We don't really think about robots in factories making mistakes. They pretty much do what they're programmed to do over and over again, right? But the more we rely on robots to do varied things, the more they're going to encounter mistakes.
Researchers at MIT have been thinking about that and, working with an industrial robot, they've discovered a novel way for the 'bot to realize it was making a mistake and correct itself: by reading a human's brain waves.
It's weird to see a robot hesitate as it sorts things into boxes, drop something into the wrong box, then pick it up again and put it in the right box. But it's not really making its own choice – it's taking a cue from the person in an EEG cap sitting across from it.
One of the curiosities of the brain makes this possible. Although our brains send out incredible noise, the signal it sends out for potential errors is much stronger than the others. Connected by an EEG cap, the robot can easily pick up on that signal to realize it's doing something wrong.
It's an incredible shortcut. Rather than having to shut everything down, examine the code, and re-program the robot, the robot can just take a split second to adjust to the brain waves it's picking up.
There are some limitations at this point, however. For one thing, it can only distinguish between two options, like where the paint goes versus where the wire goes. For another, it might still make the same mistake over and over again. Researchers would like to teach it to learn from its mistakes and not make the same ones in the future.
Nevertheless, it's a cool technology with tons of promise. Not only could a human oversee a factory, but humans in self-driving cars could let the robot at the wheel know it was making a mistake.
And just think of how this might help someone who can't verbally communicate!
0 comments: