Seen here about to get his coffee roofied as soon as he turns away, a professor from the Tokyo University of Technology shows off a robot fit with a "self-organizing incremental neural network" that allows the bot to learn, think and act on its own. You know, so robots really won't need us anymore. Now if you'll excuse me for a second...Hello, Mr. President? Forget the economy going to shit for a second -- we have an even bigger problem. CODE SILVER, MR. PRESIDENT -- CODE SILVER! (Because robots are silver)
In this experiment the robot will solve a problem by deciding what actions it should take and in what order. The robot will be told to pour a glass of water, make it cold, and give it to a person. The robot will decide how to do this while being aware of its surroundings and its own situation.
If the robot is told that cold water is wanted, it recognizes that after pouring the water, it can't pick up ice straight away, because it's hands are already full with the glass and the bottle. So it chooses to put the bottle down, and then put the ice into the glass.
As well as the robot's sensory information, in the form of visual, auditory, and tactile data, SOINN obtains information from other sources, including the Internet and other robots' experiences and knowledge. In this way, it gradually becomes smarter.
You know who else can learn, think and act for themselves? Troublesome teenagers. Food for thought. "Like a gallon of ice cream?!" HA -- a gallon of ice cream isn't food -- it's why you're fat. Hot Pockets for thought.
Hit the jump for a video demo of deathbot's capabilities.
Thanks to Michelle, Spidox, carrie, THE_DON (like, THE Don?!), Erik and anybody else who sent me this tip whose email I couldn't find because every other one in my inbox is something for boner pills and I get distracted comparing prices, who agree there are only two things robots need to learn: obey. AND DIE.