I've long wondered if we can make a humanoid robot do a 𝘄𝗮𝗹𝗹𝗳𝗹𝗶𝗽 - and we just made it happen by leveraging 𝗢𝗺𝗻𝗶𝗥𝗲𝘁𝗮𝗿𝗴𝗲𝘁 with BeyondMimic tracking!
This came after our original OmniRetarget experiments, with only minor tweaks to RL training: relaxing a termination threshold and removing one reward term.
The policy achieved a 𝟱/𝟱 success rate in our real-world experiments, showing the strength of high-quality, interaction-preserving motion retargeting combined with BeyondMimic’s minimal RL tracking.
Here is the updated arXiv: https://t.co/0dSdpYJAV6 (In Sec. V. A)
On a street in Shanghai, a humanoid robot traffic coordinator is clearing traffic jams. Judging from the purple waistline and height, this is Fourier's GR1,which is 1.65 meters tall.
The ORCA v1 hand is a 17-DoF, tendon-driven, humanoid hand with integrated tactile sensors and poppable joints.
One fully assembled hand is priced at $5,937.00. The design is open-sourced for non-commercial use.
Four teams of humanoid robots went head-to-head in China’s first 3-on-3 soccer matches. Booster T1 robots were programmed by four university teams to play fully autonomously.
The event served as a preview of the upcoming World Humanoid Robot Games, set for August in Beijing.
Imagine you had a humanoid robot. You could teach it to do anything. When it recharged, it could upload its learnings (on an anonymized basis) to all the other robots. It could be the most powerful data network effect in history.
This will likely produce a business ~100x bigger than Apple's market cap (or more)
What's clear now is we're in the right decade for humanoid robotics - this will feel like the future jumped ahead by 50 years
Humanoid robots at scale will feel like Alien technology:
→ Self-replicating factories of robots
→ Factories, job sites, hospitals, warehouses - automated
→ A personalized robot for every person
Physical work will be optional