defined. But I don't think the intent is precisely to do things that are not Turing-recognizable; that is, the goal is not to do something that is beyond the realm of typical programming logic. Instead, it's to figure out how we humans make good guesses about
    • Though I don't understand how they measured this, but a human baby is supposed to be about the same intelligence as an adult chimp (or was it a gorilla? Don't remember).

      And the article was almost devoid of technical details -- but he did allude to the fact that this was something different from most approaches that are called "AI", so I was actually expecting maybe a more fundamental breakthrough, or attempt.

      Otherwise, I actually agree with the general sentiment of your post (and I don't think it's a pop
    • I would bet that the thought of a baby would be on the order of 1,000,000 times of that of this robot.

      Better we watch out then. That's only 30 years away in terms of compute capability if Moore's Law holds up (and it will hold up, because even if we can't make one chip faster, we can definitely keep adding parallelism).

      Obviously I have absolutely no backing for those figures.

      It doesn't matter. A human could be a billion times smarter, and that's only 30*2 = 60 years away because of the exponential

      • You are assuming that increasing the computational capability increases the "intelligence" of AI constructs in some linear fashion, which of course it does not. I could give you a machine that is a billion times faster than todays fatest machine and the current AI tech would still be at roughly the same level.
    • I think there is a distinction between machine intelligence and artificial intelligence. Machine intelligence is, as you say, a very long way off and would require huge advancements in materials science. I'd even argue that it will never happen. Perhaps a century from now we'll find that nano-scale engineering is done by what we'd call a genetecist instead of a physicist. And is a 'biological information processor' just a brain?

      Artificial intelligence on the other hand is simply the simulation of the

    • when you quantify a non-quantifiable thing, it doesn't matter what number you use. A baby is 10^26 times as intelligent or twice as intelligent. All we know is that the freaky robot isn't very smart.
    • The average human has 1 billion neurons. Lucy has about 30,000. In humans, each neuron has between 1,000 and 10,000 dendrites connecting it to other neurons. So basically, you're looking at an average of at least 1 trillion connections in the human brain.

      AI will be a lot more advanced when we can find a good storage and computational means of processing that kind of volume of information in parallel.
    • If we wanted to build a bona fide organic human brain, yes we would need to quantify not only the brain's elemental composition but also how it functions. You are right in that it clearly seems beyond our reach to do that.

      But AI isn't about building an organic brain. It simply seeks to replicate the output on a particular level i.e. not at the neuronal level but at the behavioural level say. To achieve the latter do we need to understand the deep functioning of the brain? There seems to be no compellin
  • Lucy's picture on the book cover is pretty heavily airbrushed... I bet once she looks in a mirror, she'll be in for a lifetime of robot-angst.

  • Thanks, BBC (Score:5, Funny)

    by barryfandango ( 627554 ) on Friday March 19, 2004 @04:56PM (#8615272)

    "... the cutting edge of artificial intelligence or AI, a title used by Steven Spielberg for his 2001 film starring Haley Joel Osment and Jude Law."

    Now that's good journalism: a little background about the history of AI for the lay-people who might be reading this article.

  • ...or does that kind of look like Tom Servo in a halloween disguise?

    "Mike, you'll NEVER be extreme" - Crow T. Robot
    • Servo had CURVES, baby. Mmm...

      "Luncheon meats make the sawdust in your stomach explode." - Crow T. Robot
  • My research is somewhat related (and open source) so I thought I'd offer up a link [sourceforge.net]. EBLA does bottom up language acquisition based on visual perception. Here's a short paper [cornell.edu].

    The work being done by Deb Roy's Cognitive Machines Group @ MIT [mit.edu] might also be of interest.

    Sure would be nice if Grand started making bits of code and a few technical papers available. Guess he can't just give it away if its his bread and butter though.

  • Two choices available for the head :
    Chucky, from Childsplay,
    And this freaky uncombed urang utang thingy :)

    I think i can understand why they went with this one. ;)

  • This'll get modded as redundant, as well it should. But... holy crap:

    http://www.cyberlife-research.com/diary/0104.htm [cyberlife-research.com]

  • AI?? (Score:2)

    by JohnLi ( 85427 ) *
    The article mentions that they dont use software to cheat. Does that mean that any sort of matching via a database is not "real" AI??

  • Brain Farts? (Score:3, Interesting)

    by Surak_Prime ( 160061 ) on Friday March 19, 2004 @05:15PM (#8615583)
    It has occurred to me, whenever the subject of AI is broached, that scientists seem to be doing a bang-up job of heading towards replicating the proper function of a brain in computer hardware, but none of the projects I've seen try to replicate the errors that result when the brain cell sending or receiving a message dies, is replaced incorrectly, is deformed one way or another, or is subject to any of the other myriad flaws of flesh.

    Could it be that sentience, in the end, is the result of brain farts?

  • Sounds like inferior cephalopod nerves to me (Score:1, Interesting)

    by Anonymous Coward
    Your nervier (brainier) mullosks have amazing nerve fibers. They get used for experiments all the time because they're just huge, big enough to place electrodes in the axons and measure voltage changes.

    Guess flexible wiring is more pleasant to be strapped into than a squid or a cuttlefish, though I doubt it'd be as fast. Cephalopods have very fast nervous systems, they're lightning quick partly as a result.

    rjt

  • the japanese developed a nice looking actroid:
    Realistic Japanese Female Actroid Robot [i4u.com]

  • Lucy's brain (Score:2, Interesting)

    by topynate ( 694371 )
    has 50000 'neurons'. Does Steve Grand really think he'll approach mammalian intelligence with so few? I agree strongly with him giving Lucy a rich environment, but maybe he should be looking at using something like FPGAs to get more neurons on board for a reasonable cost. That's what Hugo de Garis [usu.edu] is doing, and he had much more ambitious plans. The company he was working for failed though, so I don't know whether he's still making progress in actual building of AI. Anyone?
  • When she can drink beer out of the can, lift cars and hit/give the finger to hells angels.

  • A new "ism" approaches (Score:3, Insightful)

    by Trolling4Dollars ( 627073 ) on Friday March 19, 2004 @06:06PM (#8616176) Journal
    I grow weary of all the people with cyberist attitudes who are scared of machines simply because they don't look exactly like us. If you look at the best of our CG characters today, they STILL don't look like us. I imagine that the first humanoid robots will probably look a lot like CG characters come to life. Get rid of those old fashioned attitudes... ;P

  • Creatures (Score:2, Interesting)

    by metaomni ( 667105 )
    Something that is worth noting is the 'game' that Steve Grand helped create. The entire Creatures series (1, 2 and 3) was revolutionary in the AI-software industry. It melded a game that anyone could relate to, with some serious AI running in the background. The whole concept for the games was fascinating. It's a shame that his company has now gone under. The series in its hayday had a cult following, and I'm sure there are still some out there who play it.

    Man those games were obsessions...

  • And I quote: "...conjures up fears of super-clever robots, the likes of Frankenstein..." Umm yeah.
  • Lucy isn't the most attractive robot I've ever seen. Here's my own robot-ape. http://www.fuzzgun.btinternet.co.uk/flint/index.ht m My open source AI software can also be found here: http://www.fuzzgun.btinternet.co.uk/rodney/compone nts.htm

Slashdot Top Deals

You are always doing something marginal when the boss drops by your desk.

Close