- Galbot found an unprecedented way, called LATENT, to train robots
- Using “skill snippets,” they trained a Unitree G1 robot to play tennis
- The robot developed relatively strong tennis skills based on this minimal training.
The future Wimbledons, in which a fifth-seeded tennis pro faces a sixth-seeded robot, has just moved from the realm of science fiction to something that seems inevitable.
How do we get here? The fault lies with Galbot and his LATENT innovation.
The best robotic athletes—those who can practice karate, box each other, or do parkour—are either remotely controlled or highly programmed to perform a set of pre-programmed actions. Real-time competition against, say, a human opponent is believed to be difficult or impossible. But now Galbot and a team of researchers have done it: They used minimal learning to teach a Unitree robot how to play tennis against an unpredictable human opponent.
Article continues below.
They call it “Learning humanoid tennis athletic skills from imperfect human motion data” (which they manipulated in LATENT). Instead of highly detailed robotic training that captures the full range of human tennis skills, LATENT focuses on “movement snippets that capture the primitive skills used when playing tennis.”
Look
Somehow, the researchers figured out how to use these fragments of tennis skills, or what they call “imperfect” data, to provide enough information about “primitive human skills in tennis scenarios.”
The robot, a Unitree G1, can then use those fragments to make sense of live play and, according to the researchers, “consistently hit incoming balls over a wide range of conditions and return them to target locations.”
That's a dry way to describe what happens in the amazing demo video in which a Unitree G1 robot skillfully outplays (and sometimes outperforms) a human tennis player.
As they note in the research summary: “Our method achieves surprising results in the real world and can stably sustain multi-shot rallies with human players.”
Now, you can watch the video and assume that the human is being gentle with the robot or even pointing the ball in the robot's direction. That's possible, but how do we explain the robot constantly returning the volley and purposely placing the ball where the human player is not? It seems downright competitive.
Obviously, the robot could be better. He often seems to be on the verge of disaster, and that noise seems fused with his right arm. I'm not even sure how the robot would handle a shot going over its metal and plastic head.
Still, from the ball return to a wicked backhand, the quick footwork and even the ability to stand upright, the Galbot/LATENT robot puts on quite a show.
It's not too difficult to imagine where this leads. Give it time and a robot like this could be playing exhibition matches against someone like Rafael Nadal.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds. Be sure to click the Follow button!
And of course, you can also follow TechRadar on YouTube and tiktok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp also.
The best business laptops for every budget
