Russian Robot Breaks World Record for Solving Rubik's Cube

MOSCOW, August 1 A Russian-developed robot has broken the world record for the fastest Rubik's Cube assembly — it assembled it in 0.203 seconds, while the previous record holder was a robot from the Japanese company Mitsubishi Electric Corporation with a result of 0.305 seconds, said the robot's creator, Yandex developer Alexander Krotov.
«The robot's result is 0.203 seconds. The record was recorded at a special event on July 5, 2024. I designed the robot specifically for solving the cube and used publicly available components — for example, motors for autonomous vehicles and cameras for game consoles. The previous world record for solving the cube was 0.305 seconds It was installed by a robot from the Japanese company Mitsubishi Electric Corporation. It created this robot from components of its own design,” Krotov said.

For comparison, during the time it takes the robot to develop, the human brain only has time to comprehend the image it has seen, and the eye to blink. The human record in speedcubing is 3.13 seconds.

In 10 milliseconds, the robot films the faces with its cameras and processes the images with a special program, then plans its actions for 8 milliseconds, calculating the assembly sequence in the least number of moves. Then it assembles the cube in 184 milliseconds — simultaneously moving its faces with the help of six motors.

«To break the world record, I optimized the robot's code and configured it so that its components interacted with each other as quickly as possible — for example, when shooting and processing images from cameras. I used specialized drivers to control various elements of the robot, including motors. Among the publicly available motors, I chose those that quickly gain high speed — similar models are installed in delivery robots and electric scooters,» the developer added.
Krotov worked on this project for several years, during which time he created a number of modifications of the robot. At Yandex, he develops the infrastructure of language neural networks that are used in various company products, such as Search and Neuro.