GMO Internet Group, Inc. has announced plans to launch a trial project in which a humanoid robot will learn to run by analysing data collected from elite Japanese athletes, using artificial intelligence to replicate their movements.
The company said on Thursday that the initiative marks what it claims to be the world’s first attempt to enable a robot to autonomously reproduce human running techniques based on motion data captured through sensors and other tracking systems.
As part of the project, running patterns from athletes on GMO’s track and field team will be digitised and fed into the company’s AI platform. The group includes Yuya Yoshida, who represented Japan in the marathon at the World Athletics Championships in Tokyo last year, and Asahi Kuroda, a standout performer in this year’s Tokyo-Hakone Intercollegiate Ekiden.
The AI system will then be integrated into “G1,” a humanoid robot developed by Chinese robotics firm Unitree. The robot stands 1.3 metres tall, weighs 35 kilograms, and is currently capable of reaching speeds of up to 3 metres per second—equivalent to completing 100 metres in around 30 seconds.
During a demonstration in Kawasaki on Thursday, the robot struggled to maintain balance while running and was seen falling at times, highlighting its developmental stage.
Despite this, GMO aims to eventually compete in a track and field event at the World Humanoid Robot Games scheduled to take place in China in August.
The company currently owns 10 humanoid robots and is deploying them across various organisations, with long-term plans to use them for tasks such as warehouse and factory logistics, as well as disaster rescue operations.
