The story: "Humanity is on the road to losing another game to the superior programming of a machine," says Mike Epstein at Geekosystem. Researchers at the Technical University of Darmstadt in Germany have created a robotic arm that learns ping-pong from humans and picks up new moves as it competes against them. Head researcher Katharina Muelling and her team guided a robotic arm through increasingly difficult ping-pong shots fed from a human opponent. After taking in these moves, the robot was able to use the basic information it learned to respond to new and more difficult game scenarios it wasn't trained to react to.
The reaction: While "other robots have played table tennis in the past," says Hal Hodson at New Scientist, this one is the first to use human demonstration to learn the game, therefore allowing this robot to "play more like people." Not only is that impressive, says Epstein, but "from the sound of things, it also sounds like the best player yet." After a mere hour of practice, the robot was successfully returning 88 percent of shots — "I've been playing the game for years and don't hit that many."