Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Robots assembling robot brain -- imagine this kind of robustness on every precision manufacturing line! Live demo of GPU rack assembly at #NVIDIAGTC: - end-to-end neural network (Skild Brain) finetuned with little data - memory to perform long horizon task (placing jigs, 16 screwes, removing jigs) - robust to... show more

Deepak Pathak

30,818 subscribers

46,392 просмотров • 6 месяцев назад •via X (Twitter)

Наука и технологии #NVIDIAGTC

Anya Rossi• Live Now

Private livecam show

Комментарии: 13

Фото профиля Jerry Chéng

Jerry Chéng6 месяцев назад

Super cool! But does it also generalize to other tool use? I feel this demo could also be achieved by non-learning methods using the classical robotics stack. Gripper and rigid attachment may not be good in the long run

Фото профиля Deepak Pathak

Deepak Pathak6 месяцев назад

Classical methods are no where close this kind of randomness and require <0.1mm precision sensing. As for transfer, same base model (Skild Brain) is operating across examples in this thread.

Фото профиля Sherif Zaidan

Sherif Zaidan6 месяцев назад

@chris_j_paxton Oh man, universal robot arms and RobotiQ grippers .. feels like 2012 all over again

Фото профиля Deepak Pathak

Deepak Pathak6 месяцев назад

@chris_j_paxton Haha.. unfortunately options are really scarce when you want to deploy something that can last 1000s of hrs without intervention. Do let me know if you have some good alternatives!

Фото профиля JulianSaks

JulianSaks6 месяцев назад

super impressive!

Фото профиля Futuro Martinez

Futuro Martinez6 месяцев назад

Robots assembling the very infrastructure powering AI! This precision and robustness with off-the-shelf components is game-changing for manufacturing scalability.

Фото профиля Decent Cloud

Decent Cloud6 месяцев назад

Robots building GPU racks. The constraint shifted from assembly to allocation.

Фото профиля Pushkar

Pushkar6 месяцев назад

How much real-world finetuning data was needed to reach this level of robustness?

Фото профиля Weijie Wang

Weijie Wang2 месяцев назад

live long-horizon task on off-the-shelf hardware is a genuine strength, but still an n≈1 capability demo—without success rates and disturbance metrics, "robust" is unearned.

Фото профиля 𓅋 𐎫𐎤𐎶 ‎ﷺ

𓅋 𐎫𐎤𐎶 ‎ﷺ6 месяцев назад

What model is this running, was it trained for this specific task?

Фото профиля Deepak Pathak

Deepak Pathak6 месяцев назад

Skild Brain fine-tuned with little domain data. Same base model as all tasks in this thread:

Фото профиля Flexa

Flexa6 месяцев назад

"Finetuned with little domain data" is the key signal. The base model is doing the heavy lifting — the domain data just steers it. This is the same pattern as LLM fine-tuning. The bottleneck shifts from "how much data" to "what distribution of data" covers the long tail.

Фото профиля EDWARD (❖,❖)

EDWARD (❖,❖)6 месяцев назад

This robust robot brain demo looks cool for manufacturing 🦾

Похожие видео

Brain-controlled exoskeletons to train humanoid robots! 🧠 Fourier just presented human tele-operators using brain control interfaces and exoskeletal arms to train humanoid robots on home tasks. The brain control interface is the interesting part. Instead of using a controller or joystick to teleoperate, the operator's movements and intentions are captured more naturally through the exoskeleton and BCI. This means the demonstrations are more fluid, more human-like, and better suited for training robots to perform delicate home tasks. Multiple tele-operators are simultaneously generating training data across multiple robots. This is how you build the dataset needed for eventual full autonomy, without waiting years for it to arrive. This might be the bridge between "robots that work in controlled environments" and "robots that work in homes." Not full autonomy right away, but trusted human intelligence operating through a robot body, getting better with every task completed. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Brain-controlled exoskeletons to train humanoid robots! 🧠 Fourier just presented human tele-operators using brain control interfaces and exoskeletal arms to train humanoid robots on home tasks. The brain control interface is the interesting part. Instead of using a controller or joystick to teleoperate, the operator's movements and intentions are captured more naturally through the exoskeleton and BCI. This means the demonstrations are more fluid, more human-like, and better suited for training robots to perform delicate home tasks. Multiple tele-operators are simultaneously generating training data across multiple robots. This is how you build the dataset needed for eventual full autonomy, without waiting years for it to arrive. This might be the bridge between "robots that work in controlled environments" and "robots that work in homes." Not full autonomy right away, but trusted human intelligence operating through a robot body, getting better with every task completed. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

21,042 просмотров • 8 месяцев назад

NEWS: Humanoid robotics company Figure has released Helix 02, what they claim in their most capable humanoid model yet. "A single neural system that controls the full body directly from pixels, enabling dexterous, long horizon autonomy across an entire room: • Autonomous, long‑horizon loco-manipulation: Helix 02 unloads and reloads a dishwasher across a full-sized kitchen - a four-minute, end-to-end autonomous task that integrates walking, manipulation, and balance with no resets and no human intervention. We believe this is the longest horizon, most complex task completed autonomously by a humanoid robot to date. • All sensors in. All actuators out: Helix 02 connects every onboard sensor - vision, touch, and proprioception - directly to every actuator through a single unified visuomotor neural network. • Human-like whole body control from human data: All results are enabled by System 0, a learned whole‑body controller trained on over 1,000 hours of human motion data and sim‑to‑real reinforcement learning. System 0 replaces 109,504 lines of hand‑engineered C++ with a single neural prior for stable, natural motion. • New classes of dexterity: With Figure 03’s embedded tactile sensing and palm cameras, Helix 02 performs manipulation that was previously out of reach: extracting individual pills, dispensing precise syringe volumes, and singulating small, irregular objects from clutter despite self‑occlusion. Helix 02 is trained on over 1,000 hours of human motion data and integrates vision, touch, and proprioception."

NEWS: Humanoid robotics company Figure has released Helix 02, what they claim in their most capable humanoid model yet. "A single neural system that controls the full body directly from pixels, enabling dexterous, long horizon autonomy across an entire room: • Autonomous, long‑horizon loco-manipulation: Helix 02 unloads and reloads a dishwasher across a full-sized kitchen - a four-minute, end-to-end autonomous task that integrates walking, manipulation, and balance with no resets and no human intervention. We believe this is the longest horizon, most complex task completed autonomously by a humanoid robot to date. • All sensors in. All actuators out: Helix 02 connects every onboard sensor - vision, touch, and proprioception - directly to every actuator through a single unified visuomotor neural network. • Human-like whole body control from human data: All results are enabled by System 0, a learned whole‑body controller trained on over 1,000 hours of human motion data and sim‑to‑real reinforcement learning. System 0 replaces 109,504 lines of hand‑engineered C++ with a single neural prior for stable, natural motion. • New classes of dexterity: With Figure 03’s embedded tactile sensing and palm cameras, Helix 02 performs manipulation that was previously out of reach: extracting individual pills, dispensing precise syringe volumes, and singulating small, irregular objects from clutter despite self‑occlusion. Helix 02 is trained on over 1,000 hours of human motion data and integrates vision, touch, and proprioception."

Sawyer Merritt

624,973 просмотров • 8 месяцев назад

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 просмотров • 1 год назад

Show it a single human demonstration of a long-horizon task and the robot performs it right away. No fine-tuning, no retraining. Skild has released S1, a robot foundation model where you specify the task with a video demonstration instead of a language instruction. Why video, not language Language works for "hand me the mug." It falls apart for anything delicate or long-horizon. You don't learn to fold a fitted sheet from a sentence. So S1 takes a video demo into its context window and translates it to its own body and scene. What it does - Tasks up to 10 minutes, dozens of sequential steps - Unseen tasks: potting plants, pour-over coffee, kit assembly, pancakes. Skild says even the underlying primitives, like flipping a pancake, weren't in pre-training - Recovers from its own mistakes without being trained to - Improves on flawed demos. Demonstrator drops an egg, S1 doesn't make that mistake. It reads the demo as a goal, not a trajectory to copy.

Show it a single human demonstration of a long-horizon task and the robot performs it right away. No fine-tuning, no retraining. Skild has released S1, a robot foundation model where you specify the task with a video demonstration instead of a language instruction. Why video, not language Language works for "hand me the mug." It falls apart for anything delicate or long-horizon. You don't learn to fold a fitted sheet from a sentence. So S1 takes a video demo into its context window and translates it to its own body and scene. What it does - Tasks up to 10 minutes, dozens of sequential steps - Unseen tasks: potting plants, pour-over coffee, kit assembly, pancakes. Skild says even the underlying primitives, like flipping a pancake, weren't in pre-training - Recovers from its own mistakes without being trained to - Improves on flawed demos. Demonstrator drops an egg, S1 doesn't make that mistake. It reads the demo as a goal, not a trajectory to copy.

The Humanoid Hub

19,034 просмотров • 1 месяц назад

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

467,596 просмотров • 1 год назад

What if a dexterous robot could learn a 20+ step chemistry experiment without a single on-robot training demo? Meet TwinDEX: a pair of co-designed, three-finger, nine-DoF dexterous manipulation interface: one wearable for data collection, one for robot deployment. The twinned design shares identical kinematics, contact surfaces, visual appearance, and sensors across collection and deployment — keeping observations and actions aligned end to end. Trained from scratch on only a few hundred wearable demonstrations - with zero on-robot training or intervention data - TwinDEX completed a standardized chemistry experiment involving tool switches, fine force control, and bimanual coordination. Robot-free data showed comparable learning efficiency on the multi-task evaluation, TwinDEX delivered 5.3 times effective throughput than on-robot teleoperation. TwinDEX demonstrates that high-quality robot-free data can fully substitute for on-robot teleoperation data on challenging dexterous tasks — removing the dependency on real-robot hardware that has been the central bottleneck to scaling dexterous manipulation data. This was the proof-it phase. Now comes scale: what emerges at tens of thousands, or millions, of episodes? Watch the demo and read the technical blog: #TwinDEX #Robotics #EmbodiedAI #DexterousManipulation

What if a dexterous robot could learn a 20+ step chemistry experiment without a single on-robot training demo? Meet TwinDEX: a pair of co-designed, three-finger, nine-DoF dexterous manipulation interface: one wearable for data collection, one for robot deployment. The twinned design shares identical kinematics, contact surfaces, visual appearance, and sensors across collection and deployment — keeping observations and actions aligned end to end. Trained from scratch on only a few hundred wearable demonstrations - with zero on-robot training or intervention data - TwinDEX completed a standardized chemistry experiment involving tool switches, fine force control, and bimanual coordination. Robot-free data showed comparable learning efficiency on the multi-task evaluation, TwinDEX delivered 5.3 times effective throughput than on-robot teleoperation. TwinDEX demonstrates that high-quality robot-free data can fully substitute for on-robot teleoperation data on challenging dexterous tasks — removing the dependency on real-robot hardware that has been the central bottleneck to scaling dexterous manipulation data. This was the proof-it phase. Now comes scale: what emerges at tens of thousands, or millions, of episodes? Watch the demo and read the technical blog: #TwinDEX #Robotics #EmbodiedAI #DexterousManipulation

X Square Robot

1,346,370 просмотров • 1 месяц назад

A humanoid robot just unloaded a dishwasher for 67 hours straight. By end of 2026, the same robot will perform surgery. That's not a prediction from a sci-fi writer. That's Brett Adcock. CEO of Figure AI. Said it on camera this month. Here's what makes this insane. Two years ago Figure's robot put a Keurig cup in a coffee maker. That was the breakthrough moment. The whole team celebrated. A coffee cup. Today that same robot runs 67 consecutive hours in a BMW factory. Grabbing packages. Finding barcodes. Placing them. One operation every second. Just a neural network that learned by watching. And Brett just said the hands that do that the same fingers, the same sensors, the same neural net are already capable of surgical precision from a hardware standpoint. "From a hardware perspective in 2026 we'll be able to do what surgeons can do." The AI just has to catch up to the hands. Two years ago: a coffee cup. This year: a factory. Next: the operating table. The robot didn't change, the learning did and it's not stopping.

A humanoid robot just unloaded a dishwasher for 67 hours straight. By end of 2026, the same robot will perform surgery. That's not a prediction from a sci-fi writer. That's Brett Adcock. CEO of Figure AI. Said it on camera this month. Here's what makes this insane. Two years ago Figure's robot put a Keurig cup in a coffee maker. That was the breakthrough moment. The whole team celebrated. A coffee cup. Today that same robot runs 67 consecutive hours in a BMW factory. Grabbing packages. Finding barcodes. Placing them. One operation every second. Just a neural network that learned by watching. And Brett just said the hands that do that the same fingers, the same sensors, the same neural net are already capable of surgical precision from a hardware standpoint. "From a hardware perspective in 2026 we'll be able to do what surgeons can do." The AI just has to catch up to the hands. Two years ago: a coffee cup. This year: a factory. Next: the operating table. The robot didn't change, the learning did and it's not stopping.

Shruti

47,780 просмотров • 5 месяцев назад

China's robots just started building other robots on a factory floor. 44 seconds of footage, no theatrics this time. The G1 is assembling precision motor components on a real production line, running on a new embodied AI model called UnifoLM-X1-0. This isn't a stunt. It's the same playbook every frontier AI lab uses, just with hardware attached. China's robots are running on their own assembly line specifically to generate the motion data needed to train the next generation, a closed loop where the robot's work becomes its own training set. The timing tells you what this is really for. The company behind this is heading toward a mid-2026 IPO, and until now almost every public G1 appearance has been a gala dance routine or a boxing demo. A robot doing repetitive, unglamorous assembly work is a far stronger data point for investors than another choreographed performance. The CEO has been explicit about the bar he's measuring against, what he calls the 80/80 goal, a robot completing 80% of tasks across 80% of real-world environments. One in-house factory line doesn't clear that bar on its own, but it's the first public sign anyone is building toward it instead of performing for cameras.

China's robots just started building other robots on a factory floor. 44 seconds of footage, no theatrics this time. The G1 is assembling precision motor components on a real production line, running on a new embodied AI model called UnifoLM-X1-0. This isn't a stunt. It's the same playbook every frontier AI lab uses, just with hardware attached. China's robots are running on their own assembly line specifically to generate the motion data needed to train the next generation, a closed loop where the robot's work becomes its own training set. The timing tells you what this is really for. The company behind this is heading toward a mid-2026 IPO, and until now almost every public G1 appearance has been a gala dance routine or a boxing demo. A robot doing repetitive, unglamorous assembly work is a far stronger data point for investors than another choreographed performance. The CEO has been explicit about the bar he's measuring against, what he calls the 80/80 goal, a robot completing 80% of tasks across 80% of real-world environments. One in-house factory line doesn't clear that bar on its own, but it's the first public sign anyone is building toward it instead of performing for cameras.

Shruti

26,296 просмотров • 2 месяцев назад

Humanoid robots are starting to build humanoid robots. UBTECH just put a 14,000㎡ humanoid robot factory into operation in Liuzhou, China, with a designed takt time of one humanoid every 10 minutes and planned annual capacity in the tens of thousands. What’s interesting is that humanoid robots are already part of the production process. Cruzr Y1 and Y2 handle depalletizing, palletizing, material feeding and transport, using 3D vision to adapt to shifted boxes and changing pallet patterns. On the assembly side, Walker S2 and Cruzr are produced on the same line alongside cobots, autonomous logistics vehicles and assistive manipulators. The factory also has a 65㎡ automated warehouse capable of storing 112 humanoids, with AGVs, industrial robots and stacker cranes coordinated as one system. Before production, the factory was modeled 1:1 in Siemens Plant Simulation, covering workstations, racks and AGVs. Each robot gets its own SN, with components, batches, assembly parameters and inspection data tracked all the way to delivery. Humanoid robots are now entering automated manufacturing at the 10,000-unit scale. The next milestone may be when they can take part in enough of the assembly, testing and production process to truly build robots themselves.

Humanoid robots are starting to build humanoid robots. UBTECH just put a 14,000㎡ humanoid robot factory into operation in Liuzhou, China, with a designed takt time of one humanoid every 10 minutes and planned annual capacity in the tens of thousands. What’s interesting is that humanoid robots are already part of the production process. Cruzr Y1 and Y2 handle depalletizing, palletizing, material feeding and transport, using 3D vision to adapt to shifted boxes and changing pallet patterns. On the assembly side, Walker S2 and Cruzr are produced on the same line alongside cobots, autonomous logistics vehicles and assistive manipulators. The factory also has a 65㎡ automated warehouse capable of storing 112 humanoids, with AGVs, industrial robots and stacker cranes coordinated as one system. Before production, the factory was modeled 1:1 in Siemens Plant Simulation, covering workstations, racks and AGVs. Each robot gets its own SN, with components, batches, assembly parameters and inspection data tracked all the way to delivery. Humanoid robots are now entering automated manufacturing at the 10,000-unit scale. The next milestone may be when they can take part in enough of the assembly, testing and production process to truly build robots themselves.

CyberRobo

18,047 просмотров • 24 дней назад

🚨🇺🇸FIGURE ROBOT LEARNS TO WALK IN HOURS Turns out you can teach a robot to walk—if you’ve got a few years to burn in a physics simulator (or a few hours of GPU time). Figure’s team trained a neural network using reinforcement learning to mimic natural human locomotion, and the wild part is it worked on the first try with real hardware. No tuning. No fiddling. Just plug and strut. Thanks to sim-to-real transfer and some sneaky domain randomization, the robot didn’t just wobble out of the lab—it walked like it had somewhere to be. Next up: running. Not terrifying at all. Source: Figure

🚨🇺🇸FIGURE ROBOT LEARNS TO WALK IN HOURS Turns out you can teach a robot to walk—if you’ve got a few years to burn in a physics simulator (or a few hours of GPU time). Figure’s team trained a neural network using reinforcement learning to mimic natural human locomotion, and the wild part is it worked on the first try with real hardware. No tuning. No fiddling. Just plug and strut. Thanks to sim-to-real transfer and some sneaky domain randomization, the robot didn’t just wobble out of the lab—it walked like it had somewhere to be. Next up: running. Not terrifying at all. Source: Figure

Mario Nawfal

117,866 просмотров • 1 год назад

Kelly Oubre Jr. was asked about so many Knicks fans being in Philadelphia tonight: "I could say something that could get clickbaited, but I'm not. At the end of the day, we're the show on the court. We're not necessarily focused on the crowd, the things like that. Obviously, we need it. We love it and we love our fans. But at the end of the day, we have to lock in on our task. We have to figure out our gameplan, just have a tight huddle. No matter what, it's loud. It's kind of the same no matter who the crowd is cheering for. It's loud. We have to be focused on the task at hand and just kind of have a short-term memory when it comes to things like that. But at the end of the day, we are home. This is the court that we practice on, we shootaround on and we're comfortable here. We just got to go out there and just perform."

Kelly Oubre Jr. was asked about so many Knicks fans being in Philadelphia tonight: "I could say something that could get clickbaited, but I'm not. At the end of the day, we're the show on the court. We're not necessarily focused on the crowd, the things like that. Obviously, we need it. We love it and we love our fans. But at the end of the day, we have to lock in on our task. We have to figure out our gameplan, just have a tight huddle. No matter what, it's loud. It's kind of the same no matter who the crowd is cheering for. It's loud. We have to be focused on the task at hand and just kind of have a short-term memory when it comes to things like that. But at the end of the day, we are home. This is the court that we practice on, we shootaround on and we're comfortable here. We just got to go out there and just perform."

Knicks Videos

81,963 просмотров • 5 месяцев назад

Introducing FLUX-mimic, a next-generation Video-Action Model for general purpose dexterity, developed in partnership with Black Forest Labs. Late last year we published mimic-video and introduced Video-Action Models (VAM): a new family of robotics foundation models built on top of video generation models. We showed that robot control reduces to visual prediction, and that robot capability is downstream of improvements in video modeling accuracy. The obvious implication was that advances in the video modeling frontier would directly translate to increased capabilities in end-to-end robot learning. FLUX-mimic is that thesis at frontier scale: We've applied our VAM architecture to the strongest video backbone available today, FLUX 3 from Black Forest Labs, and trained it on data from our own robots and wearables. General-purpose dexterity, running on a single GPU on premises. Because the model already understands world dynamics, it needs far fewer demonstrations to learn a new task. This is game-changing for our mission to deploy robots to factory floors, where industrial robot data is scarce and expensive to collect. We're now testing and deploying FLUX-mimic with manufacturing leaders like Audi USA, on complex, multi-step manipulation long considered impossible for conventional automation.

Introducing FLUX-mimic, a next-generation Video-Action Model for general purpose dexterity, developed in partnership with Black Forest Labs. Late last year we published mimic-video and introduced Video-Action Models (VAM): a new family of robotics foundation models built on top of video generation models. We showed that robot control reduces to visual prediction, and that robot capability is downstream of improvements in video modeling accuracy. The obvious implication was that advances in the video modeling frontier would directly translate to increased capabilities in end-to-end robot learning. FLUX-mimic is that thesis at frontier scale: We've applied our VAM architecture to the strongest video backbone available today, FLUX 3 from Black Forest Labs, and trained it on data from our own robots and wearables. General-purpose dexterity, running on a single GPU on premises. Because the model already understands world dynamics, it needs far fewer demonstrations to learn a new task. This is game-changing for our mission to deploy robots to factory floors, where industrial robot data is scarce and expensive to collect. We're now testing and deploying FLUX-mimic with manufacturing leaders like Audi USA, on complex, multi-step manipulation long considered impossible for conventional automation.

mimic

118,965 просмотров • 2 месяцев назад

Robotics has changed dramatically over the last eight years. Ted Xiao has been involved in the cutting edge of robot learning through this period, spending those eight years at Google Brain/Google Deepmind. And he’s identified three eras of robot learning. These eras are: - The Era of Existence Proofs - trying different methods like QT-Opt, on-robot RL - The Era of Foundation Models - transitioning to data collection and clean objectives (i.e. supervised learning) - The Era of Scaling - orders of magnitude more data and larger models, enabling reasoning, long-horizon actions, and cross-embodiment transfer Watch Episode 78 of RoboPapers, with Michael Cho - Rbt/Acc and Jiafei Duan to learn more!

Robotics has changed dramatically over the last eight years. Ted Xiao has been involved in the cutting edge of robot learning through this period, spending those eight years at Google Brain/Google Deepmind. And he’s identified three eras of robot learning. These eras are: - The Era of Existence Proofs - trying different methods like QT-Opt, on-robot RL - The Era of Foundation Models - transitioning to data collection and clean objectives (i.e. supervised learning) - The Era of Scaling - orders of magnitude more data and larger models, enabling reasoning, long-horizon actions, and cross-embodiment transfer Watch Episode 78 of RoboPapers, with Michael Cho - Rbt/Acc and Jiafei Duan to learn more!

RoboPapers

36,838 просмотров • 5 месяцев назад

Here’s the windmill assembly demo we showed at CES 2026 — the one no one saw coming. North executes a fully autonomous, long-horizon dexterous sequence with sustained hand–eye–tactile coordination and assembly-level precision enabled by tactile feedback. It’s also robust to disturbance: you can reposition the objects, and North will still identify them and recover the task. This is powered by CraftNet (VTLA) — using tactile feedback to continuously fine-tune the last-millimeter interaction, enabling reliable execution across 30+ steps. Read more about CraftNet: #Sharpa #SharpaWave #SharpaNorth #CraftNet #System0 #CES2026

Here’s the windmill assembly demo we showed at CES 2026 — the one no one saw coming. North executes a fully autonomous, long-horizon dexterous sequence with sustained hand–eye–tactile coordination and assembly-level precision enabled by tactile feedback. It’s also robust to disturbance: you can reposition the objects, and North will still identify them and recover the task. This is powered by CraftNet (VTLA) — using tactile feedback to continuously fine-tune the last-millimeter interaction, enabling reliable execution across 30+ steps. Read more about CraftNet: #Sharpa #SharpaWave #SharpaNorth #CraftNet #System0 #CES2026

Sharpa

88,304 просмотров • 9 месяцев назад

I don’t have my action cam rigged up right now because I’m with people, but had to pull this and share it. FSD V14.2 will literally drive you off of a completely unmapped construction site using cameras and neural nets. Understand that this is a big grey block of nothing on the map and the car did not have any routing path on the GPS until we made it out to the road at the end. Do you understand the difference yet? Are you getting it? Tesla already ships robots to consumers.

I don’t have my action cam rigged up right now because I’m with people, but had to pull this and share it. FSD V14.2 will literally drive you off of a completely unmapped construction site using cameras and neural nets. Understand that this is a big grey block of nothing on the map and the car did not have any routing path on the GPS until we made it out to the road at the end. Do you understand the difference yet? Are you getting it? Tesla already ships robots to consumers.

Mike P

20,941 просмотров • 9 месяцев назад

$BRAIN is live. the open-source fly-brain bridge has a ticker now, and the ticker is becoming the input for a physical robot. CA: 9LuwgFQemAoV9rgVBBtwbRSxBmamcRRbysEK8yL2pump what’s already done: the neural bridge is working. camera input becomes activity across eight virtual neural populations, and that activity becomes left and right motor commands IMU data feeds movement back into the model. smoothing, speed limits and a 500 ms watchdog are already built in the synthetic demo runs locally today. the ESP32 scaffold is ready for hardware integration the entire project is open source and MIT licensed github: what’s next, in order: the on-chain listener goes live. every $BRAIN transaction becomes a stimulus sent directly into the neural model then the physical robot connects. the wallet address determines which neurons activate, the amount determines the strength of the impulse, and the brain converts that reaction into movement then the livestream. every transaction, neural impulse and physical response visible in real time then user commands. spend $BRAIN to trigger a stronger reflex, request movement, temporarily control the robot, name a neuron or place your name on the stream the loop, plain: a transaction enters the neurons react the robot moves a Proof of Reflex is created with the transaction hash, neural activity map and video of the physical response the current brain is a small fly-inspired simulation. the bridge and local demo work today. the blockchain connection, physical robot and livestream are being built now the internet becomes its sensory organ. the blockchain becomes its nervous system. $BRAIN makes the body move.

$BRAIN is live. the open-source fly-brain bridge has a ticker now, and the ticker is becoming the input for a physical robot. CA: 9LuwgFQemAoV9rgVBBtwbRSxBmamcRRbysEK8yL2pump what’s already done: the neural bridge is working. camera input becomes activity across eight virtual neural populations, and that activity becomes left and right motor commands IMU data feeds movement back into the model. smoothing, speed limits and a 500 ms watchdog are already built in the synthetic demo runs locally today. the ESP32 scaffold is ready for hardware integration the entire project is open source and MIT licensed github: what’s next, in order: the on-chain listener goes live. every $BRAIN transaction becomes a stimulus sent directly into the neural model then the physical robot connects. the wallet address determines which neurons activate, the amount determines the strength of the impulse, and the brain converts that reaction into movement then the livestream. every transaction, neural impulse and physical response visible in real time then user commands. spend $BRAIN to trigger a stronger reflex, request movement, temporarily control the robot, name a neuron or place your name on the stream the loop, plain: a transaction enters the neurons react the robot moves a Proof of Reflex is created with the transaction hash, neural activity map and video of the physical response the current brain is a small fly-inspired simulation. the bridge and local demo work today. the blockchain connection, physical robot and livestream are being built now the internet becomes its sensory organ. the blockchain becomes its nervous system. $BRAIN makes the body move.

Frank

346,227 просмотров • 22 дней назад

Here is a human assembling a BMW engine by hand. Hundreds of components. Micron tolerances. Dozens of hours. So what's better here—a human or a robot❓ 🤖 Robots win at: - Repeatability. Same torque, same angle, every time. - Speed on defined tasks. Honda automates piston installation at 9,000/day. - No fatigue. Shift 47 is identical to shift 1. 👨‍🏭 Humans win at: - Feeling what they can't see. Half of engine assembly is blind. - Force intuition. Knowing "seated" from "about to crack." - Compliant manipulation. That instinctive wiggle to get a part to drop in. - Handling variance. Mercedes pulled robots off the S-Class line because they couldn't adapt to option variability. - Deformables. Wiring harnesses, gaskets, hoses—nothing predictable. - Recovery. When something's wrong, humans adapt. Robots halt. ✅ Right now, the answer is both. Robots handle the repetitive subassemblies. Humans handle everything that requires judgment, feeling, and adaptation. The day a humanoid can do this end-to-end is the day manipulation is solved.

Here is a human assembling a BMW engine by hand. Hundreds of components. Micron tolerances. Dozens of hours. So what's better here—a human or a robot❓ 🤖 Robots win at: - Repeatability. Same torque, same angle, every time. - Speed on defined tasks. Honda automates piston installation at 9,000/day. - No fatigue. Shift 47 is identical to shift 1. 👨‍🏭 Humans win at: - Feeling what they can't see. Half of engine assembly is blind. - Force intuition. Knowing "seated" from "about to crack." - Compliant manipulation. That instinctive wiggle to get a part to drop in. - Handling variance. Mercedes pulled robots off the S-Class line because they couldn't adapt to option variability. - Deformables. Wiring harnesses, gaskets, hoses—nothing predictable. - Recovery. When something's wrong, humans adapt. Robots halt. ✅ Right now, the answer is both. Robots handle the repetitive subassemblies. Humans handle everything that requires judgment, feeling, and adaptation. The day a humanoid can do this end-to-end is the day manipulation is solved.

Jack 🤖

379,855 просмотров • 10 месяцев назад

TESLA’S OPTIMUS JUST CARRIED A $10,000+ GPU LIKE IT WAS A GROCERY BAG! The robot is carrying NVIDIA’s RTX PRO 6000 Blackwell workstation GPU while walking and keeping the package stable in its hands. That sounds simple until you remember what’s happening underneath: 2 legs balancing, 2 arms controlling the load, cameras tracking the environment, and motors constantly making tiny corrections. A task that looks completely ordinary to a human is actually a serious test of real-world robotic control. This is where humanoid robots get interesting. They don’t need to learn 1,000 completely different machines if the same pair of hands can eventually pick up a GPU, move a box, open a door, carry equipment, and perform dozens of other tasks. One general-purpose robot could potentially replace an entire collection of specialized machines. The hardware is impressive, but the bigger story is versatility. The moment a robot can reliably handle arbitrary objects instead of one predefined task, the economics of automation start to change.

TESLA’S OPTIMUS JUST CARRIED A $10,000+ GPU LIKE IT WAS A GROCERY BAG! The robot is carrying NVIDIA’s RTX PRO 6000 Blackwell workstation GPU while walking and keeping the package stable in its hands. That sounds simple until you remember what’s happening underneath: 2 legs balancing, 2 arms controlling the load, cameras tracking the environment, and motors constantly making tiny corrections. A task that looks completely ordinary to a human is actually a serious test of real-world robotic control. This is where humanoid robots get interesting. They don’t need to learn 1,000 completely different machines if the same pair of hands can eventually pick up a GPU, move a box, open a door, carry equipment, and perform dozens of other tasks. One general-purpose robot could potentially replace an entire collection of specialized machines. The hardware is impressive, but the bigger story is versatility. The moment a robot can reliably handle arbitrary objects instead of one predefined task, the economics of automation start to change.

Future Memo

36,252 просмотров • 20 дней назад