正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

A humanoid robot policy trained solely on synthetic data generated by a world model. Research Scientist Joel Jang presents NVIDIA's DreamGen pipeline: ⦿ Post-train the world model Cosmos-Predict2 with a small set of real teleoperation demos. ⦿ Prompt the world model to generate synthetic video data with verbs and... show more

The Humanoid Hub

107,920 subscribers

20,968 次观看 • 11 个月前 •via X (Twitter)

科学技术教育艺术

Anya Rossi• Live Now

Private livecam show

9 条评论

𝕳𝖔𝖑𝖑𝖞 的头像

𝕳𝖔𝖑𝖑𝖞11 个月前

Mhm

Premium 的头像

Premium11 个月前

Go ad-free on X with Premium+ It's the highest return on investment you can make.

Fabien Musty 的头像

Fabien Musty11 个月前

Companies like 1x that develop their own models will cumulate benefits from in-house efforts plus everything Nvidia comes up with.

Sudodudo 的头像

Sudodudo11 个月前

Synthetic sim data training is good but it will always fail in real world due to randomness of world will overwhelm them. This deterministic approach will fail in real world , same reason tesla ain,t confident enough to create real self driving cars yet despite so much training

Alex Prompter 的头像

Alex Prompter11 个月前

the approach sounds interesting, especially using real data for fine-tuning. balancing synthetic with real-world scenarios is tricky though. curious to see how it performs in practice.

C Zhang 的头像

C Zhang11 个月前

And it directly worked on the real robot. The sim2real pipeline was just magically good at that time. Unfortunately it took us too much time to make it a publication.

Vector Wang 的头像

Vector Wang11 个月前

XLeRobot 0.2.3 (XLeVR) out! • VR (Quest 3) Whole-body Control • Reads everything: head and hand poses, joysticks, and all the buttons • Minimal dependencies, web-based • Modular, can use it on other robots • Open-source, easy install, play in 20min

Ilir Aliu - eu/acc 的头像

Ilir Aliu - eu/acc11 个月前

What if your robot could catch things mid-air... without breaking or snapping a joint? [📍Bookmark for later] IMA-Catcher is a new framework that lets robots catch nonprehensile, free-falling objects while reducing impact forces and avoiding hardware damage. Here’s what they did: ✅ Matched robot velocity to the object’s for low-impact catching ✅ Used human demos to teach smooth post-catch movements ✅ Minimized joint stress with a smart controller ✅ Generalized from 1D to full 3D catching Without velocity matching? The robot would fail the task completely. Catch the full paper here: 📍 ------------------------------ 🧠 Repost to help someone find the paper they’ve been searching for.

Junzhe (JJ) He 的头像

Junzhe (JJ) He11 个月前

Exactly 2 years ago during #Arche, my thesis supervisor @FJenelten51897 and I deployed this mpc+rl hybrid controller on very complicated collapsed buildings. Two years later its performance is still at the top tier! 😎 Paper:

相关视频

NEWS: Nvidia has just introduced "Cosmos," a world foundation model created to understand the physical world. The model can generate synthetic data to train robotics.

NEWS: Nvidia has just introduced "Cosmos," a world foundation model created to understand the physical world. The model can generate synthetic data to train robotics.

Sawyer Merritt

254,296 次观看 • 1 年前

An interactive world model developed by NVIDIA in collaboration with academic partners. - DreamDojo turns egocentric human video data into physical intelligence. - Human data is more scalable than robotics data but lacks action labels. - To solve this, a dedicated action model extracts latent actions by identifying physics and motion deltas between frames. Training - A massive 44k hours of video data are used for pre-training. - Post-training on small-scale robot datasets maps human physics to specific robot embodiments. - An additional distillation stage converts the model into an autoregressive, few-step diffusion model, enabling real-time, action-controllable simulation. Primary Use Cases - Live Teleoperation: Controlling a robot inside a world simulation in real-time. - Model-based Planning: Previewing and curating the best actions for improved success. - Policy Evaluation: Testing robot policies in realistic, out-of-distribution scenarios. Everything that's open-sourced: weights, code, post-training dataset, eval set, and details to reproduce.

An interactive world model developed by NVIDIA in collaboration with academic partners. - DreamDojo turns egocentric human video data into physical intelligence. - Human data is more scalable than robotics data but lacks action labels. - To solve this, a dedicated action model extracts latent actions by identifying physics and motion deltas between frames. Training - A massive 44k hours of video data are used for pre-training. - Post-training on small-scale robot datasets maps human physics to specific robot embodiments. - An additional distillation stage converts the model into an autoregressive, few-step diffusion model, enabling real-time, action-controllable simulation. Primary Use Cases - Live Teleoperation: Controlling a robot inside a world simulation in real-time. - Model-based Planning: Previewing and curating the best actions for improved success. - Policy Evaluation: Testing robot policies in realistic, out-of-distribution scenarios. Everything that's open-sourced: weights, code, post-training dataset, eval set, and details to reproduce.

The Humanoid Hub

11,575 次观看 • 4 个月前

Bernt Børnich says teleoperation data and synthetic data are like a crutch to bootstrap your way to robots learning in the real world.

Bernt Børnich says teleoperation data and synthetic data are like a crutch to bootstrap your way to robots learning in the real world.

The Humanoid Hub

27,977 次观看 • 1 年前

No data, no problem introducing agentic synthetic data generation with Cosmos 3 share a few examples, generate more data, automate model training, automatically deploy the latest version with no downtime in a benchmark run with Corning Incorporated's optical fiber manufacturing engineering team, a model trained on 8 real defect images plus synthetic examples generated by Cosmos reached 0.95 mean average precision and perfect recall on the toughest defect class, beating a baseline trained on real data alone. "The Roboflow Agent powered by NVIDIA allows us to generate the training data we need, fine-tune our models, and strengthen model performance and inspection quality while increasing the speed, scalability, and adoption of next-generation technologies,” - Jeremy Knopf, chief information officer, Corning Optical Communications

No data, no problem introducing agentic synthetic data generation with Cosmos 3 share a few examples, generate more data, automate model training, automatically deploy the latest version with no downtime in a benchmark run with Corning Incorporated's optical fiber manufacturing engineering team, a model trained on 8 real defect images plus synthetic examples generated by Cosmos reached 0.95 mean average precision and perfect recall on the toughest defect class, beating a baseline trained on real data alone. "The Roboflow Agent powered by NVIDIA allows us to generate the training data we need, fine-tune our models, and strengthen model performance and inspection quality while increasing the speed, scalability, and adoption of next-generation technologies,” - Jeremy Knopf, chief information officer, Corning Optical Communications

Roboflow

38,811 次观看 • 23 天前

NVIDIA has published a paper on DREAMGEN – a powerful 4-step pipeline for generating synthetic data for humanoids that enables task and environment generalization. - Step 1: Fine-tune a video generation model using a small number of human teleoperation videos - Step 2: Prompt the fine-tuned model to turn a single real image into new AI-imagined videos - Step 3: Automatically label actions in the generated videos - Step 4: Train a robot AI model with the labeled synthetic dataset This enabled humanoid robots to perform 22 novel behaviors – such as pouring, opening/closing articulated objects, and manipulating a variety of tools. The original teleoperation dataset only included pick-and-place tasks. This takes task extensibility to another level without requiring human teleoperation for every single task. The pipeline will be made open-source soon. Project page:

NVIDIA has published a paper on DREAMGEN – a powerful 4-step pipeline for generating synthetic data for humanoids that enables task and environment generalization. - Step 1: Fine-tune a video generation model using a small number of human teleoperation videos - Step 2: Prompt the fine-tuned model to turn a single real image into new AI-imagined videos - Step 3: Automatically label actions in the generated videos - Step 4: Train a robot AI model with the labeled synthetic dataset This enabled humanoid robots to perform 22 novel behaviors – such as pouring, opening/closing articulated objects, and manipulating a variety of tools. The original teleoperation dataset only included pick-and-place tasks. This takes task extensibility to another level without requiring human teleoperation for every single task. The pipeline will be made open-source soon. Project page:

The Humanoid Hub

12,074 次观看 • 1 年前

NVIDIA just introduced Cosmos, a platform for world foundation models designed for robotics. ⦿ It features advanced tokenizers, an AI-accelerated data pipeline, and integration with NVIDIA Omniverse. Humanoid makers 1X, Figure, and Agility are among the first to adopt Cosmos. ⦿ Cosmos generates synthetic, physics-based data, accelerating model training and customization. ⦿ It also features a CUDA-accelerated data processing pipeline that enables developers to process, curate, and label 20 million hours of videos in 14 days using the NVIDIA Blackwell platform.

NVIDIA just introduced Cosmos, a platform for world foundation models designed for robotics. ⦿ It features advanced tokenizers, an AI-accelerated data pipeline, and integration with NVIDIA Omniverse. Humanoid makers 1X, Figure, and Agility are among the first to adopt Cosmos. ⦿ Cosmos generates synthetic, physics-based data, accelerating model training and customization. ⦿ It also features a CUDA-accelerated data processing pipeline that enables developers to process, curate, and label 20 million hours of videos in 14 days using the NVIDIA Blackwell platform.

The Humanoid Hub

129,383 次观看 • 1 年前

CMU researchers, in collaboration with NVIDIA, present ASAP, a two-stage framework for humanoid robot agility. It pre-trains motion policies on human data, then refines them with real-world corrections using a delta action model, which adjusts for simulation mismatches.

CMU researchers, in collaboration with NVIDIA, present ASAP, a two-stage framework for humanoid robot agility. It pre-trains motion policies on human data, then refines them with real-world corrections using a delta action model, which adjusts for simulation mismatches.

The Humanoid Hub

1,826,529 次观看 • 1 年前

CMU researchers, in collaboration with NVIDIA, present ASAP, a two-stage framework for humanoid robot agility. It pre-trains motion policies on human data, refines them with real-world corrections using a delta action model, which adjusts for simulation mismatches. Early days.

CMU researchers, in collaboration with NVIDIA, present ASAP, a two-stage framework for humanoid robot agility. It pre-trains motion policies on human data, refines them with real-world corrections using a delta action model, which adjusts for simulation mismatches. Early days.

Brian Roemmele

87,593 次观看 • 1 年前

better real-world physics are coming to video models thanks to synthetic video data

better real-world physics are coming to video models thanks to synthetic video data

Dreaming Tulpa 🥓👑

16,135 次观看 • 1 年前

Turning video into humanoid robot motion! 🤳🏼 Training humanoid robots needs huge amounts of motion data, but real-world capture doesn’t scale. Mocap is expensive, dangerous edge cases are rare, and you can’t ask humans to repeatedly fall or crash. Video2Robot tackles this by converting videos into physics-grounded humanoid simulations. Motion is generated to respect balance, inertia, ground contact, and joint limits, then directly retargeted to robot simulators. One prompt can generate a full humanoid motion sequence, including multi-agent interactions and failure cases like falls or collisions, scenarios that are hard or impossible to capture safely in the real world. The pipeline is model-agnostic and works with existing video generators, making it a practical way to scale data for robots. If robots are going to operate in the real world, they need to be trained on the failures too, not just the perfect demos. Here's the GitHub: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Turning video into humanoid robot motion! 🤳🏼 Training humanoid robots needs huge amounts of motion data, but real-world capture doesn’t scale. Mocap is expensive, dangerous edge cases are rare, and you can’t ask humans to repeatedly fall or crash. Video2Robot tackles this by converting videos into physics-grounded humanoid simulations. Motion is generated to respect balance, inertia, ground contact, and joint limits, then directly retargeted to robot simulators. One prompt can generate a full humanoid motion sequence, including multi-agent interactions and failure cases like falls or collisions, scenarios that are hard or impossible to capture safely in the real world. The pipeline is model-agnostic and works with existing video generators, making it a practical way to scale data for robots. If robots are going to operate in the real world, they need to be trained on the failures too, not just the perfect demos. Here's the GitHub: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

59,645 次观看 • 6 个月前

3rd Place: GrokWorld uses Grok Imagine as a world model to generate synthetic training data for robots — augmenting or replacing months of manual collection in hours. Aptura AI

3rd Place: GrokWorld uses Grok Imagine as a world model to generate synthetic training data for robots — augmenting or replacing months of manual collection in hours. Aptura AI

xAI

846,005 次观看 • 5 个月前

Helix is now learning directly from human video data We have already trained on data collected in the real world, including Brookfield residential units To our knowledge, this is the first instance of a humanoid robot learning navigation end-to-end using only human video

Helix is now learning directly from human video data We have already trained on data collected in the real world, including Brookfield residential units To our knowledge, this is the first instance of a humanoid robot learning navigation end-to-end using only human video

Figure

46,041 次观看 • 9 个月前

Real-world robot data is expensive and slow to collect, creating a major challenge for humanoid development. 🤖 The NVIDIA GR00T N1.6 open vision language action model is pre-trained on a diverse mix of data, including thousands of hours of Stanford Vision and Learning Lab’s BEHAVIOR simulation data, which covers long-horizon everyday manipulation tasks. This diverse training is the key to robust cross-embodiment performance and real-world adaptability. 🌍 Read the blog 🔗

Real-world robot data is expensive and slow to collect, creating a major challenge for humanoid development. 🤖 The NVIDIA GR00T N1.6 open vision language action model is pre-trained on a diverse mix of data, including thousands of hours of Stanford Vision and Learning Lab’s BEHAVIOR simulation data, which covers long-horizon everyday manipulation tasks. This diverse training is the key to robust cross-embodiment performance and real-world adaptability. 🌍 Read the blog 🔗

NVIDIA Robotics

13,421 次观看 • 4 个月前

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.

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 次观看 • 1 年前

Here's some footage from my Tesla's dashcam. Just kidding. Did you believe me? The truth is, this video is completely fake. It was generated by Tesla's world simulator, a generative AI model that not only allows Tesla to generate realistic synthetic training data, but also validate new versions of their self-driving model.

Here's some footage from my Tesla's dashcam. Just kidding. Did you believe me? The truth is, this video is completely fake. It was generated by Tesla's world simulator, a generative AI model that not only allows Tesla to generate realistic synthetic training data, but also validate new versions of their self-driving model.

Whole Mars Catalog

226,004 次观看 • 8 个月前

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

42,804 次观看 • 4 个月前

The World Model as NEO's Cognitive Core 1X has revealed a major AI development where the NEO humanoid can translate any natural language prompt into robotic action. It demonstrates this capability even for novel tasks, objects, and environments not found in its robot dataset. - the 1X World Model is trained on internet-scale human interaction videos and fine-tuned with robot data to ground its understanding in physics and in NEO's embodiment - from a simple voice or text prompt, the world model generates a visualization of future actions - a built-in inverse dynamics model then translates these into precise motor movements for NEO

The World Model as NEO's Cognitive Core 1X has revealed a major AI development where the NEO humanoid can translate any natural language prompt into robotic action. It demonstrates this capability even for novel tasks, objects, and environments not found in its robot dataset. - the 1X World Model is trained on internet-scale human interaction videos and fine-tuned with robot data to ground its understanding in physics and in NEO's embodiment - from a simple voice or text prompt, the world model generates a visualization of future actions - a built-in inverse dynamics model then translates these into precise motor movements for NEO

The Humanoid Hub

68,453 次观看 • 5 个月前

🚨 Jensen Huang says everyone panicked about the AI data when MOST training data was never REAL to begin with. Ilya Sutskever told the industry pre-training was over. "Ilya said, 'We're out of data,' or something like that. 'Pre-training is over,' or something like that," Huang says. "The industry panicked, you know, that this is the end of AI." "And of course, of course that's obviously not true. We're gonna keep on scaling the amount of data that we have to train with." "A lot of that data is probably gonna be synthetic." That's where the panic came from — synthetic data sounds like cheating. "Most of the data that we are training, that we teach each other with, inform each other with, is synthetic." "It's synthetic because it didn't come out of nature." "You created it. I'm consuming it. I modify it, augment it, I regenerate it, somebody else consumes it." The textbook in your hand is synthetic. The post you're reading is synthetic. The lecture you took is synthetic. Nature didn't make any of it. Humans did. AI just learned to do the same thing — faster. "Training is now limited by compute," Huang says. "Data is now limited by compute." The data wall wasn't a wall. It was a mirror. If you're new here, follow @AiEvolutio for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

🚨 Jensen Huang says everyone panicked about the AI data when MOST training data was never REAL to begin with. Ilya Sutskever told the industry pre-training was over. "Ilya said, 'We're out of data,' or something like that. 'Pre-training is over,' or something like that," Huang says. "The industry panicked, you know, that this is the end of AI." "And of course, of course that's obviously not true. We're gonna keep on scaling the amount of data that we have to train with." "A lot of that data is probably gonna be synthetic." That's where the panic came from — synthetic data sounds like cheating. "Most of the data that we are training, that we teach each other with, inform each other with, is synthetic." "It's synthetic because it didn't come out of nature." "You created it. I'm consuming it. I modify it, augment it, I regenerate it, somebody else consumes it." The textbook in your hand is synthetic. The post you're reading is synthetic. The lecture you took is synthetic. Nature didn't make any of it. Humans did. AI just learned to do the same thing — faster. "Training is now limited by compute," Huang says. "Data is now limited by compute." The data wall wasn't a wall. It was a mirror. If you're new here, follow @AiEvolutio for the latest on ChatGPT, Claude, and the AI tools shaping how we work and create. — Jensen Huang ( NVIDIA ), NVIDIA CEO, on Lex Fridman's ( Lex Fridman ) podcast

AI Evolution

15,565 次观看 • 24 天前

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 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 次观看 • 3 个月前