Slashdot Mirror


GRACE Exceeds Expectations!

smashr writes "GRACE, the robot mentioned earlier on slashdot, has succeeded in the AAAI challenge at the conference in Canada. Her creators are saying that GRACE exceeded their expectations. The entire competition went well with only several minor hiccups (GRACE cut in front of a judge in line to register, and then demanded a conference badge several times). The team is looking forward towards refining GRACE for the competition in Mexico. Stories at: CNN.com, Yahoo, and the Edmonton Journal."

10 of 209 comments (clear)

  1. Major failing of GRACE by Tattva · · Score: 5, Funny
    Unfortunately, GRACE does not have a flexible torso, and will therefore be unable to perform a requisite skill in the academic/conference field: kissing ass.

    --
    personal attacks hurt, especially when deserved
  2. Blind leading the blind by CommieLib · · Score: 4, Funny

    Does anyone else see a problem with a bunch of robotics researchers teaching a robot social skills?

    Relax, it's a joke.

    --
    If your bitterest enemies are people who hack the heads off civilians, then I would say you're doing something right.
  3. Cutting in line and being demanding... by Microlith · · Score: 4, Funny

    Not only is it sociable, but it's rude too!

    Sounds about normal for a lot of people. In fact the blatant disregard for others should earn it points for being more human than necessary.

  4. Not a Sci-Fi convention, though. by Remus+Shepherd · · Score: 5, Funny

    Wake me when GRACE is able to sign up at a sci-fi convention. Applicable skills will have to include: Giving backrubs to others standing in line; Recognizing the registration counter people as objects to talk to even when they're wearing klingon costumes; and bitch-slapping the crowd of fanboys around her chanting "Exterminate...exterminate!"

    --
    Genocide Man -- Life is funny. Death is funnier. Mass murder can be hilarious.
  5. Beings that surpass their creators? by jethro_troll · · Score: 5, Funny

    When I went to AAAI in Philly in 1986 (in my LISP hacker days) half of my coworkers (mathematicians, linguists, logicians, all damned good AI researchers) either got lost en route to the hotel, or got on the wrong shuttle bus to the conference, or forgot their presentation slides, or...

  6. What they're good at. by drox · · Score: 5, Insightful

    The GRACE design team deserves kudos, but I still think that robots/AI should primarily be designed and programmed to do things that humans are BAD at, like searching through dangerous rubble, or performing fine manipulations in toxic or extreme-temperature environments, rather than doing things that humans are already quite GOOD at, like schmoozing. There are billions of people available who already know how to schmooze, and they can learn new schmoozing rules quickly, on the fly, without costly reprogramming. There are very few who would be willing (to say nothing of able) to work in a hazardous or tiny confining environment.

    1. Re:What they're good at. by ultramk · · Score: 5, Funny

      rather than doing things that humans are already quite GOOD at, like schmoozing.

      Don't know many scientists, eh?

      There are very few who would be willing (to say nothing of able) to work in a hazardous or tiny confining environment.

      What, like a cubicle?

      m-

      --
      You catch enchiladas by picking them up behind the head and holding them underwater until they don't kick anymore -VeGas
  7. Re:[insert intelligent comment here] by drudd · · Score: 4, Insightful

    The real problem with implementing Asimov's three laws is interpreting when your actions will break them.

    The very top of the line research in AI might be able to recognize a human being under idealized circumstances.

    So how are we supposed to tell our robots "don't run over a human's foot, since that's injuring a human, and that action violates the first law" when the robot can't distinguish between a human foot and a stain on the carpet.

    Doug

    --
    Venn ist das nurnstuck git und Slotermeyer? Ya! Beigerhund das oder die Flipperwaldt gersput!
  8. Not really autonomous by Devlin-du-GEnie · · Score: 4, Informative

    The Edmonton Journal article says that GRACE's handlers used hand signals and voice cues to help her navigate and perform. That makes her registering for the conference merely interesting, not jaw-dropping amazing.

  9. cutting in line... by furchin · · Score: 4, Interesting

    The CNN article states that the robot bumped into a judge, rather than cut in line. That's a significant difference. Cutting in line indicates a tempramental personality, with some true intelligence perhaps. Bumping into a judge indicates that the programmers in charge of GRACE failed basic obstacle avoidance -- which boils down to the following for loop:

    for (int i = 0; i<num_sensors; i++)
    if (sensor_distance[i] < 5 inches)
    motors = off;

    I'm involved in a lot of robotics work, and while I believe that robots should eventually attain very intelligent behavior, I also believe that the first priority in programming a robot is to ensure it does not harm humans. By bumping a judge, GRACE has shown that it is not capable of functioning safely in society. If it bumps a judge, what's to keep it from running a judge over and killing him? Standard robots the size of GRACE are 300 lbs, quite capable of inflicting significant damage.

    As a side note, most robots have touch sensors on their side panels that automatically shut off power to the motors when they are triggered. I'm willing to bet that this is what kept GRACE from running over the judge.