For those of us who have no friends, we at least have video games. And thanks to artificial intelligence, the computer-controlled adversaries we play against are always getting better. Most games aren’t at the point where the AI can give a skilled human serious competition without having the playing field tilted in its direction, but that’s okay, because we still derive pleasure and satisfaction from beating them anyway.
What AI is missing is physical embodiment. You know, something that you can scream at when you lose and gloat over when you win. The little humanoid NAO fills that role nicely, and researchers from the University of Tsukuba in Japan have taught the robot to play Wii Tennis, with a Wiimote and all.
NAO is actually playing the game, although it’s not using onboard visual processing to do it. The video output from the Nintendo Wii is passed through a computer that does all of the heavy lifting, recognizing the locations of the players and changing its strategy accordingly. The computer then sends control signals to NAO, which performs pre-designed gestures, just like a human player does.
What’s the point of all this when you could just play against the built-in AI in a game? By using an AI that’s not a part of the game, you can upgrade it yourself, and (hypothetically) use the same robot to play a whole bunch of different games. The robot can learn how you play, what you like, and what you hate, and use this knowledge to challenge and/or frustrate you. Perhaps more important, if you have a cool video game-playing robot at home, you can bet that lots of people will want to be your friend.
“Gaming Humanoids for Facilitating Social Interaction Among People,” by Junya Hirose, Masakazu Hirokawa, and Kenji Suzuki from University of Tsukuba in Japan, was presented at the 2015 ACM/IEEE International Conference on Human-Robot Interaction.
[ HRI 2015 ]