Students at ETH Zurich from Switzerland here have designed a robot that operates 30 times quicker than the fastest human. The “Sensors Group” of Institute of Neuroinformatics at the ETH Zurich has employed a camera and brain-motivated neural network to design “Dextra – a robotic hand.” The robot essentially reads your mind predicting gestures of your hand to beat you in the game of rock-paper-scissors.
“Dextra looks at the world in extremely slow motion. It is so quick at viewing the symbol you are about to show that it is capable of both determining your next move and executing a winning symbol 30 times quicker than the fastest human,” claimed Tobi Delbruck, ETH Zurich Professor, to the media in an interview. “Unlike traditional AI (artificial intelligence) vision systems working on picture frames, Dextra wins by being independent of frames,” he claimed.
Speaking of robots, future robots can be helpful in offices and homes, thanks to MIT researchers who have designed an enhanced computer vision that allows devices to accomplish specific tasks and inspect random objects. Innovations in computer vision have allowed robots to make basic differences among objects.
On the other hand, the systems do not really recognize shapes of the objects, so there is little the machines can do after a swift pick-up. The new system invented by scientists at MIT (Massachusetts Institute of Technology) in the U.S., dubbed as DON (Dense Object Nets), sees objects as series of points that serve as type of visual maps.
This method allows robots better manipulate and understand items, and, most significantly, permits them to even pick up a particular object among a mess of similar objects. “Many methods to manipulation cannot recognize particular parts of the object across the different orientations,” claimed a PhD student at MIT, Lucas Manuelli, to the media in an interview.