Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

New model: your robot can now pack your suitcase 🧳 Xiaomi has released a new robot foundation model. Called Xiaomi-Robotics-1, it is designed to have a robot pick things up and move them around. But first, DEFINITIONS: - Mixture-of-Transformers (MoT): An architecture where separate transformer "experts" (e.g., one for... vision-language, one for actions) share a single attention stream, so each modality gets specialized parameters without losing joint reasoning. - Vision-language model (VLM): A model that jointly understands images and text. - Diffusion transformer: A transformer trained to turn noise into structured outputs by iterative denoising, here generating robot actions rather than images. - Action chunks: Short sequences of future actions (e.g., the next ~50 motor commands) predicted in one shot instead of one step at a time. - Flow matching: A faster version of diffusion. The model learns a straight-line velocity field from noise to the target action, so it needs only a few integration steps instead of many denoising ones. Its peculiarity comes from its two stage training: 1. 100,000 hours of video shot through a UMI rig: a handheld 3D-printed gripper with a camera, worn by humans doing ordinary tasks in homes, shops, factories and offices. 2. Adapt to actual robot bodies with ~10,000 hours of real-robot data. It replaces the standard approach of teleoperating a real robot for every hour of training data. Its architecture is a Mixture-of-Transformers pairing a pre-trained Qwen3-VL vision-language model with a diffusion transformer that emits action chunks via flow matching, released in 2.6B, 5.1B and 10.5B parameter variants. However, if you read the entire paper ("Scaling VLA Models with over 100K Hours"), you realize that all of the scaling experiments on 20k hours. Therefore the headline "out-of-the-box success climbing 26% → 75% as pre-training data grows" tops out at 100% of 20k hours! What the full corpus does to that curve is never shown -> and this where things would become interesting! Xiaomi's own conclusion is that model size has stopped mattering and data is the binding constraint. The performance gap among different model sizes are less pronounced than those observed across different data scales. This result suggests that model capacity at the billions-parameter scale may already be sufficient to capture the current dataset's distribution. Which further asks the same question: why not use the 100k video hours? Anyway, I would definitely love to have a couple robots at home that can cooperate to pack my suitcase with items relevant to my next destination:show more

Léo

1,205 subscribers

14,771 Aufrufe • vor 3 Tagen •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

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 Aufrufe • vor 5 Monaten

Not the flashiest demos, but what’s under the hood represents a foundational shift for general-purpose robotics. World models are the next-gen foundation of Physical AI, not the VLM backbones found in typical VLAs. DreamZero is a 14B-parameter World Action Model (WAM) by NVIDIA that treats robotics as a joint video-and-action prediction task. Unlike traditional Vision-Language-Action (VLA) models that map images directly to motor commands, DreamZero leverages a pretrained video diffusion backbone to predict future world states and actions simultaneously. - achieves 2× better zero-shot generalization to unseen tasks and environments compared to state-of-the-art VLAs. - learns effectively from heterogeneous, non-repetitive data (500 hours), breaking the need for thousands of repeated demonstrations. - adapts to new robot embodiments with just 30 minutes of play data. - enables 7Hz closed-loop control via system optimizations and "DreamZero-Flash," making high-capacity diffusion models viable for real-time use.

Not the flashiest demos, but what’s under the hood represents a foundational shift for general-purpose robotics. World models are the next-gen foundation of Physical AI, not the VLM backbones found in typical VLAs. DreamZero is a 14B-parameter World Action Model (WAM) by NVIDIA that treats robotics as a joint video-and-action prediction task. Unlike traditional Vision-Language-Action (VLA) models that map images directly to motor commands, DreamZero leverages a pretrained video diffusion backbone to predict future world states and actions simultaneously. - achieves 2× better zero-shot generalization to unseen tasks and environments compared to state-of-the-art VLAs. - learns effectively from heterogeneous, non-repetitive data (500 hours), breaking the need for thousands of repeated demonstrations. - adapts to new robot embodiments with just 30 minutes of play data. - enables 7Hz closed-loop control via system optimizations and "DreamZero-Flash," making high-capacity diffusion models viable for real-time use.

The Humanoid Hub

35,204 Aufrufe • vor 6 Monaten

NEW RESEARCH: You can now create a new robot optimized for any given task! I love this new project by Huy Ha, Shuran Song, and others. Called "Transformer Transformer: A Unified Model for Motion-Conditioned Robot Co-design", it generates a robot's physical design and its controller together from a task spec. DEFINITIONS: - Reward function: A scoring rule that assigns a number to how well a behavior achieves the task. Here, it is the objective the generated design is pushed to maximize (e.g., track the target motion with low error). - Tokenizing: dividing continuous or structured data (a robot's links, joints, motor specs, states, actions) into a discrete vocabulary of symbols a transformer can process, the same step that turned pixels and audio into "language" for these models. - Diffusion transformer (DiT): A transformer trained to turn random noise into structured output through iterative denoising. Here, it generates robot bodies and trajectories instead of images. - MuJoCo: The standard fast physics simulator for robotics research (DeepMind-maintained). The Menagerie is its curated zoo of ready-to-use robot models. - CMA-ES: Covariance Matrix Adaptation Evolution Strategy, the workhorse black-box optimizer: it evolves a population of candidate designs, keeps the best, and needs thousands of simulator rollouts. - Bimanual multi-trajectory optimization: Finding one design/controller that performs well across several target motions for a two-armed robot at once, harder than optimizing for a single arm and a single motion. - BERT/MAE masked-modeling trick: Train one model to fill in whatever parts of the input you hide (words for BERT, image patches for MAE); at inference, choosing what to mask chooses the task, so masking the body makes it a designer and masking the actions makes it a controller. In practice, you give it a target end-effector motion and a reward function, and it outputs a complete embodiment (link, joint, motor, and inertial property), as well as a controller to drive it. It works by tokenizing both the body (links/joints/motors) and the dynamics (states/actions) into a compact scheme called RoboTokens, training a diffusion transformer (DiT) over them. The same model predicts dynamics using those predictions ("Dynamics Self-Guidance") to push generated designs toward higher reward at inference time. Masking different token types (using the BERT/MAE masked-modeling trick) lets the one model do three jobs: generate an embodiment, control an arbitrary embodiment, or design one conditioned on a motion. It is trained on 11 robots from the MuJoCo Menagerie (0.65 kg hand to 67.5 kg quadruped, 6–35 joints), and validated in sim and on a physical ALOHA doing cloth flinging. I like the fact that this approach inverts the entire recent robotics ideas: designing a policy for a fixed robot -> designing the robot for a fixed task. Every other approach assumes the body is given and learns a controller. Transformer Transformer takes the task (target motion + reward), then generates the body and controller jointly. In practice, it is a ~180× speedup over the standard optimizer at equal-or-better quality. It reaches "CMA-ES-level quality in seconds" and finishes bimanual multi-trajectory optimization in that is worth underlining nowadays! Also worth mentioning: this is the lab behind UMI and Handroid, that I mentioned here previously! The team seems extremely creative, i love these out-of-the-box approaches. Enjoy watching the demo of robot optimization in 3D, data acquisition, then real-life testing:

NEW RESEARCH: You can now create a new robot optimized for any given task! I love this new project by Huy Ha, Shuran Song, and others. Called "Transformer Transformer: A Unified Model for Motion-Conditioned Robot Co-design", it generates a robot's physical design and its controller together from a task spec. DEFINITIONS: - Reward function: A scoring rule that assigns a number to how well a behavior achieves the task. Here, it is the objective the generated design is pushed to maximize (e.g., track the target motion with low error). - Tokenizing: dividing continuous or structured data (a robot's links, joints, motor specs, states, actions) into a discrete vocabulary of symbols a transformer can process, the same step that turned pixels and audio into "language" for these models. - Diffusion transformer (DiT): A transformer trained to turn random noise into structured output through iterative denoising. Here, it generates robot bodies and trajectories instead of images. - MuJoCo: The standard fast physics simulator for robotics research (DeepMind-maintained). The Menagerie is its curated zoo of ready-to-use robot models. - CMA-ES: Covariance Matrix Adaptation Evolution Strategy, the workhorse black-box optimizer: it evolves a population of candidate designs, keeps the best, and needs thousands of simulator rollouts. - Bimanual multi-trajectory optimization: Finding one design/controller that performs well across several target motions for a two-armed robot at once, harder than optimizing for a single arm and a single motion. - BERT/MAE masked-modeling trick: Train one model to fill in whatever parts of the input you hide (words for BERT, image patches for MAE); at inference, choosing what to mask chooses the task, so masking the body makes it a designer and masking the actions makes it a controller. In practice, you give it a target end-effector motion and a reward function, and it outputs a complete embodiment (link, joint, motor, and inertial property), as well as a controller to drive it. It works by tokenizing both the body (links/joints/motors) and the dynamics (states/actions) into a compact scheme called RoboTokens, training a diffusion transformer (DiT) over them. The same model predicts dynamics using those predictions ("Dynamics Self-Guidance") to push generated designs toward higher reward at inference time. Masking different token types (using the BERT/MAE masked-modeling trick) lets the one model do three jobs: generate an embodiment, control an arbitrary embodiment, or design one conditioned on a motion. It is trained on 11 robots from the MuJoCo Menagerie (0.65 kg hand to 67.5 kg quadruped, 6–35 joints), and validated in sim and on a physical ALOHA doing cloth flinging. I like the fact that this approach inverts the entire recent robotics ideas: designing a policy for a fixed robot -> designing the robot for a fixed task. Every other approach assumes the body is given and learns a controller. Transformer Transformer takes the task (target motion + reward), then generates the body and controller jointly. In practice, it is a ~180× speedup over the standard optimizer at equal-or-better quality. It reaches "CMA-ES-level quality in seconds" and finishes bimanual multi-trajectory optimization in that is worth underlining nowadays! Also worth mentioning: this is the lab behind UMI and Handroid, that I mentioned here previously! The team seems extremely creative, i love these out-of-the-box approaches. Enjoy watching the demo of robot optimization in 3D, data acquisition, then real-life testing:

Léo

24,267 Aufrufe • vor 3 Tagen

Robots might learn better from video than from language! 📼 Most Vision-Language-Action (VLA) models learn what to do from text, but still struggle with how things move in the real world. That makes them data-hungry and slow to train. mimic video takes a different route. Instead of grounding robot control in text, it grounds it in video, using large pre-trained video models that already capture physical motion and dynamics. The idea is straightforward: let the video model handle “what will happen next,” and let a smaller control model focus only on turning that visual plan into robot actions. The result is big gains in practice. Robots trained this way need 10× less data, converge twice as fast, and perform better on both simulated benchmarks and real bimanual manipulation tasks. If robots can “imagine” motion using video, control becomes a much simpler problem. Shoutout to Jonas Pai, Liam Achenbach, Oier Mees, Elvis Nava and the rest of the team! Here's the project page: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Robots might learn better from video than from language! 📼 Most Vision-Language-Action (VLA) models learn what to do from text, but still struggle with how things move in the real world. That makes them data-hungry and slow to train. mimic video takes a different route. Instead of grounding robot control in text, it grounds it in video, using large pre-trained video models that already capture physical motion and dynamics. The idea is straightforward: let the video model handle “what will happen next,” and let a smaller control model focus only on turning that visual plan into robot actions. The result is big gains in practice. Robots trained this way need 10× less data, converge twice as fast, and perform better on both simulated benchmarks and real bimanual manipulation tasks. If robots can “imagine” motion using video, control becomes a much simpler problem. Shoutout to Jonas Pai, Liam Achenbach, Oier Mees, Elvis Nava and the rest of the team! Here's the project page: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

50,063 Aufrufe • vor 7 Monaten

X Square Robot Unveils New Embodied AI Model, Says Robots Will Arrive in Homes in 35 Days Backed by Alibaba, ByteDance, Xiaomi and Meituan, X Square Robot unveiled a next-generation embodied AI foundation model for home robots and said its first deployments in everyday households will begin within 35 days. X Square Robot on Tuesday unveiled WALL-B, a new embodied AI foundation model designed for deployment in real-world homes, marking what the company described as a major step toward bringing general-purpose robots into daily family life. At a launch event themed "Born to Bot, Bot to Family," the company also introduced its World Unified Model (WUM) architecture, a training framework that combines vision, language, action and physical prediction within a single system from the outset. X Square said the model is intended to help robots operate in the far more unpredictable setting of a home, where tasks, layouts and interactions vary from moment to moment. "Robots in factories and in homes are completely different. In factories, they repeat the same action 10,000 times without variation. In a home, however, they need to perform 10,000 different actions, each unique and non-repetitive. Therefore, the challenge of a truly intelligent robot lies not in repeating a single action, but in the ability to execute new, untrained movements within unstructured environments. Deploying robots in the home is one of the most significant technical hurdles of our time," said Qian Wang, founder and CEO of X Square Robot. WALL-B is the first real-world implementation of the World Unified Model architecture. Unlike modular systems that train perception, language and control separately, X Square Robot said World Unified Model optimizes those capabilities jointly from the very beginning. The company said that allows physical prediction — including force, friction and collision dynamics — to emerge as part of the model itself, rather than being layered on afterward. "We train all capabilities—vision, language, action, and prediction—within the same network from day one. Much like infants, who do not learn to see, move and speak in isolated, sequential stages, but instead see, move listen and act simultaneously while receiving feedback, we have integrated all these capabilities into a unified whole," said Wang Hao, CTO of X Square. X Square Robot said the development of WALL-B rests on two pillars. The first is a data strategy that prioritizes training on authentic, non-staged home environments to cover the “long-tail” distribution of real-world scenarios, such as misplaced objects and temporary occlusions. Unlike models primarily trained on synthetic data or laboratory datasets, this strategy exposes WALL-B to the natural clutter of lived-in spaces—misplaced items, unexpected obstacles, and spontaneous human activity—ensuring that the training data reflects real-world conditions rather than a simplified version. The second is a physics-aware predictive mechanism that anticipates physical outcomes before an action is taken, enabling the model to respond to contact dynamics instead of just reacting. The development of the self-developed WUM architecture on physical robotic platforms highlights the company’s accumlated experience in bridging sim-to-real gaps across varied operational contexts. Wang commented that the current AI model is still in an "intern" stage, subject to errors requiring remote assistance. For instance, it may mistakenly place slippers in the kitchen or pause while wiping a table to "think". However, the model operates nonstop 24 hours a day, becoming increasingly "intelligent" as each day of operation generates new data. In 35 days, on May 25, X Square Robot will officially bring its robots into everyday homes, underscoring the company’s long-term commitment to the home robotics sector.

X Square Robot Unveils New Embodied AI Model, Says Robots Will Arrive in Homes in 35 Days Backed by Alibaba, ByteDance, Xiaomi and Meituan, X Square Robot unveiled a next-generation embodied AI foundation model for home robots and said its first deployments in everyday households will begin within 35 days. X Square Robot on Tuesday unveiled WALL-B, a new embodied AI foundation model designed for deployment in real-world homes, marking what the company described as a major step toward bringing general-purpose robots into daily family life. At a launch event themed "Born to Bot, Bot to Family," the company also introduced its World Unified Model (WUM) architecture, a training framework that combines vision, language, action and physical prediction within a single system from the outset. X Square said the model is intended to help robots operate in the far more unpredictable setting of a home, where tasks, layouts and interactions vary from moment to moment. "Robots in factories and in homes are completely different. In factories, they repeat the same action 10,000 times without variation. In a home, however, they need to perform 10,000 different actions, each unique and non-repetitive. Therefore, the challenge of a truly intelligent robot lies not in repeating a single action, but in the ability to execute new, untrained movements within unstructured environments. Deploying robots in the home is one of the most significant technical hurdles of our time," said Qian Wang, founder and CEO of X Square Robot. WALL-B is the first real-world implementation of the World Unified Model architecture. Unlike modular systems that train perception, language and control separately, X Square Robot said World Unified Model optimizes those capabilities jointly from the very beginning. The company said that allows physical prediction — including force, friction and collision dynamics — to emerge as part of the model itself, rather than being layered on afterward. "We train all capabilities—vision, language, action, and prediction—within the same network from day one. Much like infants, who do not learn to see, move and speak in isolated, sequential stages, but instead see, move listen and act simultaneously while receiving feedback, we have integrated all these capabilities into a unified whole," said Wang Hao, CTO of X Square. X Square Robot said the development of WALL-B rests on two pillars. The first is a data strategy that prioritizes training on authentic, non-staged home environments to cover the “long-tail” distribution of real-world scenarios, such as misplaced objects and temporary occlusions. Unlike models primarily trained on synthetic data or laboratory datasets, this strategy exposes WALL-B to the natural clutter of lived-in spaces—misplaced items, unexpected obstacles, and spontaneous human activity—ensuring that the training data reflects real-world conditions rather than a simplified version. The second is a physics-aware predictive mechanism that anticipates physical outcomes before an action is taken, enabling the model to respond to contact dynamics instead of just reacting. The development of the self-developed WUM architecture on physical robotic platforms highlights the company’s accumlated experience in bridging sim-to-real gaps across varied operational contexts. Wang commented that the current AI model is still in an "intern" stage, subject to errors requiring remote assistance. For instance, it may mistakenly place slippers in the kitchen or pause while wiping a table to "think". However, the model operates nonstop 24 hours a day, becoming increasingly "intelligent" as each day of operation generates new data. In 35 days, on May 25, X Square Robot will officially bring its robots into everyday homes, underscoring the company’s long-term commitment to the home robotics sector.

X Square Robot

52,968 Aufrufe • vor 3 Monaten

Excited to announce GR00T N1, the world’s first open foundation model for humanoid robots! We are on a mission to democratize Physical AI. The power of general robot brain, in the palm of your hand - with only 2B parameters, N1 learns from the most diverse physical action dataset ever compiled and punches above its weight: - Real humanoid teleoperation data. - Large-scale simulation data: we are open-sourcing 300K+ trajectories! - Neural trajectories: we apply SOTA video generation models to “hallucinate” new synthetic data that features accurate physics in pixels. Using Jensen’s words, “systematically infinite data”! - Latent actions: we develop novel algorithms to extract action tokens from in-the-wild human videos and neural generated videos. GR00T N1 is a single end-to-end neural net, from photons to actions: - Vision-Language Model (System 2) that interprets the physical world through vision and language instructions, enabling robots to reason about their environment and instructions, and plan the right actions. - Diffusion Transformer (System 1) that “renders” smooth and precise motor actions at 120 Hz, executing the latent plan made by System 2. We deploy N1 on GR1 robot, 1X Neo robot, and a large collection of simulation benchmarks. N1 achieves up to +30% boost in diverse manipulation tasks for household and industrial settings. While humanoid robots are the main focus of N1, our model also supports cross-embodiment. We finetune it to work on the $110 HuggingFace LeRobot SO100 robot arm! Open robot brain runs on open hardware. Sounds just right. Let’s solve robotics, together, one token at a time. Links to our Whitepaper, Github repo, HuggingFace model, and open dataset page in the thread: 🧵

Excited to announce GR00T N1, the world’s first open foundation model for humanoid robots! We are on a mission to democratize Physical AI. The power of general robot brain, in the palm of your hand - with only 2B parameters, N1 learns from the most diverse physical action dataset ever compiled and punches above its weight: - Real humanoid teleoperation data. - Large-scale simulation data: we are open-sourcing 300K+ trajectories! - Neural trajectories: we apply SOTA video generation models to “hallucinate” new synthetic data that features accurate physics in pixels. Using Jensen’s words, “systematically infinite data”! - Latent actions: we develop novel algorithms to extract action tokens from in-the-wild human videos and neural generated videos. GR00T N1 is a single end-to-end neural net, from photons to actions: - Vision-Language Model (System 2) that interprets the physical world through vision and language instructions, enabling robots to reason about their environment and instructions, and plan the right actions. - Diffusion Transformer (System 1) that “renders” smooth and precise motor actions at 120 Hz, executing the latent plan made by System 2. We deploy N1 on GR1 robot, 1X Neo robot, and a large collection of simulation benchmarks. N1 achieves up to +30% boost in diverse manipulation tasks for household and industrial settings. While humanoid robots are the main focus of N1, our model also supports cross-embodiment. We finetune it to work on the $110 HuggingFace LeRobot SO100 robot arm! Open robot brain runs on open hardware. Sounds just right. Let’s solve robotics, together, one token at a time. Links to our Whitepaper, Github repo, HuggingFace model, and open dataset page in the thread: 🧵

Jim Fan

466,333 Aufrufe • vor 1 Jahr

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

42,864 Aufrufe • vor 6 Monaten

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 Aufrufe • vor 6 Monaten

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Léo

32,078 Aufrufe • vor 16 Tagen

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

RoboPapers

38,334 Aufrufe • vor 1 Monat

Foundation models are enough to solve robotics! Unfortunately, this is not true. We keep hearing that Vision-Language-Action (VLA) models struggle because of the gap between static training and the dynamic real world. A German startup (Sereact) just released a solution that bridges this gap perfectly. They are introducing a new paradigm called Interactive RL Policy Patching. It's a distributed framework that allows robots to self-learn from human corrections without needing full retraining. When a robot fails, a human operator provides a brief "patch" or demonstration. The system then uses online off-policy reinforcement learning to update the behavior instantly. This is powered by a massive foundation model trained on hundreds of millions of interactions from over 100 deployed robot stations. The best part is the distributed parameter synchronization... When one robot learns a fix, the update is published fleet-wide... meaning the entire swarm gets smarter from a single human intervention. They are already proving this on complex manipulation tasks like shoe unboxing and screw sorting, drastically reducing the data needed to handle edge cases. Real-world environments are unforgiving, and I love seeing systems that can actually adapt on the fly! 📍 More info:

Foundation models are enough to solve robotics! Unfortunately, this is not true. We keep hearing that Vision-Language-Action (VLA) models struggle because of the gap between static training and the dynamic real world. A German startup (Sereact) just released a solution that bridges this gap perfectly. They are introducing a new paradigm called Interactive RL Policy Patching. It's a distributed framework that allows robots to self-learn from human corrections without needing full retraining. When a robot fails, a human operator provides a brief "patch" or demonstration. The system then uses online off-policy reinforcement learning to update the behavior instantly. This is powered by a massive foundation model trained on hundreds of millions of interactions from over 100 deployed robot stations. The best part is the distributed parameter synchronization... When one robot learns a fix, the update is published fleet-wide... meaning the entire swarm gets smarter from a single human intervention. They are already proving this on complex manipulation tasks like shoe unboxing and screw sorting, drastically reducing the data needed to handle edge cases. Real-world environments are unforgiving, and I love seeing systems that can actually adapt on the fly! 📍 More info:

Ilir Aliu

18,210 Aufrufe • vor 6 Monaten

We’re excited to share DiT4DiT, an end-to-end Video-Action Model for robot learning that unifies a video Diffusion Transformer and an action Diffusion Transformer in a single cascaded framework. By leveraging the rich spatiotemporal and physical dynamics learned through video generation, rather than static image-text priors, DiT4DiT achieves state-of-the-art results on LIBERO (98.6%) and RoboCasa GR1 (50.8%) with far less training data, delivering over 10× better sample efficiency and up to 7× faster convergence. Real-world deployment on a humanoid robot further shows robust generalization. We believe this is a step toward making video generation a powerful backbone for robot policy learning. This work builds upon the brilliant foundations laid by Nvidia's GR00T and Cosmos. Project: Paper: Code: Coming soon. In the meantime, you can ask your coding agent to reproduce the method based on GR00T/Cosmos.

We’re excited to share DiT4DiT, an end-to-end Video-Action Model for robot learning that unifies a video Diffusion Transformer and an action Diffusion Transformer in a single cascaded framework. By leveraging the rich spatiotemporal and physical dynamics learned through video generation, rather than static image-text priors, DiT4DiT achieves state-of-the-art results on LIBERO (98.6%) and RoboCasa GR1 (50.8%) with far less training data, delivering over 10× better sample efficiency and up to 7× faster convergence. Real-world deployment on a humanoid robot further shows robust generalization. We believe this is a step toward making video generation a powerful backbone for robot policy learning. This work builds upon the brilliant foundations laid by Nvidia's GR00T and Cosmos. Project: Paper: Code: Coming soon. In the meantime, you can ask your coding agent to reproduce the method based on GR00T/Cosmos.

Shuo Yang

31,872 Aufrufe • vor 4 Monaten

Pretraining is essential for good performance on a wide variety of robotics tasks, and so most vision-language-action models build off of a vision language model (VLM) trained on a wide variety of image-language data. But how does the choice of VLM translate to downstream robotics performance? Jianke Zhang and @GYanjiang join us to talk about this key part of the robot policy, looking at a wide variety of different VLMs and how they perform. Interestingly, they see that performance on auxiliary tasks like quesiton answering did not lead to downstream improvements in control. To learn more, watch episode 65 of RoboPapers now, with Chris Paxton and Jiafei Duan!

Pretraining is essential for good performance on a wide variety of robotics tasks, and so most vision-language-action models build off of a vision language model (VLM) trained on a wide variety of image-language data. But how does the choice of VLM translate to downstream robotics performance? Jianke Zhang and @GYanjiang join us to talk about this key part of the robot policy, looking at a wide variety of different VLMs and how they perform. Interestingly, they see that performance on auxiliary tasks like quesiton answering did not lead to downstream improvements in control. To learn more, watch episode 65 of RoboPapers now, with Chris Paxton and Jiafei Duan!

RoboPapers

23,905 Aufrufe • vor 5 Monaten

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 (R² = 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:

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 (R² = 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

294,395 Aufrufe • vor 5 Monaten

Karpathy told Dwarkesh that a 1 billion parameter model, trained on clean data, could hit the intelligence of today's 1.8 trillion parameter frontier. That is a 1,800x compression claim. The math behind it is more defensible than it sounds. When researchers at frontier labs look at random samples from their training corpus, they see stock ticker symbols, broken HTML, forum spam, autogenerated gibberish. Not Wikipedia. Not the Wall Street Journal. The actual pretraining dataset is mostly noise, and the model is burning parameters to vaguely remember all of it. One estimate pegs Llama 3's information compression at 0.07 bits per token. Well-structured English carries around 1.5 bits per token of real information. The trillion-parameter model is holding a roughly 5% resolution image of the internet it trained on. So when a lab ships a 1.8 trillion parameter model, the overwhelming majority of those weights are handling rough memorization. They are compression overhead for a noisy training set, taking up capacity that could be doing reasoning instead. Karpathy's proposal is to separate the two. Build a cognitive core: a small model that contains only the algorithms for reasoning and problem-solving, stripped of encyclopedic memorization. Pair it with external memory the model queries when it needs a fact. A 1 billion parameter reasoner plus retrieval beats a 1.8 trillion parameter model trying to do both. The data already supports this direction. GPT-4o runs at roughly 200 billion parameters and outperforms the original 1.8 trillion GPT-4. Inference costs for GPT-3.5 level performance fell 280x between 2022 and 2024, driven almost entirely by smaller, cleaner, better-architected models. The trend line is pointing where Karpathy says it should. The real implication for anyone tracking the AI trade: data quality is the actual constraint. The companies winning the next phase will be the ones who figured out what to train on, and what to throw away.

Karpathy told Dwarkesh that a 1 billion parameter model, trained on clean data, could hit the intelligence of today's 1.8 trillion parameter frontier. That is a 1,800x compression claim. The math behind it is more defensible than it sounds. When researchers at frontier labs look at random samples from their training corpus, they see stock ticker symbols, broken HTML, forum spam, autogenerated gibberish. Not Wikipedia. Not the Wall Street Journal. The actual pretraining dataset is mostly noise, and the model is burning parameters to vaguely remember all of it. One estimate pegs Llama 3's information compression at 0.07 bits per token. Well-structured English carries around 1.5 bits per token of real information. The trillion-parameter model is holding a roughly 5% resolution image of the internet it trained on. So when a lab ships a 1.8 trillion parameter model, the overwhelming majority of those weights are handling rough memorization. They are compression overhead for a noisy training set, taking up capacity that could be doing reasoning instead. Karpathy's proposal is to separate the two. Build a cognitive core: a small model that contains only the algorithms for reasoning and problem-solving, stripped of encyclopedic memorization. Pair it with external memory the model queries when it needs a fact. A 1 billion parameter reasoner plus retrieval beats a 1.8 trillion parameter model trying to do both. The data already supports this direction. GPT-4o runs at roughly 200 billion parameters and outperforms the original 1.8 trillion GPT-4. Inference costs for GPT-3.5 level performance fell 280x between 2022 and 2024, driven almost entirely by smaller, cleaner, better-architected models. The trend line is pointing where Karpathy says it should. The real implication for anyone tracking the AI trade: data quality is the actual constraint. The companies winning the next phase will be the ones who figured out what to train on, and what to throw away.

Aakash Gupta

508,200 Aufrufe • vor 3 Monaten

Autonomous driving with Chain of Thought - autopilot thinking out loud in text! LINGO-1 is the most interesting work I've read in autodriving for a while. Before: perception -> driving action After: perception -> textual reasoning -> action LINGO-1 trains a video-language model that comments on the ongoing scene. You can ask it to explain its decisions ("why are you stopped?") and planning ("what are you gonna do next?"). The explicit reasoning step comes with key benefits: - Explainability: driving models are no longer a mysterious blackbox that you pray for safety. - Counterfactuals: it's able to imagine scenarios that are not in the training data, and reason through how to handle them correctly. - Long-tail programming: there are soooo many edge cases in driving. It's impossible to have good data coverage on everything. Instead of collecting 1000s of examples to "neural program" a case, you can now have a human teacher write prompts to explain a handful of examples. LINGO-1 is closely related to a few works in game AI: - MineDojo (my team's work at NVIDIA, learns a reward model that aligns Minecraft gameplay videos with their transcripts. The model, called "MineCLIP", is able to ground commentary text in the video pixels. - Thought Cloning (Jeff Clune): pixel -> language -> action loop in gridworlds.

Autonomous driving with Chain of Thought - autopilot thinking out loud in text! LINGO-1 is the most interesting work I've read in autodriving for a while. Before: perception -> driving action After: perception -> textual reasoning -> action LINGO-1 trains a video-language model that comments on the ongoing scene. You can ask it to explain its decisions ("why are you stopped?") and planning ("what are you gonna do next?"). The explicit reasoning step comes with key benefits: - Explainability: driving models are no longer a mysterious blackbox that you pray for safety. - Counterfactuals: it's able to imagine scenarios that are not in the training data, and reason through how to handle them correctly. - Long-tail programming: there are soooo many edge cases in driving. It's impossible to have good data coverage on everything. Instead of collecting 1000s of examples to "neural program" a case, you can now have a human teacher write prompts to explain a handful of examples. LINGO-1 is closely related to a few works in game AI: - MineDojo (my team's work at NVIDIA, learns a reward model that aligns Minecraft gameplay videos with their transcripts. The model, called "MineCLIP", is able to ground commentary text in the video pixels. - Thought Cloning (Jeff Clune): pixel -> language -> action loop in gridworlds.

Jim Fan

552,790 Aufrufe • vor 2 Jahren

$Introducing HRM-Text. An ultra-lean 1B-parameter reasoning language model designed to deliver strong general performance with a fraction of the data, compute, and infrastructure. Trained on just 40B structured tokens, HRM-Text achieves competitive performance while using ~1/1000 of the training data of comparable models. The kicker? The full model trains in roughly one day on a $1,000 budget. This opens the door to a new generation of AI that is powerful, accessible, and radically easier to adapt. Theories and research concepts once deemed too expensive to test are officially back in the game. Sapient Intelligence invites you to help us shape a new paradigm for general intelligence.$

Introducing HRM-Text. An ultra-lean 1B-parameter reasoning language model designed to deliver strong general performance with a fraction of the data, compute, and infrastructure. Trained on just 40B structured tokens, HRM-Text achieves competitive performance while using ~1/1000 of the training data of comparable models. The kicker? The full model trains in roughly one day on a $1,000 budget. This opens the door to a new generation of AI that is powerful, accessible, and radically easier to adapt. Theories and research concepts once deemed too expensive to test are officially back in the game. Sapient Intelligence invites you to help us shape a new paradigm for general intelligence.

Sapient Intelligence

515,089 Aufrufe • vor 2 Monaten

🇨🇳 It has started. A new home service in China pairs human cleaners with autonomous AI robots to tackle household chores. Residents in Shenzhen can now book a service where a human professional and an autonomous robot arrive together to clean their home. Real houses present a chaotic mess of dropped toys and random furniture that confuse traditional machines. X Square Robot and a major service platform named 58[.]com decided to tackle this chaos by launching China's first robot cleaner service in March-26. Customers use an application to hire a cleaning crew that consists of 1 human worker and 1 robot. The human takes care of the tricky chores that require complex judgment. The robot handles the repetitive physical work like picking up trash and wiping down flat surfaces. This machine runs on a system called WALL-A, which acts as a single continuous AI brain rather than a list of pre-written rules. They built this AI foundation model to perceive its surroundings and make its own decisions without human guidance. It processes visual data and plans multi-step actions. And deploying these robots into actual homes now provides the massive amounts of extremely important training data to improve it continuously. Alibaba and ByteDance backed this project. IMO, if the foundational model behind it figures out how to navigate a messy living room without getting stuck, it can learn to operate in almost any other physical environment.

🇨🇳 It has started. A new home service in China pairs human cleaners with autonomous AI robots to tackle household chores. Residents in Shenzhen can now book a service where a human professional and an autonomous robot arrive together to clean their home. Real houses present a chaotic mess of dropped toys and random furniture that confuse traditional machines. X Square Robot and a major service platform named 58[.]com decided to tackle this chaos by launching China's first robot cleaner service in March-26. Customers use an application to hire a cleaning crew that consists of 1 human worker and 1 robot. The human takes care of the tricky chores that require complex judgment. The robot handles the repetitive physical work like picking up trash and wiping down flat surfaces. This machine runs on a system called WALL-A, which acts as a single continuous AI brain rather than a list of pre-written rules. They built this AI foundation model to perceive its surroundings and make its own decisions without human guidance. It processes visual data and plans multi-step actions. And deploying these robots into actual homes now provides the massive amounts of extremely important training data to improve it continuously. Alibaba and ByteDance backed this project. IMO, if the foundational model behind it figures out how to navigate a messy living room without getting stuck, it can learn to operate in almost any other physical environment.

Rohan Paul

52,996 Aufrufe • vor 4 Monaten

This is THE moment of Physical AI! We are officially announcing Cosmos 3: Omnimodal World Models for Physical AI 🚀 - Cosmos 3 is an omnimodal world model: within a unified architecture, it can understand and generate language, images, video, audio, and actions. - It is not just a VLM, not just a video generator, not just an audio-visual generative model, and not just a physics simulator / world-action model. It can understand images and videos, generate images, videos, and audio, simulate future worlds, predict actions, and generate robot policies—enabling models to truly begin to “touch the world.” - Cosmos 3 is the #1 open-weight reasoner / T2I / I2V / robot policy across many benchmarks. Huge thanks to every teammate who fought side by side on this journey—from architecture, data, training, infra, serving, and evaluation to post-training. Every part of this project carries an incredible amount of hard work. This was my first time leading a project as Tech Lead, and I feel truly fortunate. The future of Physical AI needs models that can not only “see” and “describe” the world, but also “imagine,” “simulate,” and “act”—and eventually close the loop with the real world. I hope Cosmos 3 can become an important starting point for this direction, and I’m excited to push Physical AI into its next stage together with the open-source community. Welcome to the era of Physical AI. HuggingFace: Project Website: Code:

This is THE moment of Physical AI! We are officially announcing Cosmos 3: Omnimodal World Models for Physical AI 🚀 - Cosmos 3 is an omnimodal world model: within a unified architecture, it can understand and generate language, images, video, audio, and actions. - It is not just a VLM, not just a video generator, not just an audio-visual generative model, and not just a physics simulator / world-action model. It can understand images and videos, generate images, videos, and audio, simulate future worlds, predict actions, and generate robot policies—enabling models to truly begin to “touch the world.” - Cosmos 3 is the #1 open-weight reasoner / T2I / I2V / robot policy across many benchmarks. Huge thanks to every teammate who fought side by side on this journey—from architecture, data, training, infra, serving, and evaluation to post-training. Every part of this project carries an incredible amount of hard work. This was my first time leading a project as Tech Lead, and I feel truly fortunate. The future of Physical AI needs models that can not only “see” and “describe” the world, but also “imagine,” “simulate,” and “act”—and eventually close the loop with the real world. I hope Cosmos 3 can become an important starting point for this direction, and I’m excited to push Physical AI into its next stage together with the open-source community. Welcome to the era of Physical AI. HuggingFace: Project Website: Code:

Max Zhaoshuo Li 李赵硕

1,078,282 Aufrufe • vor 2 Monaten