A computer at Carnegie Mellon University has been reading the internet for the past 10 months, spotting patterns in language and sorting things into categories.  It knows Peyton Manning is a football player and the Indianapolis Colts is a football team, for example.  It can go one further and estimate that Manning plays for the Colts and tell you how certain it is of that connection.

The Never-Ending Language Learning system, or NELL, has made an impressive showing so far. NELL scans hundreds of millions of Web pages for text patterns that it uses to learn facts, 390,000 to date, with an estimated accuracy of 87 percent. These facts are grouped into semantic categories — cities, companies, sports teams, actors, universities, plants and 274 others. The category facts are things like “San Francisco is a city” and “sunflower is a plant.” [NewYorkTimes]

The technology holds great potential: search results as a written reply instead of just links, computerized personal assistants that respond to questions, and other important things I can’t be arsed to look up.  I tried to reach this fancypants computer to get a comment, but he called me a n00b and told me “T*ts or GTFO”.  Hey, you’re only ten months old, mister.  Don’t think I won’t wash your ports out with soap.

Naw, just kidding.  NELL didn’t say that, and she’s not even a he.  What she said was, “Tay ina win.”

Tay ina win.

Read the rest of this entry »

If the swimming robot wasn’t terrifying enough for you, there’s good news, as Carnegie Mellon University has released a new video of their tree-climbing robot.  This biomimetic robot mimics the movements of a snake to reach places people and other robots can’t reach.

Moreover, these highly articulated devices can coordinate their internal degrees of freedom to perform a variety of locomotion capabilities that go beyond the capabilities of conventional wheeled and the recently developed legged robots. The true power of these devices is that they are versatile, achieving behaviors not limited to crawling, climbing, and swimming. [CMU Biorobotics Lab]

Wait, not limited to crawling, climbing, and swimming?  What else?  Do they jump?  Holy crap they totally jump don’t they?  I knew it.  I bet they can fly, too.  Oh lawd flying snakebots, I need to lie down.  Oh god, it can do Inception too can’t it?

Read the rest of this entry »

While the Tetris was only meant to demonstrate the abilities of a water screen, the Tetris is what everybody’s focusing on, because as a species, we’re so obsessed with Tetris we’ve built robots out of Legos to play it for us. Although we’ll admit that the water screen is pretty neat, if incredibly low resolution.

If you were wondering what the deal with the water screen is, it uses three layers, dripping water out of needles, while a projector uses precise timing and projecting onto the three different layers to create the illusion of a consistent screen. Unfortunately, the resolution is absolute crap, about 50 by 60 drops, but that’s more than enough to score four lines.

In Tetris, we mean. Expect to see this technology in overly elaborate office buildings around the world in about five years.

Read the rest of this entry »

Wonk-Eye Cat reads the double entendre headline.  And the side banner.

The geniuses at the Biorobotics lab and Manipulation lab at Carnegie Mellon University have come up with an awesome new design for a vibrating robot that can climb up different-sized circular and square tubes.  It can travel at a speed of 20 times its body length per second (holy crap) and carry five times its own weight.  Somewhere an ant just crapped its little ant pants.  You’ve been usurped, ant.  Also, who gave you pants?  You’re an ant.  That’s just silly.

Its simple motor turns an unbalanced mass at a uniform velocity. As the mass swings around, it causes the robot to bounce back and forth between the tube walls. Two rubber o-rings let the researchers specify the exact contact points and increase friction with the walls.  [IEEE Spectrum via NextBigFuture]

Previous dynamic climbing bots have used bristles to create anisotropic friction, making the bot travel in the direction with less friction.  Getting it back out of the tube was a problem with these robots.  This new design doesn’t use anisotropic friction, as you can see in the video below, which was the winner for best video at ICRA this month.  Really?  Man, just wait till IEEE gets a load of our animated videos.  We’ll sweep the competition next year with our design for a microbot that removes pants from ants, because that sh*t needs to stop immediately.

Read the rest of this entry »