Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Eric Jang weighs in on RL vs. supervised learning for humanoid robot manipulation tasks.

The Humanoid Hub

115,812 subscribers

11,835 views • 1 year ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

6 Comments

The Humanoid Hub1 year ago

Full interview with @ericjang11, 1X’s VP of AI, on how his team is building scalable learning systems for home-use humanoids.

Yev Marusenko1 year ago

Eric thinking 10 years ahead

Ant A. 🇺🇸1 year ago

Perhaps the idea is to have the learning humanoids RL in real-life with human partners for training in every kind of situation.

Ether Whales1 year ago

Finally a robot that do the laundry and dish 😁

Aakash1 year ago

Great Humanoid

Himanshu Kumar1 year ago

Intriguing take.. Practical robotics often favors hybrid approaches for sim2real robustness..

Related Videos

Sim2Real RL for Vision-Based Dexterous Manipulation on Humanoids TLDR - we train a humanoid robot with two multifingered hands to perform a range of dexterous manipulation tasks robust generalization and high performance without human demonstration :D

Sim2Real RL for Vision-Based Dexterous Manipulation on Humanoids TLDR - we train a humanoid robot with two multifingered hands to perform a range of dexterous manipulation tasks robust generalization and high performance without human demonstration :D

Toru

49,892 views • 1 year ago

Introducing HiLMa-Res: a hierarchical RL framework for quadrupeds to tackle loco-manipulation tasks with sustained mobility! Designed for general learning tasks (vision-based, state-based, real-world data, etc), the robot now can step over stones🐾/navigate boxes📦/dribble⚽.

Introducing HiLMa-Res: a hierarchical RL framework for quadrupeds to tackle loco-manipulation tasks with sustained mobility! Designed for general learning tasks (vision-based, state-based, real-world data, etc), the robot now can step over stones🐾/navigate boxes📦/dribble⚽.

Zhongyu Li

11,720 views • 1 year ago

Excited to share a new humanoid robot platform we’ve been working on. Berkeley Humanoid is a reliable and low-cost mid-scale research platform for learning-based control. We demonstrate the robot walking on various terrains and dynamic hopping with a simple RL controller.

Excited to share a new humanoid robot platform we’ve been working on. Berkeley Humanoid is a reliable and low-cost mid-scale research platform for learning-based control. We demonstrate the robot walking on various terrains and dynamic hopping with a simple RL controller.

Qiayuan Liao

80,564 views • 2 years ago

Introducing RL-100: Performant Robotic Manipulation with Real-World Reinforcement Learning. 7 real robot tasks, 900/900 successes. Up to 250 consecutive trials in one task, running 2 hours nonstop without failure. High success rate against physical disturbances, zero-shot, and few-shot adaptation Our first step toward a deployable robot learning system.

Introducing RL-100: Performant Robotic Manipulation with Real-World Reinforcement Learning. 7 real robot tasks, 900/900 successes. Up to 250 consecutive trials in one task, running 2 hours nonstop without failure. High success rate against physical disturbances, zero-shot, and few-shot adaptation Our first step toward a deployable robot learning system.

Kun Lei

92,064 views • 11 months ago

🤖 How can we scale up humanoid robot learning? Introducing 🌟VLK🌟: generating large-scale synthetic data with paired egocentric observations, text, and full-body G1 kinematics for learning humanoid loco-manipulation. No teleoperation needed! Website:

🤖 How can we scale up humanoid robot learning? Introducing 🌟VLK🌟: generating large-scale synthetic data with paired egocentric observations, text, and full-body G1 kinematics for learning humanoid loco-manipulation. No teleoperation needed! Website:

Jiaman Li

974,350 views • 2 months ago

Wobbly humanoid robot is learning to use hands for balance

Wobbly humanoid robot is learning to use hands for balance

Mashable

16,374 views • 3 years ago

Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learning investigated the application of Deep Reinforcement Learning (Deep RL) for low-cost, miniature humanoid hardware in a dynamic environment, showing the method can synthesize sophisticated and safe movement skills making up complex behavioral strategies in a simplified one-versus-one (1v1) soccer game abs: project page:

Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learning investigated the application of Deep Reinforcement Learning (Deep RL) for low-cost, miniature humanoid hardware in a dynamic environment, showing the method can synthesize sophisticated and safe movement skills making up complex behavioral strategies in a simplified one-versus-one (1v1) soccer game abs: project page:

AK

293,474 views • 3 years ago

Engineers at 1X have developed a 25-degree-of-freedom humanoid robot hand for advanced dexterous manipulation. It uses force sensors in every joint and fingertips to detect pressure, slipping, and automatically adjust its grip. Its lightweight, waterproof design allows the NEO humanoid robot to safely wash its hands after cooking or other messy tasks.

Engineers at 1X have developed a 25-degree-of-freedom humanoid robot hand for advanced dexterous manipulation. It uses force sensors in every joint and fingertips to detect pressure, slipping, and automatically adjust its grip. Its lightweight, waterproof design allows the NEO humanoid robot to safely wash its hands after cooking or other messy tasks.

Space and Technology

26,780 views • 2 months ago

The humanoid robot 100-meter group race. First place went to Shandong Yobotics' Taishan Walker. This humanoid robot is primarily used for specialized tasks, such as high-risk inspections and rescue operations. That's why it was the fastest.（World Humanoid Robot Games2025）

The humanoid robot 100-meter group race. First place went to Shandong Yobotics' Taishan Walker. This humanoid robot is primarily used for specialized tasks, such as high-risk inspections and rescue operations. That's why it was the fastest.（World Humanoid Robot Games2025）

CyberRobo

94,611 views • 1 year ago

Not just boxing… humanoid robots are now learning Chinese Kung Fu! Imagine future robot fights: Muay Thai vs. Brazilian Jiu-Jitsu vs. Boxing vs. Chinese Kung Fu Robot UFC is coming. Who’s your pick?

Not just boxing… humanoid robots are now learning Chinese Kung Fu! Imagine future robot fights: Muay Thai vs. Brazilian Jiu-Jitsu vs. Boxing vs. Chinese Kung Fu Robot UFC is coming. Who’s your pick?

Reborn

15,878 views • 1 year ago

Introducing TRLC-DK1-X: An accessible humanoid data collection platform. It features two 6+1 DOF leader-follower arm pairs and three ultra-wide FOV cameras, integrated for learning static, bimanual manipulation tasks. Order now for $4,999.

Introducing TRLC-DK1-X: An accessible humanoid data collection platform. It features two 6+1 DOF leader-follower arm pairs and three ultra-wide FOV cameras, integrated for learning static, bimanual manipulation tasks. Order now for $4,999.

Jannik Grothusen

44,302 views • 10 months ago

More and more humanoid robot kiosks are appearing in China. Yesterday, I passed by a subway entrance in Guangzhou and experienced a retail kiosk operated by a humanoid robot again, buying a bottle of water. The process was the same: select items on a screen, scan a QR code for mobile payment, and the humanoid robot retrieves the product. (However, it didn't interact with me at all.) I looked it up, and the wheeled humanoid robot is named S1, from AutoLife Robotics in Shenzhen. It uses omnidirectional wheels for movement and a lifting body. It's 1-1.6m tall, weighs 72kg, and has a 10-hour battery life. A single arm can carry 3-5kg. Furthermore, this humanoid robot is suitable for lightweight tasks in retail, convenience stores, supermarkets, and warehouse sorting.

More and more humanoid robot kiosks are appearing in China. Yesterday, I passed by a subway entrance in Guangzhou and experienced a retail kiosk operated by a humanoid robot again, buying a bottle of water. The process was the same: select items on a screen, scan a QR code for mobile payment, and the humanoid robot retrieves the product. (However, it didn't interact with me at all.) I looked it up, and the wheeled humanoid robot is named S1, from AutoLife Robotics in Shenzhen. It uses omnidirectional wheels for movement and a lifting body. It's 1-1.6m tall, weighs 72kg, and has a 10-hour battery life. A single arm can carry 3-5kg. Furthermore, this humanoid robot is suitable for lightweight tasks in retail, convenience stores, supermarkets, and warehouse sorting.

CyberRobo

52,565 views • 6 months ago

🇩🇪 - Germany is BACK! Agile Robots launched Agile One, a Humanoid Robot that is designed for industry. It can handle many tasks autonomously. It learns quickly by doing the tasks, weighs 69kg, and is powered by their own AI foundation models! Europe will win 🇪🇺

🇩🇪 - Germany is BACK! Agile Robots launched Agile One, a Humanoid Robot that is designed for industry. It can handle many tasks autonomously. It learns quickly by doing the tasks, weighs 69kg, and is powered by their own AI foundation models! Europe will win 🇪🇺

NXT EU

28,619 views • 9 months ago

.Eric Jang is spearheading AI at 1X. Eric believes 1X is on the path to becoming an AGI company — not just a robotics company. I sat down with him at 1X’s new Palo Alto office to discuss the key AI building blocks needed to bring humanoid robots into homes at scale. 00:00 Eric Jang joins Humanoid Hub 00:27 Move to Palo Alto 01:02 Why 1X targets the home 02:01 The AGI company 02:49 Redwood AI overview 04:01 Unified control, modular stack 06:07 Data labeling at scale 07:10 RL mobility update for NEO 10:03 GPU and onboard compute 11:09 World model explained 13:34 Trusting world models for evaluation 15:05 Synthetic data generation 16:30 A world model Turing Test 17:41 RL in manipulation 20:08 Lessons from early NEO deployments 21:00 Full autonomy 22:18 1X’s unique edge in AI 24:02 Eric asks Devang a question

.Eric Jang is spearheading AI at 1X. Eric believes 1X is on the path to becoming an AGI company — not just a robotics company. I sat down with him at 1X’s new Palo Alto office to discuss the key AI building blocks needed to bring humanoid robots into homes at scale. 00:00 Eric Jang joins Humanoid Hub 00:27 Move to Palo Alto 01:02 Why 1X targets the home 02:01 The AGI company 02:49 Redwood AI overview 04:01 Unified control, modular stack 06:07 Data labeling at scale 07:10 RL mobility update for NEO 10:03 GPU and onboard compute 11:09 World model explained 13:34 Trusting world models for evaluation 15:05 Synthetic data generation 16:30 A world model Turing Test 17:41 RL in manipulation 20:08 Lessons from early NEO deployments 21:00 Full autonomy 22:18 1X’s unique edge in AI 24:02 Eric asks Devang a question

The Humanoid Hub

96,465 views • 1 year ago

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Max Fu

40,466 views • 2 years ago

The Robot Learning Company (The Robot Learning Company) makes automation accessible beyond factories by building affordable AI-powered robot arms to automate repetitive, stationary tasks. Congrats on the launch, Jannik Grothusen!

The Robot Learning Company (The Robot Learning Company) makes automation accessible beyond factories by building affordable AI-powered robot arms to automate repetitive, stationary tasks. Congrats on the launch, Jannik Grothusen!

Y Combinator

46,872 views • 1 year ago

Excited to introduce SOLE-R1, a video-language reasoning model for zero-shot reward prediction for robot manipulation tasks! SOLE-R1 reasoning can serve as the SOLE signal for learning new tasks (completely from scratch) through online RL - i.e., robots start with random actions and learn previously unseen tasks guided only by SOLE-R1 rewards, without any demonstrations, ground-truth rewards, success indicators, or task-specific tuning. SOLE-R1 significantly outperforms strong baselines (e.g., Robometer, RoboReward, TOPReward, GPT-5, Gemini-3-Pro) in zero-shot online RL when evaluated across 40 tasks - including a real-world tabletop manipulation setting and 4 sim environments (LIBERO, ManiSkill, Meta-World, RoboSuite). We open source all models, training data, and code. Website, demos, and paper at: 🧵 (1/6)

Excited to introduce SOLE-R1, a video-language reasoning model for zero-shot reward prediction for robot manipulation tasks! SOLE-R1 reasoning can serve as the SOLE signal for learning new tasks (completely from scratch) through online RL - i.e., robots start with random actions and learn previously unseen tasks guided only by SOLE-R1 rewards, without any demonstrations, ground-truth rewards, success indicators, or task-specific tuning. SOLE-R1 significantly outperforms strong baselines (e.g., Robometer, RoboReward, TOPReward, GPT-5, Gemini-3-Pro) in zero-shot online RL when evaluated across 40 tasks - including a real-world tabletop manipulation setting and 4 sim environments (LIBERO, ManiSkill, Meta-World, RoboSuite). We open source all models, training data, and code. Website, demos, and paper at: 🧵 (1/6)

Philip Schroeder

19,275 views • 3 months ago

🚨 Xiaomi has released a video of its deployment of humanoid robots in its electric vehicle factory. The humanoid robot performed some tasks repeatedly for 3 hours in the vehicle die casting part.

🚨 Xiaomi has released a video of its deployment of humanoid robots in its electric vehicle factory. The humanoid robot performed some tasks repeatedly for 3 hours in the vehicle die casting part.

Tsla Chan

17,290 views • 6 months ago