Group Actions
Engagement
  • Author Author: Catwell
  • Date Created: Date Created
  • Views 741 views
  • Likes 1 like
  • Comments 1 comment
Related
Recommended

Robots already dominate humans, at rock paper scissors

Catwell

 

Robots have already beat humans in chess, and at the University of Tokyo, Ishikawa Oku has developed a robot that will always win at rock paper scissors (or “janken” in Japan).

 

It achieves a 100% winning rate by analyzing 1000 frames per second that allow it to react milliseconds before the match is over. The robot only needs 1 ms to see the shape your hand is starting to make and it reacts to counter your move before you can tell it cheated.

 

Watching 1000 frames per second, the robot is patient enough to see the very last shape you are going to throw so tricking it is not an option. The robot is able to coordinate its robotic wrist joint with the 1,2, 3 motion of rock paper scissors as the human player moves. It will not shoot until the last possible millisecond to assure its winning decision, and it’s timing is impeccable.

 

This type of robot could be used to assist humans by taking advantage of information that humans usually miss, like what happens right before an accident. This high speed visual recognition could also be used for reading lips or to instantly coordinate jobs that are too much for a single person to handle.

 

Cabe

http://twitter.com/Cabe_e14

  • I am not surprised by the results.

    With enough samples, you can build up a reasonably accurate predictive algorithm to immitate your personnal random solection process.

    Everyone can become predictable for this level of simple task.  Most people are not aware that they have individual tendencies in most things.

    Some people have learned to observe and learn these patterns quickly, which is why some people can win most of the time while others seem to lose.

     

    I worked with Dr. Peter Raeth on using a radial basis function to build up a frequency model for predicting background signal levels.  We demonstrated that the algorithm could improve your signal extraction capabilities significantly.  Even in a supposed high noise environment, we found a predictable multifrequency pattern.

     

    As they say, you have to understand the math to understand the game.  Even complex games can become predictable.  Otherwise no computer could defeat a skilled chess master.

     

    Just a thought,

    DAB