Vision-Superior Robot Trumps Humans At Rock-Paper-Scissors, Ping Pong Balls

For the record, I'm pretty darn amazed at how speedily and accurately the embedded vision demo co-developed by Analog Devices and BDTI accomplishes its dice-counting task: much faster and more precise than me. With that said, I'm really amazed at a robot named Janken, the Japanese word for the rock-paper-scissors game also known as roshambo.

Within a millisecond, as IEEE Spectrum explains, Janken (developed by the Ishikawa Oku Lab at the University of Toyko) determines what hand shape you're in the process of making; within a few milliseconds more, it's determined what shape its "hand" needs to make to beat you. But it all happens so fast that, as the video above shows, it seems like Janken is playing against you fair and square…and beating you every single time. More coverage comes from Boing Boing and Slashdot.

Believe it or not, however, Professor Ishikawa from the Ishikawa Oku Lab felt that the above clip "lacked enough technical details and wasn't satisfying for robotics researchers". So, a few weeks later, the lab published another video (below), which shows how their system's 1,000 frame-per-second camera and pan-and-tilt subsystems are able to accurately track an in-flight, high-speed ping pong ball, dynamically adjusting to keep it perfectly centered in the camera's frame. Amazing. And, for the record, befitting from way more available compute horsepower (PDF), along with a much higher associated bill-of-materials cost, than that which exists in the ADI/BDTI demo.

Here you’ll find a wealth of practical technical insights and expert advice to help you bring AI and visual intelligence into your products without flying blind.



1646 N. California Blvd.,
Suite 360
Walnut Creek, CA 94596 USA

Phone: +1 (925) 954-1411
Scroll to Top