A curling stone slides on the ice and knocks the opposition's stone from the target circle.
The stone was thrown by Curly, a pair of artificial-intelligence-driven robots.
The skip robot acts as the captain and formulates strategies after inspecting the game situation with its camera.
It sends the information to the pitcher robot, which pushes the stone towards the target.
Just like AlphaGo, Curly studied thousands of deliveries from curling matches.
"Using deep learning based on 160-thousand deliveries from more than 13-hundred curling matches, the robots can control the direction, speed and rotation of the delivery."
The robots also learned to take into account the stadium's temperature and humidity, which affect the ice.
And for the first time ever, the machines faced human players in a curling game but fell to a 3-nil defeat.
"The curling robots formulated strategies pretty well, but failed to adjust to variables of the ice."
The AI robots will be used for developing strategies and providing training support.
Park Se-young, Arirang News.