Watch an AI robot learn how to demolish humans at a marble maze game

Researchers have developed an AI-powered robot they claim can beat the physical marble game Labyrinth faster than humans are capable of. Thomas Bi and Raffaello D’Andrea of ETH Zurich created CyberRunner, which combines model-based reinforcement with the dexterity required to beat a game that requires physical skill, coordination and precision.

For the uninitiated, the aim of the game is to guide a marble through a maze without falling into any holes. The player controls the movement of the ball by rotating two dials, which tilt the board.

CyberRunner learns through experience. A camera observes the game and an algorithm learns more about it from each attempt. "Based on its understanding of the game it recognizes which strategies and behaviors are more promising," the researchers say. As such, the robot kept getting better at the game.

The researchers gave CyberRunner and several humans around six hours of practice with the game. While the humans by and large struggled to beat Labyrinth after that time, CyberRunner was able to conquer it in just under 14.5 seconds. The researchers claim that's faster than any previously recorded time.

CyberRunner became so adept at the game that it was able to use some unintended shortcuts. The researchers had to step in and instruct the AI to follow the maze's correct path.

We've seen AI models outperform humans in other games, such as chess, Go and Dota 2. However, we haven't seen too many instances of AI performing better than humans in games that require a physical skill component.

Meanwhile, the researchers are open-sourcing CyberRunner. Maybe I can help train it to be better at Screwball Scramble than I ever was as a kid.