When two people need to decide who will take out the trash or who will get the last bottle of soda in the refrigerator, they sometimes resort to flipping a coin. Of course, that requires one of them to have a coin, which is why a quick round of the traditional hand game “rock-paper-scissors” is often used instead. It requires no extra equipment, each player has control over half of the inputs, and the results are decisive, unlike when the coin hits the floor and disappears under the sofa.
[Credit: Ishikawa Oku Laboratory, University of Tokyo]
But what if you could turn to a robotic surrogate to play on your behalf? Researchers at the Graduate School of Information Science and Technology at the University of Tokyo have developed a me... that plays the familiar game against a human opponent. And it has the remarkable trait of never losing.
How does it achieve this perfect record? You might conclude that it has some marvelous new mind-reading ability, when in fact the system is much simpler than that. The hand is linked to a digital camera that watches the opponent’s hand movements. This provides the information that the robot needs to synchronize the up and down motions required for the preliminary stages of play; you move your fist up and down three times before displaying your hand shape for rock, paper, or scissors.
The machine vision system can actually recognize a hand shape in about a millisecond, and the robot is able to display its move in about another millisecond. The net effect is that the robot’s move appears to occur simultaneously with the human opponent’s. When you slow down the video by about 50 times that you can detect that the robot makes its move only after the human’s move is revealed.
While this game is an entertaining technology demonstration, it points the way toward interesting applications for machine vision leading to mechanical responses. We are already seeing similar innovations in cars that can brake automatically to avoid obstacles ahead or can maneuver to avoid crossing lanes inadvertently. Such mechanical devices could be used to make systems more energy efficient and safer than a human operator.
There is no word, however, whether the University of Tokyo researchers plan to add “lizard-Spock” to their robot’s abilities.