Loading video...

Video Failed to Load

Go Home

๐Ÿ”ฅ INTERESTING: A humanoid robot learned tennis from just 5 hours of motion data, now rallying at 90% success.

Cointelegraph's profile picture

Cointelegraph

2,940,506 subscribers

127,951 views โ€ข 5 months ago โ€ขvia X (Twitter)

Health & WellnessNews & PoliticsScience & Technology
LIVE

Anya Rossiโ€ข Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Scientists have taught a humanoid robot to play tennis using just 5 hours of motion capture data. After training, the robot is capable of sustaining long rallies with a human and returning balls at speeds exceeding 15 m/s with an accuracy of about 90%.
1:29

Scientists have taught a humanoid robot to play tennis using just 5 hours of motion capture data. After training, the robot is capable of sustaining long rallies with a human and returning balls at speeds exceeding 15 m/s with an accuracy of about 90%.

Massimo

49,108 views โ€ข 4 months ago

๐ŸŽพYour humanoid tennis player is here!๐Ÿค– Introducing LATENT (Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data) โ€” the worldโ€™s first real-time whole-body planning and control algorithm for athletic humanoid tennis. For the first time, a humanoid robot can sustain high-dynamic, long-horizon tennis rallies with millisecond-level reactions, precise ball striking, and natural whole-body motion. This marks a leap from mechanical motion imitation to intelligent, decision-driven athletic interaction. #Galbot #Robotics #Innovation #AI #Technology #Humanoid
1:25

๐ŸŽพYour humanoid tennis player is here!๐Ÿค– Introducing LATENT (Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data) โ€” the worldโ€™s first real-time whole-body planning and control algorithm for athletic humanoid tennis. For the first time, a humanoid robot can sustain high-dynamic, long-horizon tennis rallies with millisecond-level reactions, precise ball striking, and natural whole-body motion. This marks a leap from mechanical motion imitation to intelligent, decision-driven athletic interaction. #Galbot #Robotics #Innovation #AI #Technology #Humanoid

Galbot

45,902 views โ€ข 5 months ago

๐Ÿš€ Galbot just released AstraBrain-WBC 0.5, the worldโ€™s first humanoid robot cerebellum foundation model. Trained on 20,000 hours of human motion data, it validates a Scaling Law for robot motion control and demonstrates zero-shot execution of unseen motions on real humanoids. The era of robot control foundation models is getting closer.
3:05

๐Ÿš€ Galbot just released AstraBrain-WBC 0.5, the worldโ€™s first humanoid robot cerebellum foundation model. Trained on 20,000 hours of human motion data, it validates a Scaling Law for robot motion control and demonstrates zero-shot execution of unseen motions on real humanoids. The era of robot control foundation models is getting closer.

Eren Chen

10,270 views โ€ข 2 months ago

NVIDIA has open-sourced SONIC, a humanoid behavior foundation model that gives robots a core set of motor skills learned from large-scale human motion data.
0:48

NVIDIA has open-sourced SONIC, a humanoid behavior foundation model that gives robots a core set of motor skills learned from large-scale human motion data.

The Humanoid Hub

33,542 views โ€ข 6 months ago

Motion at Will, Power in Balance: The DR02 humanoid robot from DEEP Robotics
0:53

Motion at Will, Power in Balance: The DR02 humanoid robot from DEEP Robotics

DEEP Robotics

2,128,127 views โ€ข 8 months ago

A step forward to be more "human." Chinese tech company Galbot unveiled its humanoid tennis robot that has leaped from mechanical motion imitation to real-time decision-making.
1:01

A step forward to be more "human." Chinese tech company Galbot unveiled its humanoid tennis robot that has leaped from mechanical motion imitation to real-time decision-making.

China Xinhua News

17,479 views โ€ข 5 months ago

Researchers just taught a robot to play tennis. From just clips of a few amateur players performing basic forehands, backhands, and shuffles... ...a robot learned one of the fastest, most coordinated physical skills there is. Insane!
1:09

Researchers just taught a robot to play tennis. From just clips of a few amateur players performing basic forehands, backhands, and shuffles... ...a robot learned one of the fastest, most coordinated physical skills there is. Insane!

Rowan Cheung

39,376 views โ€ข 4 months ago

๐Ÿค– INTERESTING: Chinaโ€™s robot dogs are now dropping kids off at school. Real or AI?
0:34

๐Ÿค– INTERESTING: Chinaโ€™s robot dogs are now dropping kids off at school. Real or AI?

Cointelegraph

39,209 views โ€ข 3 months ago

โšก๏ธ NEW: The Unitree G1 humanoid robot showcased advanced walking, balance, and movement capabilities powered by AI-driven real-time motion control.
0:27

โšก๏ธ NEW: The Unitree G1 humanoid robot showcased advanced walking, balance, and movement capabilities powered by AI-driven real-time motion control.

Cointelegraph

49,997 views โ€ข 2 months ago

๐Ÿค– INTERESTING: Torobo is demonstrating real-time human motion imitation, bringing humanoid robots one step closer to natural human-like movement.
0:23

๐Ÿค– INTERESTING: Torobo is demonstrating real-time human motion imitation, bringing humanoid robots one step closer to natural human-like movement.

Cointelegraph

21,794 views โ€ข 2 months ago

๐Ÿค– LATEST: A "prettier-than-human" bionic humanoid robot by Noetix Robotics sold for $15,400 on JD Auctions after starting at just $0.14 and drawing nearly 40,000 viewers.
0:48

๐Ÿค– LATEST: A "prettier-than-human" bionic humanoid robot by Noetix Robotics sold for $15,400 on JD Auctions after starting at just $0.14 and drawing nearly 40,000 viewers.

Cointelegraph

41,430 views โ€ข 2 months ago

How can we build a general-purpose โ€œfoundation modelโ€ for robot motion? Zhengyi โ€œZenโ€ Luo joins us to talk about SONIC, which uses motion tracking as a foundational task for humanoid robot control, and scales humanoid control training to 9k GPU hours and 100 million frames worth of data. The result: a model with a generally-useful embedding space that can be controlled by a VLA, or from human video, to perform a wide variety of humanoid whole-body-control tasks, including with zero-shot transfer to previously unseen motions. Watch episode 72 of RoboPapers, with Michael Cho - Rbt/Acc and Jiafei Duan, now!
59:44

How can we build a general-purpose โ€œfoundation modelโ€ for robot motion? Zhengyi โ€œZenโ€ Luo joins us to talk about SONIC, which uses motion tracking as a foundational task for humanoid robot control, and scales humanoid control training to 9k GPU hours and 100 million frames worth of data. The result: a model with a generally-useful embedding space that can be controlled by a VLA, or from human video, to perform a wide variety of humanoid whole-body-control tasks, including with zero-shot transfer to previously unseen motions. Watch episode 72 of RoboPapers, with Michael Cho - Rbt/Acc and Jiafei Duan, now!

RoboPapers

24,802 views โ€ข 4 months ago

The problem with humanoid teleoperation is that it is expensive and difficult to scale Enter NVIDIA's EgoScale: - A VLA model pretrained on thousands hours of egocentric human videos. - Mid-trained via 50 hours of human + 4 hours of robot "play" data for human-robot alignment. - Fine-tuned with very few examples of task-specific robot teleoperation (100 or fewer per task). - Successfully transfers across 5-finger (Sharpa) and 3-finger (Unitree G1) robot hands. - Performance scales predictably as data increases.
0:50

The problem with humanoid teleoperation is that it is expensive and difficult to scale Enter NVIDIA's EgoScale: - A VLA model pretrained on thousands hours of egocentric human videos. - Mid-trained via 50 hours of human + 4 hours of robot "play" data for human-robot alignment. - Fine-tuned with very few examples of task-specific robot teleoperation (100 or fewer per task). - Successfully transfers across 5-finger (Sharpa) and 3-finger (Unitree G1) robot hands. - Performance scales predictably as data increases.

The Humanoid Hub

44,441 views โ€ข 5 months ago

Researchers developed a system called LATENT that teaches humanoid robots to play tennis using small pieces of human movement data. It learns basic motion fragments, then corrects and combines them to create natural movements and swings. The system was tested on the Unitree G1 humanoid robot, enabling it to hit balls and rally with human players.
0:22

Researchers developed a system called LATENT that teaches humanoid robots to play tennis using small pieces of human movement data. It learns basic motion fragments, then corrects and combines them to create natural movements and swings. The system was tested on the Unitree G1 humanoid robot, enabling it to hit balls and rally with human players.

Space and Technology

45,748 views โ€ข 3 months ago

We trained a humanoid with 22-DoF dexterous hands to assemble model cars, operate syringes, sort poker cards, fold/roll shirts, all learned primarily from 20,000+ hours of egocentric human video with no robot in the loop. Humans are the most scalable embodiment on the planet. We discovered a near-perfect log-linear scaling law (Rยฒ = 0.998) between human video volume and action prediction loss, and this loss directly predicts real-robot success rate. Humanoid robots will be the end game, because they are the practical form factor with minimal embodiment gap from humans. Call it the Bitter Lesson of robot hardware: the kinematic similarity lets us simply retarget human finger motion onto dexterous robot hand joints. No learned embeddings, no fancy transfer algorithms needed. Relative wrist motion + retargeted 22-DoF finger actions serve as a unified action space that carries through from pre-training to robot execution. Our recipe is called "EgoScale": - Pre-train GR00T N1.5 on 20K hours of human video, mid-train with only 4 hours (!) of robot play data with Sharpa hands. 54% gains over training from scratch across 5 highly dexterous tasks. - Most surprising result: a *single* teleop demo is sufficient to learn a never-before-seen task. Our recipe enables extreme data efficiency. - Although we pre-train in 22-DoF hand joint space, the policy transfers to a Unitree G1 with 7-DoF tri-finger hands. 30%+ gains over training on G1 data alone. The scalable path to robot dexterity was never more robots. It was always us. Deep dives in thread:
1:17

We trained a humanoid with 22-DoF dexterous hands to assemble model cars, operate syringes, sort poker cards, fold/roll shirts, all learned primarily from 20,000+ hours of egocentric human video with no robot in the loop. Humans are the most scalable embodiment on the planet. We discovered a near-perfect log-linear scaling law (Rยฒ = 0.998) between human video volume and action prediction loss, and this loss directly predicts real-robot success rate. Humanoid robots will be the end game, because they are the practical form factor with minimal embodiment gap from humans. Call it the Bitter Lesson of robot hardware: the kinematic similarity lets us simply retarget human finger motion onto dexterous robot hand joints. No learned embeddings, no fancy transfer algorithms needed. Relative wrist motion + retargeted 22-DoF finger actions serve as a unified action space that carries through from pre-training to robot execution. Our recipe is called "EgoScale": - Pre-train GR00T N1.5 on 20K hours of human video, mid-train with only 4 hours (!) of robot play data with Sharpa hands. 54% gains over training from scratch across 5 highly dexterous tasks. - Most surprising result: a *single* teleop demo is sufficient to learn a never-before-seen task. Our recipe enables extreme data efficiency. - Although we pre-train in 22-DoF hand joint space, the policy transfers to a Unitree G1 with 7-DoF tri-finger hands. 30%+ gains over training on G1 data alone. The scalable path to robot dexterity was never more robots. It was always us. Deep dives in thread:

Jim Fan

298,223 views โ€ข 5 months ago

Georgia Tech's research uses Metaโ€™s Project Aria glasses to train humanoid robots with egocentric data. PhD student Simar Kareer developed an algorithm that leverages human data to enhance robot learning, achieving a 400% performance boost with just 90 minutes of recordings.
3:00

Georgia Tech's research uses Metaโ€™s Project Aria glasses to train humanoid robots with egocentric data. PhD student Simar Kareer developed an algorithm that leverages human data to enhance robot learning, achieving a 400% performance boost with just 90 minutes of recordings.

The Humanoid Hub

65,618 views โ€ข 1 year ago

Chinese humanoid robot is selling popcorn at the cinema. It can work 14 hours a day.
0:35

Chinese humanoid robot is selling popcorn at the cinema. It can work 14 hours a day.

Li Zexin ๆŽๆณฝๆฌฃ

23,627 views โ€ข 7 months ago

TWIST: a real-time teleoperation system for humanoid robots to mimic whole-body motions. Reference motion data is generated by retargeting human motion-capture data to the robot. Then, the controller is trained in simulation using reinforcement learning and behavior cloning.
0:58

TWIST: a real-time teleoperation system for humanoid robots to mimic whole-body motions. Reference motion data is generated by retargeting human motion-capture data to the robot. Then, the controller is trained in simulation using reinforcement learning and behavior cloning.

The Humanoid Hub

46,715 views โ€ข 1 year ago

asked claude fable 5 to design a humanoid robot 2 hours and 1.4m tokens later, i got this absolutely insane
0:42

asked claude fable 5 to design a humanoid robot 2 hours and 1.4m tokens later, i got this absolutely insane

Jake Fitzgerald

906,118 views โ€ข 2 months ago

Perhaps the "GPT-3 moment" for humanoid robots lies within vast amounts of human video data. UC Berkeley (Abbeel + Malik group) just released Do as I Do: a pipeline that takes monocular RGB human videos (ego + exo)->reconstructs hand-object interactions->retargets them to multi-fingered robotic hands, generating executable robot trajectories. It pushes success rates on real-world clips from ~25% to 71%, outperforming prior SOTA. This kind of human video->robot-complete data approach feels increasingly important for closing the data flywheel in humanoid/embodied AI.
0:29

Perhaps the "GPT-3 moment" for humanoid robots lies within vast amounts of human video data. UC Berkeley (Abbeel + Malik group) just released Do as I Do: a pipeline that takes monocular RGB human videos (ego + exo)->reconstructs hand-object interactions->retargets them to multi-fingered robotic hands, generating executable robot trajectories. It pushes success rates on real-world clips from ~25% to 71%, outperforming prior SOTA. This kind of human video->robot-complete data approach feels increasingly important for closing the data flywheel in humanoid/embodied AI.

CyberRobo

12,107 views โ€ข 2 months ago