A computer vision system helps the robot recognize an object’s shape, size, color and orientation. ITRI’s computer vision robot serves coffee to its opponent during a game of chess at CES in Las Vegas on January 5, 2017. The robot uses a computer vision system as well as deep-learning features to gently handle the chess pieces and react to the moves of its human opponent.
Player and robot communicate via a tablet, with the robot frequently asking for time to think before it decides on a move. The robot’s movements aren’t perfect (it failed to set down a chess piece when the chessboard was slanted), but it still does really well with gripping and precise movements. During the short time I was at the booth, the robot defeated its human opponent twice.
But the robot has a softer side, too: It served its opponent coffee as a demonstration of its vision system and dexterity. Though it was a bit slow, the robot smoothly filled the coffee cup on the table without spilling a drop.
ITRI says it envisions the technology being used in assembly lines as well as in hospitals to care for the elderly.
Robots that learn from experience and smart, autonomous drones are quickly moving from science fiction to reality and are on display at CES in Las Vegas. The Industrial Technology Research Institute (ITRI), based out of Taiwan, is showcasing these two technologies that could one day help robots take over the world.
The first is ITRI’s Intelligent Vision System, which allows robots to interpret the visual world, act on visual information, and learn from experience. That’s right, learn.
Many robots are programmed to perform a task repeatedly at a specific time and location. In other words, they don’t learn— they just do. The Intelligent Vision System allows the robots to adapt to their conditions.