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.

Introducing Neural Jacobian Fields, robot 3D kinematic models learned only from vision! They can model & control robots from just a single RGB camera, even those w/ intractable kinematics & no embedded sensors such as soft, 3D-printed pneumatic hands! 1/n

Vincent Sitzmann

19,445 subscribers

54,204 Aufrufe • vor 2 Jahren •via X (Twitter)

Bildung Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

14 Kommentare

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

This project was led by the outstanding @sizhe_lester_li, in a multi-disciplinary collaboration with @annan__zhang, @BoyuanChen0, Hanna Matusik, Chao Liu, and Daniela Rus! Amazing team - truly grateful for the soft robot chops hands-on contributed by @annan__zhang & co! 2/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

While we humans constantly use vision for closed-loop control of our hands and limbs, today’s robots instead rely completely on embedded sensors, precision manufacturing and expert-designed forward kinematic models to understand where they are in 3D: 3/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Conventional robots are “blind” when it comes to state estimation! This constrains their design to be “simple” such that an expert can design a model & requires expensive manufacturing & sensors. But bio-inspired, multi-material, and cheap robots cannot be modeled easily! 4/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Our approach enables this pneumatic bio-inspired hand to perform physical tasks and allows the $220 poppy arm to draw “MIT” in the air, all with using a single RGB camera as the only sensor! None of these motion trajectories are prescribed as part of our training data. 5/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Neural Jacobian Fields enable us to equip any robot with vision-based control, irrespective of its sensors, fabrication, material, or actuation. Each point is mapped to its “system Jacobian”, which maps a change in motor commands to the 3D motion of that point! 6/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

The Jacobian Field can be directly used for inverse dynamics control. Given desired motions, our model solves for the corresponding control command at interactive speeds. 7/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Neural Jacobian Fields are trained completely self-supervised, from multi-view videos of a robot executing random commands - no human labels or intervention. At test time, a *single* image suffices to reconstruct them, for closed-loop control from a single RGB camera. 8/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

The learned Jacobian Field is interpretable: we can visualize it by color-coding sensitivity to the different control channels. In this way, we can see that the Jacobian field correctly discovers the 3D kinematics of the robot. 9/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

We’re not advocating for *only* using vision. In this paper, that’s what we do, to show how *surprisingly powerful* it can be! But: the more sensors, the better. It is straightforward to add more sensors - just add them to the input to reconstruct the jacobian field! 10/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Neural Jacobian Fields could make robotic automation more affordable by enabling control of cheap robots, as they allow us to skip expensive sensors, enable new fabrication techniques such as 3D printing, and are robust to imprecise manufacturing! 11/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

There are still ways to go - but we think that this direction is super promising and just the beginning of an exciting arc of building representations of robotic embodiments that can bridge vision, proprioception, touch, and more! 12/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Again, shoutout to the amazing team at MIT - make sure to check them out! @sizhe_lester_li, @annan__zhang, @BoyuanChen0, Hanna Matusik, Chao Liu, and Daniela Rus! Code & everything is already released: 13/n

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Also, I should add we are aware of another paper that is named "Neural Jacobian Fields": It's a completely different idea and problem and not directly related, but still really, really cool (check it out!)...

Profilbild von Vincent Sitzmann

Vincent Sitzmannvor 2 Jahren

Our paper is actually called "Unifying 3D Representation and Control of Diverse Robots with a Single Camera". We thought about not naming the core method "Neural Jacobian Fields", but it makes the most sense, b/c we are really modeling the "System Jacobian", a staple in robotics.

Ähnliche Videos

Supervised learning has held 3D Vision back for too long. Meet RayZer — a self-supervised 3D model trained with zero 3D labels: ❌ No supervision of camera & geometry ✅ Just RGB images And the wild part? RayZer outperforms supervised methods (as 3D labels from COLMAP is noisy) 🌐 Project: (1/4)

Supervised learning has held 3D Vision back for too long. Meet RayZer — a self-supervised 3D model trained with zero 3D labels: ❌ No supervision of camera & geometry ✅ Just RGB images And the wild part? RayZer outperforms supervised methods (as 3D labels from COLMAP is noisy) 🌐 Project: (1/4)

Hanwen Jiang

70,830 Aufrufe • vor 1 Jahr

3D-LLM: Injecting the 3D World into Large Language Models paper page: Large language models (LLMs) and Vision-Language Models (VLMs) have been proven to excel at multiple tasks, such as commonsense reasoning. Powerful as these models can be, they are not grounded in the 3D physical world, which involves richer concepts such as spatial relationships, affordances, physics, layout, and so on. In this work, we propose to inject the 3D world into large language models and introduce a whole new family of 3D-LLMs. Specifically, 3D-LLMs can take 3D point clouds and their features as input and perform a diverse set of 3D-related tasks, including captioning, dense captioning, 3D question answering, task decomposition, 3D grounding, 3D-assisted dialog, navigation, and so on. Using three types of prompting mechanisms that we design, we are able to collect over 300k 3D-language data covering these tasks. To efficiently train 3D-LLMs, we first utilize a 3D feature extractor that obtains 3D features from rendered multi- view images. Then, we use 2D VLMs as our backbones to train our 3D-LLMs. By introducing a 3D localization mechanism, 3D-LLMs can better capture 3D spatial information. Experiments on ScanQA show that our model outperforms state-of-the-art baselines by a large margin (e.g., the BLEU-1 score surpasses state-of-the-art score by 9%). Furthermore, experiments on our held-in datasets for 3D captioning, task composition, and 3D-assisted dialogue show that our model outperforms 2D VLMs. Qualitative examples also show that our model could perform more tasks beyond the scope of existing LLMs and VLMs.

3D-LLM: Injecting the 3D World into Large Language Models paper page: Large language models (LLMs) and Vision-Language Models (VLMs) have been proven to excel at multiple tasks, such as commonsense reasoning. Powerful as these models can be, they are not grounded in the 3D physical world, which involves richer concepts such as spatial relationships, affordances, physics, layout, and so on. In this work, we propose to inject the 3D world into large language models and introduce a whole new family of 3D-LLMs. Specifically, 3D-LLMs can take 3D point clouds and their features as input and perform a diverse set of 3D-related tasks, including captioning, dense captioning, 3D question answering, task decomposition, 3D grounding, 3D-assisted dialog, navigation, and so on. Using three types of prompting mechanisms that we design, we are able to collect over 300k 3D-language data covering these tasks. To efficiently train 3D-LLMs, we first utilize a 3D feature extractor that obtains 3D features from rendered multi- view images. Then, we use 2D VLMs as our backbones to train our 3D-LLMs. By introducing a 3D localization mechanism, 3D-LLMs can better capture 3D spatial information. Experiments on ScanQA show that our model outperforms state-of-the-art baselines by a large margin (e.g., the BLEU-1 score surpasses state-of-the-art score by 9%). Furthermore, experiments on our held-in datasets for 3D captioning, task composition, and 3D-assisted dialogue show that our model outperforms 2D VLMs. Qualitative examples also show that our model could perform more tasks beyond the scope of existing LLMs and VLMs.

AK

249,798 Aufrufe • vor 3 Jahren

1/N 🎨🤖Given only a static image of an oil painting by an expert artist, can a robot infer the corresponding control actions, such as trajectory, orientation, and applied force, to accurately reproduce the painting? 🖌️Introducing IMPASTO: a robotic oil-painting system that integrates learned pixel dynamics with model-based planning to reproduce paintings from target images. This project is led by Yingke Wang ( – a roboticist and an artist. Project website:

1/N 🎨🤖Given only a static image of an oil painting by an expert artist, can a robot infer the corresponding control actions, such as trajectory, orientation, and applied force, to accurately reproduce the painting? 🖌️Introducing IMPASTO: a robotic oil-painting system that integrates learned pixel dynamics with model-based planning to reproduce paintings from target images. This project is led by Yingke Wang ( – a roboticist and an artist. Project website:

Ruohan Zhang

51,285 Aufrufe • vor 9 Monaten

How precise can a robot get… without ever seeing the object move? Turns out, just a geometric model is enough. A robot hit 73 precise 3D goals in a row, using only a geometric model. No demonstrations. No imitation learning. No fancy vision models. The secret? Sample where the robot could make contact. Let local contact-implicit control take over from there. That’s it. ✅ Real-time performance ✅ Natural contact making and breaking ✅ Works on tricky tasks like pushing and rotating This approach mixes global planning with local MPC. Fast. Accurate. Sample-efficient. Thanks for sharing, Michael Posa ! 📍Watch it in action: Paper:

How precise can a robot get… without ever seeing the object move? Turns out, just a geometric model is enough. A robot hit 73 precise 3D goals in a row, using only a geometric model. No demonstrations. No imitation learning. No fancy vision models. The secret? Sample where the robot could make contact. Let local contact-implicit control take over from there. That’s it. ✅ Real-time performance ✅ Natural contact making and breaking ✅ Works on tricky tasks like pushing and rotating This approach mixes global planning with local MPC. Fast. Accurate. Sample-efficient. Thanks for sharing, Michael Posa ! 📍Watch it in action: Paper:

Ilir Aliu

19,012 Aufrufe • vor 1 Jahr

Truly general-purpose robots must be able to navigate spaces they have never seen before. ⦿ Skild Brain enables end-to-end autonomous locomotion from raw vision and joint inputs, without mapping or pre-planning. ⦿ The model adapts in real time to new terrain such as stairs, gaps, and clutter, reacting instinctively like a human. There is no “stair mode” or “step-over mode”: the transition between walking behaviors is continuous and instinctive. ⦿ A single neural policy supports a wide range of locomotion tasks and can transfer across different robot platforms.

Truly general-purpose robots must be able to navigate spaces they have never seen before. ⦿ Skild Brain enables end-to-end autonomous locomotion from raw vision and joint inputs, without mapping or pre-planning. ⦿ The model adapts in real time to new terrain such as stairs, gaps, and clutter, reacting instinctively like a human. There is no “stair mode” or “step-over mode”: the transition between walking behaviors is continuous and instinctive. ⦿ A single neural policy supports a wide range of locomotion tasks and can transfer across different robot platforms.

The Humanoid Hub

28,439 Aufrufe • vor 1 Jahr

📢📢 𝐏𝐞𝐫𝐜𝐇𝐞𝐚𝐝: 𝐏𝐞𝐫𝐜𝐞𝐩𝐭𝐮𝐚𝐥 𝐇𝐞𝐚𝐝 𝐌𝐨𝐝𝐞𝐥 𝐟𝐨𝐫 𝐒𝐢𝐧𝐠𝐥𝐞-𝐈𝐦𝐚𝐠𝐞 𝟑𝐃 𝐇𝐞𝐚𝐝 𝐑𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧 & 𝐄𝐝𝐢𝐭𝐢𝐧𝐠📢📢 PercHead reconstructs realistic 3D heads from a single image and enables disentangled 3D editing via geometric controls and style inputs from images or text. At its core is a generalized 3D head decoder trained with perceptual supervision from DINOv2 and SAM 2.1. We find that our new perceptual loss formulation improves reconstruction fidelity compared to commonly-used methods such as LPIPS. Our trained reconstruction model is able to generate 3D-consistent heads from a single input image. Even with challenging side-view inputs, the model robustly infers missing regions for a coherent, high-fidelity output. In addition, our architecture seamlessly adapts to downstream tasks: by swapping the encoder, we can transform the model into a disentangled 3D editing pipeline. In this scenario, we can control geometry through - potentially hand-drawn - segmentation maps, and condition style via image or text prompt. We also provide an interactive GUI to enable the exploration of our editing pipeline. 🌍 📽️ Great work by Antonio Oroz and Tobias Kirschstein

📢📢 𝐏𝐞𝐫𝐜𝐇𝐞𝐚𝐝: 𝐏𝐞𝐫𝐜𝐞𝐩𝐭𝐮𝐚𝐥 𝐇𝐞𝐚𝐝 𝐌𝐨𝐝𝐞𝐥 𝐟𝐨𝐫 𝐒𝐢𝐧𝐠𝐥𝐞-𝐈𝐦𝐚𝐠𝐞 𝟑𝐃 𝐇𝐞𝐚𝐝 𝐑𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧 & 𝐄𝐝𝐢𝐭𝐢𝐧𝐠📢📢 PercHead reconstructs realistic 3D heads from a single image and enables disentangled 3D editing via geometric controls and style inputs from images or text. At its core is a generalized 3D head decoder trained with perceptual supervision from DINOv2 and SAM 2.1. We find that our new perceptual loss formulation improves reconstruction fidelity compared to commonly-used methods such as LPIPS. Our trained reconstruction model is able to generate 3D-consistent heads from a single input image. Even with challenging side-view inputs, the model robustly infers missing regions for a coherent, high-fidelity output. In addition, our architecture seamlessly adapts to downstream tasks: by swapping the encoder, we can transform the model into a disentangled 3D editing pipeline. In this scenario, we can control geometry through - potentially hand-drawn - segmentation maps, and condition style via image or text prompt. We also provide an interactive GUI to enable the exploration of our editing pipeline. 🌍 📽️ Great work by Antonio Oroz and Tobias Kirschstein

Matthias Niessner

18,899 Aufrufe • vor 10 Monaten

THE WOVEN ROBOT! Wow! The woven robot hand is here and it is absolutely game-changing. Allonic just braided thousands of tendon-like fibers around a 3D-printed skeleton in minutes. No joints. No screws. No human assembly. Pure biomimetic tissue that flexes, grips, and lasts like biology itself. This isn’t incremental. This is the breakthrough that collapses robot hand costs from $30,000 to pennies. The old world of expensive, fragile robotic limbs is over. Now humanoid robots can build more of themselves scalable, replicable, anywhere. Robots building Robots building Robots, yeah you get it. In your freaking garage! Massive opportunities are about to explode as work as we know it fades. Rent one, have it build your fleet, send it back, repeat. Thousands if you want. The age of woven machines begins.

THE WOVEN ROBOT! Wow! The woven robot hand is here and it is absolutely game-changing. Allonic just braided thousands of tendon-like fibers around a 3D-printed skeleton in minutes. No joints. No screws. No human assembly. Pure biomimetic tissue that flexes, grips, and lasts like biology itself. This isn’t incremental. This is the breakthrough that collapses robot hand costs from $30,000 to pennies. The old world of expensive, fragile robotic limbs is over. Now humanoid robots can build more of themselves scalable, replicable, anywhere. Robots building Robots building Robots, yeah you get it. In your freaking garage! Massive opportunities are about to explode as work as we know it fades. Rent one, have it build your fleet, send it back, repeat. Thousands if you want. The age of woven machines begins.

Brian Roemmele

21,747 Aufrufe • vor 5 Monaten

A new 30-minute presentation from Ashok Elluswamy, Tesla’s VP of AI, has been released, where he talks about FSD, AI and the team’s latest progress. Highlight from the presentation: • Tesla's vehicle fleet can provide 500 years of driving data every single day. Curse of Dimensionality: • 8 cameras at high frame rate = billions of tokens per 30 seconds of driving context. • Tesla must compress and extract the right correlations between sensory input and control actions. Data Advantage: • Tesla has access to a “Niagara Falls of data” — hundreds of years’ worth of collective fleet driving. • Uses smart data triggers to capture rare corner cases (e.g., complex intersections, unpredictable behavior). Quality and Efficiency: • Extracts only the essential data needed to train models efficiently. Debugging and Interpretability: • Even though the system is end-to-end, Tesla can still prompt the model to output interpretable data: 3D occupancy, road boundaries, objects, signs, traffic lights, etc. • Natural language querying: ask the model why it made a certain decision. • These auxiliary predictions don’t drive the car but help engineers debug and ensure safety. Tesla’s Advanced Gaussian Splatting (3D Scene Modeling): • Tesla developed a custom, ultra-fast Gaussian splatting system to reconstruct 3D scenes from limited camera views. • Produces crisp, accurate 3D renderings even from few camera angles — far better than standard NeRF/splatting approaches. • Enables rapid visual debugging of the driving environment in 3D. Evaluation & World Models: • Evaluation is the hardest challenge: models may perform well offline but fail in real-world conditions. • Tesla builds balanced, diverse evaluation datasets focusing on edge cases — not just easy highway driving. Introduced a learned world simulator (neural network-generated video engine): • Can simulate 8 Tesla camera feeds simultaneously — fully synthetic. • Used for testing, training, and reinforcement learning. • Allows adversarial event injection (e.g., adding a pedestrian or vehicle cutting in). • Enables replaying past failures to verify new model improvements. • Can run in near real-time, letting testers “drive” inside a simulated world. What’s Next: • Scale robotaxi service globally. • Unlock full autonomy across the entire Tesla fleet. • Cybercab: next-gen 2-seat vehicle designed specifically for robotaxi use, targeting lowest transportation cost (cheaper than public transit). • Same neural networks will power Optimus humanoid robot. • The same video generation system is now being applied to Optimus. • The system can simulate and plan movement for robots, adapting easily to new forms. via the International Conference on Computer Vision (ICCV). Full presentation:

A new 30-minute presentation from Ashok Elluswamy, Tesla’s VP of AI, has been released, where he talks about FSD, AI and the team’s latest progress. Highlight from the presentation: • Tesla's vehicle fleet can provide 500 years of driving data every single day. Curse of Dimensionality: • 8 cameras at high frame rate = billions of tokens per 30 seconds of driving context. • Tesla must compress and extract the right correlations between sensory input and control actions. Data Advantage: • Tesla has access to a “Niagara Falls of data” — hundreds of years’ worth of collective fleet driving. • Uses smart data triggers to capture rare corner cases (e.g., complex intersections, unpredictable behavior). Quality and Efficiency: • Extracts only the essential data needed to train models efficiently. Debugging and Interpretability: • Even though the system is end-to-end, Tesla can still prompt the model to output interpretable data: 3D occupancy, road boundaries, objects, signs, traffic lights, etc. • Natural language querying: ask the model why it made a certain decision. • These auxiliary predictions don’t drive the car but help engineers debug and ensure safety. Tesla’s Advanced Gaussian Splatting (3D Scene Modeling): • Tesla developed a custom, ultra-fast Gaussian splatting system to reconstruct 3D scenes from limited camera views. • Produces crisp, accurate 3D renderings even from few camera angles — far better than standard NeRF/splatting approaches. • Enables rapid visual debugging of the driving environment in 3D. Evaluation & World Models: • Evaluation is the hardest challenge: models may perform well offline but fail in real-world conditions. • Tesla builds balanced, diverse evaluation datasets focusing on edge cases — not just easy highway driving. Introduced a learned world simulator (neural network-generated video engine): • Can simulate 8 Tesla camera feeds simultaneously — fully synthetic. • Used for testing, training, and reinforcement learning. • Allows adversarial event injection (e.g., adding a pedestrian or vehicle cutting in). • Enables replaying past failures to verify new model improvements. • Can run in near real-time, letting testers “drive” inside a simulated world. What’s Next: • Scale robotaxi service globally. • Unlock full autonomy across the entire Tesla fleet. • Cybercab: next-gen 2-seat vehicle designed specifically for robotaxi use, targeting lowest transportation cost (cheaper than public transit). • Same neural networks will power Optimus humanoid robot. • The same video generation system is now being applied to Optimus. • The system can simulate and plan movement for robots, adapting easily to new forms. via the International Conference on Computer Vision (ICCV). Full presentation:

Sawyer Merritt

1,288,166 Aufrufe • vor 10 Monaten

📢📢 𝐀𝐯𝐚𝐭𝟑𝐫 📢📢 Avat3r creates high-quality 3D head avatars from just a few input images in a single forward pass with a new dynamic 3DGS reconstruction model. Video: Project: Our core idea is to make Gaussian Reconstruction Models animatable. We find that a simple cross-attention to an expression code sequence is already sufficient to model complex facial expressions. We then incorporate position maps from DUSt3R and feature maps from Sapiens to facilitate the prediction task. While DUSt3R's position maps act as a pixel-aligned initialization for the Gaussians' positions, the Sapiens feature maps help the cross-view transformer to match corresponding image tokens in the 4 input images. One major challenge in creating a 3D head avatar from smartphone images comes from inconsistent facial expressions when the subject could not remain perfectly static during the capture. We eliminate this static requirement by simply showing our model input images with different facial expressions during training. This technique makes our model robust to inconsistent input images later on. Finally, we show that despite the model has been trained with 4 input images, one can even create a 3D head avatar when only a single image is available. To achieve this, we employ a pre-trained 3D GAN to lift the single image to 3D and then render the 4 input images for our model. This allows us to create 3D head avatars from single images and even highly out-of-distribution examples like AI generated faces, paintings or statues. Great work by Tobias Kirschstein from his internship at Meta with Javier Romero, Artem Sevastopolsky, and Shunsuke Saito

Matthias Niessner

74,763 Aufrufe • vor 1 Jahr

Picking up laundry with a 3D printed robot arm and LeRobot. It operates using SO-ARM101 arm mounted on a mobile base, relying on just a wrist and base camera. The base is controlled with an Xbox controller by a human operator. The arm is fully autonomous thanks to the ACT model. The robot can generalize to previously unseen environments. The technical stack: - Model: Action Chunking Transformer (ACT) with Temporal Ensemble - Data: Learned via teleoperation including negative examples and failure recoveries - Framework: Built and trained locally using LeRobot on a single consumer GPU in 3.5h This is a perfect example that real-world robotics isn't about perfect lab conditions - it's about messy living rooms, confusing shadows, and learning to recover from mistakes. And with open tools, anyone can tackle these edge cases right from their desk.

Picking up laundry with a 3D printed robot arm and LeRobot. It operates using SO-ARM101 arm mounted on a mobile base, relying on just a wrist and base camera. The base is controlled with an Xbox controller by a human operator. The arm is fully autonomous thanks to the ACT model. The robot can generalize to previously unseen environments. The technical stack: - Model: Action Chunking Transformer (ACT) with Temporal Ensemble - Data: Learned via teleoperation including negative examples and failure recoveries - Framework: Built and trained locally using LeRobot on a single consumer GPU in 3.5h This is a perfect example that real-world robotics isn't about perfect lab conditions - it's about messy living rooms, confusing shadows, and learning to recover from mistakes. And with open tools, anyone can tackle these edge cases right from their desk.

LeRobot

68,814 Aufrufe • vor 29 Tagen

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

300,036 Aufrufe • vor 6 Monaten

Introducing Open-TeleVision: with Fully Autonomous policy video👇. We can conduct a long-horizon task with inserting 12 cans nonstop without any interruptions. We offer: 🤖 Highly precise and smooth bimanual manipulation. 📺 Active egocentric vision (with a moving neck) feedback. It is achieved by imitation learning from teleoperation: We propose a VR-based REAL-TIME teleoperation that streams the stereo video observation from the robot camera to the VR device. The robot neck moves as the human head moves, the robot hands move as the human hands move, offering the operator an intuitive experience as the human herself becomes the robot. The devils are all in the details, and how to implement things right: ✅How to perform IK/retargeting for smooth and precise control. ✅How to do all these and also stream stereo video without no latency, all in real time. We released our code here: Active head hardware design: 1/n

Introducing Open-TeleVision: with Fully Autonomous policy video👇. We can conduct a long-horizon task with inserting 12 cans nonstop without any interruptions. We offer: 🤖 Highly precise and smooth bimanual manipulation. 📺 Active egocentric vision (with a moving neck) feedback. It is achieved by imitation learning from teleoperation: We propose a VR-based REAL-TIME teleoperation that streams the stereo video observation from the robot camera to the VR device. The robot neck moves as the human head moves, the robot hands move as the human hands move, offering the operator an intuitive experience as the human herself becomes the robot. The devils are all in the details, and how to implement things right: ✅How to perform IK/retargeting for smooth and precise control. ✅How to do all these and also stream stereo video without no latency, all in real time. We released our code here: Active head hardware design: 1/n

Xiaolong Wang

25,572 Aufrufe • vor 2 Jahren

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

50,090 Aufrufe • vor 8 Monaten

Introducing CADGenBench: measure how well AI systems produce engineering-grade 3D parts! While current models can generate 3D parts, they are far from precise enough to build functional parts. We built a benchmark to systematically measure their capabilities on two tasks: 1. Generation from an engineering drawing of a part 2. Editing: given an existing STEP file and a requested change The benchmark is tool-agnostic. It makes no assumptions about how you build the model. You can vary the LLM, and you can vary the environment. Use build123d, Onshape, Autodesk, or a model without an LLM entirely. We open sourced the scoring engine and a reference baseline on top of build123d. A collaboration between Hugging Face and Mecado! Submission space: Code repository:

Introducing CADGenBench: measure how well AI systems produce engineering-grade 3D parts! While current models can generate 3D parts, they are far from precise enough to build functional parts. We built a benchmark to systematically measure their capabilities on two tasks: 1. Generation from an engineering drawing of a part 2. Editing: given an existing STEP file and a requested change The benchmark is tool-agnostic. It makes no assumptions about how you build the model. You can vary the LLM, and you can vary the environment. Use build123d, Onshape, Autodesk, or a model without an LLM entirely. We open sourced the scoring engine and a reference baseline on top of build123d. A collaboration between Hugging Face and Mecado! Submission space: Code repository:

Michael Rabinovich

65,367 Aufrufe • vor 2 Monaten

The next evolution: VLA+ models Just yesterday Microsoft Research released Rho-alpha (ρα) – their first robotics model, built on the Phi family. While most Vision-Language-Action (VLA) models stop at vision and language, Rho-alpha adds: ▪️ Tactile sensing to feel objects during manipulation ▪️ Online learning that lets it improve from human corrections (via teleoperation, 3D mouse or other tools) in real-time even after deployment. Both these sides make adaptability central rather than incidental. Microsoft calls it a VLA+ model, positioning it as an extension beyond what current VLA systems support. ➡️ Today Rho-alpha can control dual-arm robot setups to perform tasks such as: • Manipulating the BusyBox following natural-language instructions • Plug insertion • Toolbox packing and object arrangement with bimanual coordination But to understand why this "plus" matters, we need to understand what came before. Here, we'll take you through the entire landscape of VLA models – Gemini Robotics, π0, SmolVLA, Helix, ACoT-VLA and others:

The next evolution: VLA+ models Just yesterday Microsoft Research released Rho-alpha (ρα) – their first robotics model, built on the Phi family. While most Vision-Language-Action (VLA) models stop at vision and language, Rho-alpha adds: ▪️ Tactile sensing to feel objects during manipulation ▪️ Online learning that lets it improve from human corrections (via teleoperation, 3D mouse or other tools) in real-time even after deployment. Both these sides make adaptability central rather than incidental. Microsoft calls it a VLA+ model, positioning it as an extension beyond what current VLA systems support. ➡️ Today Rho-alpha can control dual-arm robot setups to perform tasks such as: • Manipulating the BusyBox following natural-language instructions • Plug insertion • Toolbox packing and object arrangement with bimanual coordination But to understand why this "plus" matters, we need to understand what came before. Here, we'll take you through the entire landscape of VLA models – Gemini Robotics, π0, SmolVLA, Helix, ACoT-VLA and others:

Turing Post

62,494 Aufrufe • vor 7 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

466,965 Aufrufe • vor 1 Jahr

Open-source magnetic tactile sensor for $5! 🧲 Researchers introduced a magnetic tactile sensor that's low-cost, and easy to fabricate, democratizing tactile sensing for robotics. Operating in unstructured environments like homes and offices requires robots to sense forces during physical interaction. Yet the lack of a versatile, accessible tactile sensor has led to fragmented solutions and often force-unaware, sensorless approaches. Building an eFlesh sensor requires four components: a hobbyist 3D printer, off-the-shelf magnets (less than $5), a CAD model, and a magnetometer circuit board. The sensor is 3D printed with magnets embedded in the middle layer. Based on chosen mechanical properties, magnets displace in response to contact forces, measured by a magnetometer underneath. An open-source design tool converts simple OBJ/STL files into 3D-printable STLs. This enables application-specific sensors for robot hands, grippers, quadruped feet, and more. Slip detection generalizes to unseen objects with 95% accuracy. Visual-tactile control policies improve manipulation by 40% over vision-only baselines, achieving 90% success on precise tasks like plug insertion and credit card swiping. All design files, code, trained models, and conversion tools are openly available. Project page: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Open-source magnetic tactile sensor for $5! 🧲 Researchers introduced a magnetic tactile sensor that's low-cost, and easy to fabricate, democratizing tactile sensing for robotics. Operating in unstructured environments like homes and offices requires robots to sense forces during physical interaction. Yet the lack of a versatile, accessible tactile sensor has led to fragmented solutions and often force-unaware, sensorless approaches. Building an eFlesh sensor requires four components: a hobbyist 3D printer, off-the-shelf magnets (less than $5), a CAD model, and a magnetometer circuit board. The sensor is 3D printed with magnets embedded in the middle layer. Based on chosen mechanical properties, magnets displace in response to contact forces, measured by a magnetometer underneath. An open-source design tool converts simple OBJ/STL files into 3D-printable STLs. This enables application-specific sensors for robot hands, grippers, quadruped feet, and more. Slip detection generalizes to unseen objects with 95% accuracy. Visual-tactile control policies improve manipulation by 40% over vision-only baselines, achieving 90% success on precise tasks like plug insertion and credit card swiping. All design files, code, trained models, and conversion tools are openly available. Project page: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

289,626 Aufrufe • vor 4 Monaten

🚀 Announcing Echo — our new frontier model for 3D world generation. Echo turns a simple text prompt or image into a fully explorable, 3D-consistent world. Instead of disconnected views, the result is a single, coherent spatial representation you can move through freely. This is part of a bigger shift in AI: from generating pixels and tokens to generating spaces. Echo predicts a geometry-grounded 3D scene at metric scale, meaning every novel view, depth map, and interaction comes from the same underlying world — not independent hallucinations. Once generated, the world is interactive in real time. You control the camera, explore from any angle, and render instantly — even on low-end hardware, directly in the browser. High-quality 3D world exploration is no longer gated by expensive equipment. Under the hood, Echo infers a physically grounded 3D representation and converts it into a renderable format. For our web demo, we use 3D Gaussian Splatting (3DGS) for fast, GPU-friendly rendering — but the representation itself is flexible and can be easily adapted. Why this matters: consistent 3D worlds unlock real workflows — digital twins, 3D design, game environments, robotics simulation, and more. From a single photo or a line of text, Echo builds worlds that are reliable, editable, and spatially faithful. Echo also enables scene editing and restyling. Change materials, remove or add objects, explore design variations — all while preserving global 3D consistency. Editing no longer breaks the world. This is only the beginning. Echo is the foundation for future world models with dynamics, physical reasoning, and richer interaction — environments that don’t just look right, but behave right. Explore the generated worlds on our website and sign up for the closed beta. The era of spatial intelligence starts here. 🌍 #Echo #WorldModels #SpatialAI #3DFoundationModels Check it out:

🚀 Announcing Echo — our new frontier model for 3D world generation. Echo turns a simple text prompt or image into a fully explorable, 3D-consistent world. Instead of disconnected views, the result is a single, coherent spatial representation you can move through freely. This is part of a bigger shift in AI: from generating pixels and tokens to generating spaces. Echo predicts a geometry-grounded 3D scene at metric scale, meaning every novel view, depth map, and interaction comes from the same underlying world — not independent hallucinations. Once generated, the world is interactive in real time. You control the camera, explore from any angle, and render instantly — even on low-end hardware, directly in the browser. High-quality 3D world exploration is no longer gated by expensive equipment. Under the hood, Echo infers a physically grounded 3D representation and converts it into a renderable format. For our web demo, we use 3D Gaussian Splatting (3DGS) for fast, GPU-friendly rendering — but the representation itself is flexible and can be easily adapted. Why this matters: consistent 3D worlds unlock real workflows — digital twins, 3D design, game environments, robotics simulation, and more. From a single photo or a line of text, Echo builds worlds that are reliable, editable, and spatially faithful. Echo also enables scene editing and restyling. Change materials, remove or add objects, explore design variations — all while preserving global 3D consistency. Editing no longer breaks the world. This is only the beginning. Echo is the foundation for future world models with dynamics, physical reasoning, and richer interaction — environments that don’t just look right, but behave right. Explore the generated worlds on our website and sign up for the closed beta. The era of spatial intelligence starts here. 🌍 #Echo #WorldModels #SpatialAI #3DFoundationModels Check it out:

SpAItial AI

176,524 Aufrufe • vor 8 Monaten