Loading video...

Video Failed to Load

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

🚀UPDATE: #Xiaomi humanoid robots advance in EV factory operations Self-tapping nut loading reached 98% success rate, just 1% behind human after one quarter of optimization. First long-duration flexible manipulation tasks reach 90% success in panel sorting and bin folding👇

智东西China AI News

1,125 subscribers

14,297 views • 1 month ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Xiaomi's humanoid robotic intern has achieved over 90% success across multiple factory tasks at its EV factory in Beijing. After four months at the self-piercing nut station, the robot reached 98% task accuracy. It is now handling center console side-cover sorting and returnable box folding, with both tasks already achieving over 90% accuracy.

Xiaomi's humanoid robotic intern has achieved over 90% success across multiple factory tasks at its EV factory in Beijing. After four months at the self-piercing nut station, the robot reached 98% task accuracy. It is now handling center console side-cover sorting and returnable box folding, with both tasks already achieving over 90% accuracy.

Space and Technology

46,938 views • 27 days ago

Xiaomi’s robot just got handed two harder jobs at its EV factory. 🤖 It already hits 98% on self-tapping nut loading. Now it’s sorting large, flexible center-console panels and folding totes — both at 90%. The bigger step isn’t the numbers. It’s the control stack: whole-body reach, bimanual regrasping, active compliance, precise fingertip control, and multi-robot coordination. From one station to several. From doing a task to coordinating with other robots. That’s the real shift.

Xiaomi’s robot just got handed two harder jobs at its EV factory. 🤖 It already hits 98% on self-tapping nut loading. Now it’s sorting large, flexible center-console panels and folding totes — both at 90%. The bigger step isn’t the numbers. It’s the control stack: whole-body reach, bimanual regrasping, active compliance, precise fingertip control, and multi-robot coordination. From one station to several. From doing a task to coordinating with other robots. That’s the real shift.

RoboHub🤖

29,686 views • 1 month ago

Today we’re introducing KinetIQ Ascend: our reinforcement learning approach designed to reach 99.9% manipulation reliability at human speed and beyond. It enables our robots to learn from real production tasks and improve through trial and error: • Machine feeding: throughput +42% • Item handover: throughput +85%, success from 80% to 98% • Tote handling: throughput more than doubled, success from 78% to 99% More in the blog post below 👇

Today we’re introducing KinetIQ Ascend: our reinforcement learning approach designed to reach 99.9% manipulation reliability at human speed and beyond. It enables our robots to learn from real production tasks and improve through trial and error: • Machine feeding: throughput +42% • Item handover: throughput +85%, success from 80% to 98% • Tote handling: throughput more than doubled, success from 78% to 99% More in the blog post below 👇

Humanoid

42,922 views • 2 months ago

Generalist AI has just broken the "impossible triangle" of speed, reliability, and intelligence in robotics. They’ve pushed their embodied foundation model forward in a big way: GEN-1 hits “Mastery” on simple physical tasks,99% average success rate, and it can improvise when things go wrong. Smooth, natural, fast(factory owners must be very interested)👇 >Folding boxes: 200 runs at 99% success in ~12 seconds ( ~34s before) >Packing phones, folding T-shirts, servicing robot vacuums,hundreds of reps with almost zero failures > Intelligent improvisation, handling uncertain tasks >Needs only about 1 hour of robot-specific data per task This feels like a real step toward humanoid robots that are actually reliable enough for factories, warehouse or homes. Physical AGI is accelerating

Generalist AI has just broken the "impossible triangle" of speed, reliability, and intelligence in robotics. They’ve pushed their embodied foundation model forward in a big way: GEN-1 hits “Mastery” on simple physical tasks,99% average success rate, and it can improvise when things go wrong. Smooth, natural, fast(factory owners must be very interested)👇 >Folding boxes: 200 runs at 99% success in ~12 seconds ( ~34s before) >Packing phones, folding T-shirts, servicing robot vacuums,hundreds of reps with almost zero failures > Intelligent improvisation, handling uncertain tasks >Needs only about 1 hour of robot-specific data per task This feels like a real step toward humanoid robots that are actually reliable enough for factories, warehouse or homes. Physical AGI is accelerating

CyberRobo

25,519 views • 4 months ago

Robots that actually WORK in real factories. Without endless retraining. > Just 20 minutes of data. > 99.4% success rate. > 108 motors soldered in 5+ hours straight. Sub-0.6 mm precision on messy, deformable (!) cables. This hybrid “learning-augmented” system adds neural brains + 3D safety monitoring to ordinary cobots… and suddenly complex factory tasks become reliable, human-safe, and stupidly fast. This is for all of you who are into manufacturing. Huge credit to robotics engineer Yunho Kim Yunho Kim 📍Paper:

Robots that actually WORK in real factories. Without endless retraining. > Just 20 minutes of data. > 99.4% success rate. > 108 motors soldered in 5+ hours straight. Sub-0.6 mm precision on messy, deformable (!) cables. This hybrid “learning-augmented” system adds neural brains + 3D safety monitoring to ordinary cobots… and suddenly complex factory tasks become reliable, human-safe, and stupidly fast. This is for all of you who are into manufacturing. Huge credit to robotics engineer Yunho Kim Yunho Kim 📍Paper:

Ilir Aliu

16,956 views • 3 months ago

Robots from Humanoid passed real factory test at Siemens! 😮‍💨 Humanoid from the UK just completed a successful proof of concept with Siemens, deploying their HMND 01 wheeled humanoid robot in a real working factory. The task was simple but important: pick totes from a storage stack, transport them to a conveyor, and place them at the pickup point for human operators. Repeat until the stack is empty. The robot achieved 60 tote moves per hour, handled two different tote sizes, ran autonomously for more than 30 minutes at a time, stayed operational for over eight hours, and maintained over 90% success rate on pick and place tasks. This matters because it's not a demo video. It's a real deployment in Siemens' Electronics Factory in Erlangen, handling actual production work, measured against real operational metrics. The test ran in two phases. First, Humanoid built a physical twin in-house to test and optimize the system. Then they deployed on-site at Siemens for two weeks, where the robot operated in the real production environment. We can see a wheeled humanoid robot in action! So, now the ultimate question: wheels or legs? 👀 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Robots from Humanoid passed real factory test at Siemens! 😮‍💨 Humanoid from the UK just completed a successful proof of concept with Siemens, deploying their HMND 01 wheeled humanoid robot in a real working factory. The task was simple but important: pick totes from a storage stack, transport them to a conveyor, and place them at the pickup point for human operators. Repeat until the stack is empty. The robot achieved 60 tote moves per hour, handled two different tote sizes, ran autonomously for more than 30 minutes at a time, stayed operational for over eight hours, and maintained over 90% success rate on pick and place tasks. This matters because it's not a demo video. It's a real deployment in Siemens' Electronics Factory in Erlangen, handling actual production work, measured against real operational metrics. The test ran in two phases. First, Humanoid built a physical twin in-house to test and optimize the system. Then they deployed on-site at Siemens for two weeks, where the robot operated in the real production environment. We can see a wheeled humanoid robot in action! So, now the ultimate question: wheels or legs? 👀 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

21,993 views • 6 months ago

A new member joins the European humanoid robot team. German company Agile Robotics has launched its first industrial humanoid robot, Agile One, featuring intuitive human-robot interaction, dexterous hands (for grasping small screws and touching the screen), and AI-driven operation trained in the real world. It performs tasks such as material collection, handling, picking and placing, loading and unloading machine tools, tool handling, and precision operations. The emergence of humanoid robots by industrial robot companies is a natural progression; they are designed for areas inaccessible to traditional robotic arms and industrial robots, extending operational systems and enhancing business collaboration capabilities. Agile One will be manufactured in Germany in early 2026 and deployed for on-site training at customers to enhance its AI model capabilities. Soon, Europe will enter its second wave of automation driven by intelligent technology.

A new member joins the European humanoid robot team. German company Agile Robotics has launched its first industrial humanoid robot, Agile One, featuring intuitive human-robot interaction, dexterous hands (for grasping small screws and touching the screen), and AI-driven operation trained in the real world. It performs tasks such as material collection, handling, picking and placing, loading and unloading machine tools, tool handling, and precision operations. The emergence of humanoid robots by industrial robot companies is a natural progression; they are designed for areas inaccessible to traditional robotic arms and industrial robots, extending operational systems and enhancing business collaboration capabilities. Agile One will be manufactured in Germany in early 2026 and deployed for on-site training at customers to enhance its AI model capabilities. Soon, Europe will enter its second wave of automation driven by intelligent technology.

CyberRobo

72,567 views • 9 months ago

NVIDIA did it again.. they trained a GPT that generates human movement instead of text.. next-token prediction, but the tokens are body motions. it's called GPC and it hit a 99.98% success rate reproducing a massive corpus of motion clips.. cartwheels, vaults, flips, all of it. → 99.98% success rate reproducing motion clips → learns cartwheels, flips, vaulting from raw data → adapts to new tasks with <1% extra parameters studios spend 6-12 months and millions of dollars building custom controllers for one character in one game. GPC does it in one forward pass. Pixar, Ubisoft, and every mocap studio on the planet just felt a chill down their spine. the same next-token trick that gave us ChatGPT is now doing parkour inside a physics engine.

NVIDIA did it again.. they trained a GPT that generates human movement instead of text.. next-token prediction, but the tokens are body motions. it's called GPC and it hit a 99.98% success rate reproducing a massive corpus of motion clips.. cartwheels, vaults, flips, all of it. → 99.98% success rate reproducing motion clips → learns cartwheels, flips, vaulting from raw data → adapts to new tasks with <1% extra parameters studios spend 6-12 months and millions of dollars building custom controllers for one character in one game. GPC does it in one forward pass. Pixar, Ubisoft, and every mocap studio on the planet just felt a chill down their spine. the same next-token trick that gave us ChatGPT is now doing parkour inside a physics engine.

How To Prompt

101,820 views • 1 month ago

500 humanoid robots replacing humans in high-voltage operations What does that look like? Steel against steel,instead of flesh and blood. This marks a turning point for China’s State Grid, shifting from human-based maintenance to autonomous operations. This year, State Grid announced plans to procure 8,500 embodied AI robots, with a total budget of RMB 6.8 billion (~$1 billion). These robots will be deployed across four major scenarios: power inspection, live-line operations, emergency response, and warehouse logistics,covering more than 600 specific task scenarios. Among them, humanoid robots for live-line operations are the most expensive and strategically critical: 500 units with a budget of RMB 2.5 billion (~$370 million). They will be deployed in distribution network live-line work and ultra-high-voltage (UHV) projects, replacing humans in high-risk tasks. Workers will transition into supervisory roles, ready to take over remotely when needed. As early as last year, State Grid had already validated the feasibility of humanoid robots for substation inspection. Tienkung can autonomously perform inspection tasks at a State Grid substation in Beijing. Of course, suppliers are not limited to X-Humanoid,players like Unitree, AGIBOT, DeepRobotics, UBTECH, and Fourier are all involved. These 500 humanoid robots will also collaborate with 5,000 inspection quadruped robots and 3,000 dual-arm wheeled robots for indoor substation maintenance,together forming an intelligent, automated, and collaborative network for autonomous grid operations. What does this change? According to State Grid, each embodied AI unit can save RMB 500,000 to 800,000 (~$70,000–$110,000) in annual labor costs, with a payback period of around 2–3 years. Inspection efficiency increases by 5x, fault response time is reduced by 60%, and power supply reliability improves by 0.5 percentage points. More importantly, over 90% of human exposure to high-risk operations can be eliminated, reducing safety incidents by 80%. At another level, for humanoid robot companies, the center of R&D and iteration is shifting to the customer site. Real-world physical interaction becomes the fastest feedback loop,accelerating innovation and evolution. And 8,500 units are just the beginning of scaled deployment. Based on current plans, embodied AI robots will cover 30% of key areas in State Grid by 2026, 80% of high-risk operation scenarios by 2027, and enable fully autonomous operations by 2030. The demand roadmap is clear: define use cases ->deploy at scale->improve models and robots->expand further. 8,500… 50,000… 100,000… But remember,power grids are just one part of China’s vast infrastructure system. The experience of autonomous robotic operations here can be replicated across other sectors, such as broader energy systems. That, in itself, is another story. P.S.The video shows Tienkung 1.0 autonomously performing substation inspection tasks (2025).

500 humanoid robots replacing humans in high-voltage operations What does that look like? Steel against steel,instead of flesh and blood. This marks a turning point for China’s State Grid, shifting from human-based maintenance to autonomous operations. This year, State Grid announced plans to procure 8,500 embodied AI robots, with a total budget of RMB 6.8 billion (~$1 billion). These robots will be deployed across four major scenarios: power inspection, live-line operations, emergency response, and warehouse logistics,covering more than 600 specific task scenarios. Among them, humanoid robots for live-line operations are the most expensive and strategically critical: 500 units with a budget of RMB 2.5 billion (~$370 million). They will be deployed in distribution network live-line work and ultra-high-voltage (UHV) projects, replacing humans in high-risk tasks. Workers will transition into supervisory roles, ready to take over remotely when needed. As early as last year, State Grid had already validated the feasibility of humanoid robots for substation inspection. Tienkung can autonomously perform inspection tasks at a State Grid substation in Beijing. Of course, suppliers are not limited to X-Humanoid,players like Unitree, AGIBOT, DeepRobotics, UBTECH, and Fourier are all involved. These 500 humanoid robots will also collaborate with 5,000 inspection quadruped robots and 3,000 dual-arm wheeled robots for indoor substation maintenance,together forming an intelligent, automated, and collaborative network for autonomous grid operations. What does this change? According to State Grid, each embodied AI unit can save RMB 500,000 to 800,000 (~$70,000–$110,000) in annual labor costs, with a payback period of around 2–3 years. Inspection efficiency increases by 5x, fault response time is reduced by 60%, and power supply reliability improves by 0.5 percentage points. More importantly, over 90% of human exposure to high-risk operations can be eliminated, reducing safety incidents by 80%. At another level, for humanoid robot companies, the center of R&D and iteration is shifting to the customer site. Real-world physical interaction becomes the fastest feedback loop,accelerating innovation and evolution. And 8,500 units are just the beginning of scaled deployment. Based on current plans, embodied AI robots will cover 30% of key areas in State Grid by 2026, 80% of high-risk operation scenarios by 2027, and enable fully autonomous operations by 2030. The demand roadmap is clear: define use cases ->deploy at scale->improve models and robots->expand further. 8,500… 50,000… 100,000… But remember,power grids are just one part of China’s vast infrastructure system. The experience of autonomous robotic operations here can be replicated across other sectors, such as broader energy systems. That, in itself, is another story. P.S.The video shows Tienkung 1.0 autonomously performing substation inspection tasks (2025).

CyberRobo

46,782 views • 4 months ago

Holy…S😳 Xiaomi's New CyberOne is so human-like Although this update features a bionic hand, I was immediately drawn to it. Let's look at the changes in the hand： It can handle industrial precision tasks like turning screws, plus delicate operations such as pinching feathers and throwing balloons. Behind the performance: >Volume cut by 60%:now almost identical in size/shape to a real human hand >Big leap in degrees of freedom (+50% total, +83% active:22-27DOF) >Full-palm tactile sensors over 8200 mm² for precise grip even without vision >150,000+ grip cycles durability (61-hour test) And a major innovation:Smart bionic sweat gland cooling: evaporates water for ~10W active heat dissipation Using tactile gloves to capture real human data, they’re training smoother, human-like grasps with imitation and reinforcement learning. Elon has said that humanoid hands and true AI are the most difficult aspects of building humanoid robots. It seems that Xiaomi is also getting close.

Holy…S😳 Xiaomi's New CyberOne is so human-like Although this update features a bionic hand, I was immediately drawn to it. Let's look at the changes in the hand： It can handle industrial precision tasks like turning screws, plus delicate operations such as pinching feathers and throwing balloons. Behind the performance: >Volume cut by 60%:now almost identical in size/shape to a real human hand >Big leap in degrees of freedom (+50% total, +83% active:22-27DOF) >Full-palm tactile sensors over 8200 mm² for precise grip even without vision >150,000+ grip cycles durability (61-hour test) And a major innovation:Smart bionic sweat gland cooling: evaporates water for ~10W active heat dissipation Using tactile gloves to capture real human data, they’re training smoother, human-like grasps with imitation and reinforcement learning. Elon has said that humanoid hands and true AI are the most difficult aspects of building humanoid robots. It seems that Xiaomi is also getting close.

CyberRobo

111,430 views • 5 months ago

Jensen just announced NVIDIA’s Isaac GR00T N1.5 and GR00T-Dreams blueprint at COMPUTEX 2025: ⦿ Isaac GR00T N1.5 is the first update to NVIDIA’s open, generalized, fully customizable foundation model for humanoid reasoning and skills. ⦿ “Human demonstrations aren’t scalable — limited by the number of hours in a day,” said Jensen. The GR00T-Dreams blueprint enables the generation of vast synthetic motion data from single images, accelerating robot behavior learning via compressed action tokens. ⦿ Synthetic training data generated by GR00T-Dreams was used to develop GR00T N1.5 in just 36 hours — what would have taken nearly three months without the blueprint. ⦿ This update significantly improves the foundation model’s success rate for common material handling and manufacturing tasks. GR00T N1.5 can be deployed on Jetson Thor, launching later this year.

Jensen just announced NVIDIA’s Isaac GR00T N1.5 and GR00T-Dreams blueprint at COMPUTEX 2025: ⦿ Isaac GR00T N1.5 is the first update to NVIDIA’s open, generalized, fully customizable foundation model for humanoid reasoning and skills. ⦿ “Human demonstrations aren’t scalable — limited by the number of hours in a day,” said Jensen. The GR00T-Dreams blueprint enables the generation of vast synthetic motion data from single images, accelerating robot behavior learning via compressed action tokens. ⦿ Synthetic training data generated by GR00T-Dreams was used to develop GR00T N1.5 in just 36 hours — what would have taken nearly three months without the blueprint. ⦿ This update significantly improves the foundation model’s success rate for common material handling and manufacturing tasks. GR00T N1.5 can be deployed on Jetson Thor, launching later this year.

The Humanoid Hub

69,557 views • 1 year ago

I've long wondered if we can make a humanoid robot do a 𝘄𝗮𝗹𝗹𝗳𝗹𝗶𝗽 - and we just made it happen by leveraging 𝗢𝗺𝗻𝗶𝗥𝗲𝘁𝗮𝗿𝗴𝗲𝘁 with BeyondMimic tracking! This came after our original OmniRetarget experiments, with only minor tweaks to RL training: relaxing a termination threshold and removing one reward term. The policy achieved a 𝟱/𝟱 success rate in our real-world experiments, showing the strength of high-quality, interaction-preserving motion retargeting combined with BeyondMimic’s minimal RL tracking. Here is the updated arXiv: (In Sec. V. A)

I've long wondered if we can make a humanoid robot do a 𝘄𝗮𝗹𝗹𝗳𝗹𝗶𝗽 - and we just made it happen by leveraging 𝗢𝗺𝗻𝗶𝗥𝗲𝘁𝗮𝗿𝗴𝗲𝘁 with BeyondMimic tracking! This came after our original OmniRetarget experiments, with only minor tweaks to RL training: relaxing a termination threshold and removing one reward term. The policy achieved a 𝟱/𝟱 success rate in our real-world experiments, showing the strength of high-quality, interaction-preserving motion retargeting combined with BeyondMimic’s minimal RL tracking. Here is the updated arXiv: (In Sec. V. A)

Zhen Wu

1,051,448 views • 10 months ago

OPTIMUS: TESLA’S HUMANOID ROBOT IS ABOUT TO TRANSFORM LABOR & ABUNDANCE Optimus isn’t just a cool prototype—it’s Tesla’s bet on solving the biggest economic constraint of the future: physical labor shortages. Designed from the ground up as a general-purpose humanoid, Optimus will handle unsafe, repetitive, or boring tasks at scale, freeing humans for higher-value work and accelerating abundance across industries. Key impacts already in motion: •Factory deployment — Early units are performing real tasks at Tesla Gigafactories, proving reliability in chaotic, real-world environments. •Dangerous & dull jobs — Welding, heavy lifting, sorting, cleaning hazardous sites—Optimus takes the risk so people don’t have to. •24/7 productivity — No fatigue, no breaks, no unions—constant output that scales with demand. •Home & care applications — Helping the elderly, assisting with disabilities, or handling household chores—turning science fiction into everyday reality. •Economic multiplier — When humanoid robots become cheaper than human labor, costs collapse in manufacturing, logistics, agriculture, and services—unlocking massive productivity gains and lower prices for everyone. From first walking demos to factory trials, Optimus is progressing faster than most realize. It’s the physical embodiment of Tesla’s AI + robotics vision: a world where physical work is abundant, safe, and optional. Which Optimus application excites you most—factory scaling, elder care, or seeing it cook breakfast? The age of humanoid helpers is closer than you think. 🤖⚡🚀

OPTIMUS: TESLA’S HUMANOID ROBOT IS ABOUT TO TRANSFORM LABOR & ABUNDANCE Optimus isn’t just a cool prototype—it’s Tesla’s bet on solving the biggest economic constraint of the future: physical labor shortages. Designed from the ground up as a general-purpose humanoid, Optimus will handle unsafe, repetitive, or boring tasks at scale, freeing humans for higher-value work and accelerating abundance across industries. Key impacts already in motion: •Factory deployment — Early units are performing real tasks at Tesla Gigafactories, proving reliability in chaotic, real-world environments. •Dangerous & dull jobs — Welding, heavy lifting, sorting, cleaning hazardous sites—Optimus takes the risk so people don’t have to. •24/7 productivity — No fatigue, no breaks, no unions—constant output that scales with demand. •Home & care applications — Helping the elderly, assisting with disabilities, or handling household chores—turning science fiction into everyday reality. •Economic multiplier — When humanoid robots become cheaper than human labor, costs collapse in manufacturing, logistics, agriculture, and services—unlocking massive productivity gains and lower prices for everyone. From first walking demos to factory trials, Optimus is progressing faster than most realize. It’s the physical embodiment of Tesla’s AI + robotics vision: a world where physical work is abundant, safe, and optional. Which Optimus application excites you most—factory scaling, elder care, or seeing it cook breakfast? The age of humanoid helpers is closer than you think. 🤖⚡🚀

Tesla Owners Silicon Valley

46,045 views • 6 months ago

🚨 BREAKING: Swiss startup mimic just raised €13.8M to teach robots the one thing they still struggle with human-like dexterity. 🇨🇭 The Zurich-based company is developing physical AI systems that allow robots to learn directly from human demonstrations. Skilled operators wear custom motion-tracking gear on real factory floors, generating data that trains robots to grasp, move, and adapt with human precision. Founded in 2024 as a spin-off from ETH Zurich, mimic pairs AI-driven dexterous robotic hands with standard industrial robot arms, skipping the humanoid form factor in favor of practical, deployable automation. The oversubscribed seed round was led by Elaia with participation from Speedinvest, Founderful, 1st Kind, 10x Founders, 2100 Ventures, and the Sequoia Scout Fund, bringing total funding to over $20M. Their technology is already being piloted with Fortune 500 manufacturers and logistics operators, where robots are learning complex manipulation tasks once reserved for humans. Why it matters: Europe is stepping up in the AI robotics race. By making dexterity scalable, mimic bridges the gap between lab breakthroughs and real factory deployment, where automation still meets its hardest challenges. As CEO Stefan Weirich put it: “We make dexterity deployable at scale, closing the gap between what AI can do in the lab and what factories actually need.” Let's go robotics in Europe! 🔥

🚨 BREAKING: Swiss startup mimic just raised €13.8M to teach robots the one thing they still struggle with human-like dexterity. 🇨🇭 The Zurich-based company is developing physical AI systems that allow robots to learn directly from human demonstrations. Skilled operators wear custom motion-tracking gear on real factory floors, generating data that trains robots to grasp, move, and adapt with human precision. Founded in 2024 as a spin-off from ETH Zurich, mimic pairs AI-driven dexterous robotic hands with standard industrial robot arms, skipping the humanoid form factor in favor of practical, deployable automation. The oversubscribed seed round was led by Elaia with participation from Speedinvest, Founderful, 1st Kind, 10x Founders, 2100 Ventures, and the Sequoia Scout Fund, bringing total funding to over $20M. Their technology is already being piloted with Fortune 500 manufacturers and logistics operators, where robots are learning complex manipulation tasks once reserved for humans. Why it matters: Europe is stepping up in the AI robotics race. By making dexterity scalable, mimic bridges the gap between lab breakthroughs and real factory deployment, where automation still meets its hardest challenges. As CEO Stefan Weirich put it: “We make dexterity deployable at scale, closing the gap between what AI can do in the lab and what factories actually need.” Let's go robotics in Europe! 🔥

Lukas Ziegler

77,152 views • 9 months ago

NVIDIA just announced EgoScale 🤖🧠 NVIDIA Research has uncovered a log-linear scaling law for robot dexterity by pretraining VLA models on over 20,000 hours of egocentric human video This massive dataset is 20 times larger than previous efforts and proves that robot intelligence follows a predictable path: the more human data, the lower the loss The secret is a simple recipe combining large-scale human pretraining with a small amount of aligned human-robot mid-training to bridge the gap In testing, this method boosted the average success rate by 54% on a 22-DoF robotic hand compared to policies built without pretraining EgoScale also enables one-shot task adaptation and works across different hardware, suggesting that human motion is a universal motor prior for robots Website: Paper: Source: NVIDIA Research #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #NVIDIA #EgoScale #GR00T

NVIDIA just announced EgoScale 🤖🧠 NVIDIA Research has uncovered a log-linear scaling law for robot dexterity by pretraining VLA models on over 20,000 hours of egocentric human video This massive dataset is 20 times larger than previous efforts and proves that robot intelligence follows a predictable path: the more human data, the lower the loss The secret is a simple recipe combining large-scale human pretraining with a small amount of aligned human-robot mid-training to bridge the gap In testing, this method boosted the average success rate by 54% on a 22-DoF robotic hand compared to policies built without pretraining EgoScale also enables one-shot task adaptation and works across different hardware, suggesting that human motion is a universal motor prior for robots Website: Paper: Source: NVIDIA Research #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #NVIDIA #EgoScale #GR00T

RoboHub🤖

43,752 views • 5 months ago

The wait is over. Figure 03 is here, "Designed for Helix, the home, and the world at scale." ⦿ Shown testing at home, performing long-horizon tasks fully autonomously (nothing teleoperated). ⦿ A completely redesigned sensory suite and hand system, which is purpose-built to enable Helix, Figure's proprietary VLA AI. ⦿ 9% lighter than Figure 02. ⦿ Features 2kW wireless charging, an improved audio system for voice reasoning, and battery safety advancements. ⦿ Figure established a new supply chain and an entirely new process for manufacturing humanoid robots. ⦿ A new compact camera architecture delivers twice the frame rate, one-quarter the latency, and a 60% wider field of view per camera. ⦿ An embedded palm camera with a wide field of view and low-latency sensing. ⦿ First-generation in-house developed tactile sensors for fingertips, capable of detecting forces as small as three grams of pressure. ⦿ Includes 10 Gbps mmWave data offload capability. ⦿ Better protection against pinch points. Figure's manufacturing facility, BotQ, has a first-generation manufacturing line that will initially be capable of producing up to 12,000 humanoid robots per year, with the goal of producing a total of 100,000 robots over the next four years.

The wait is over. Figure 03 is here, "Designed for Helix, the home, and the world at scale." ⦿ Shown testing at home, performing long-horizon tasks fully autonomously (nothing teleoperated). ⦿ A completely redesigned sensory suite and hand system, which is purpose-built to enable Helix, Figure's proprietary VLA AI. ⦿ 9% lighter than Figure 02. ⦿ Features 2kW wireless charging, an improved audio system for voice reasoning, and battery safety advancements. ⦿ Figure established a new supply chain and an entirely new process for manufacturing humanoid robots. ⦿ A new compact camera architecture delivers twice the frame rate, one-quarter the latency, and a 60% wider field of view per camera. ⦿ An embedded palm camera with a wide field of view and low-latency sensing. ⦿ First-generation in-house developed tactile sensors for fingertips, capable of detecting forces as small as three grams of pressure. ⦿ Includes 10 Gbps mmWave data offload capability. ⦿ Better protection against pinch points. Figure's manufacturing facility, BotQ, has a first-generation manufacturing line that will initially be capable of producing up to 12,000 humanoid robots per year, with the goal of producing a total of 100,000 robots over the next four years.

The Humanoid Hub

169,159 views • 10 months ago

The power of generative models — now embodied in humanoids. Announcing DreamControl –– After a year-long research effort at General Robotics — we present a scalable framework for whole-body humanoid control that fuses diffusion priors with reinforcement learning to unlock real-world scene interaction. Diffusion + RL → natural whole-body skills on real robots. DreamControl enables humanoids to move beyond locomotion demos → performing natural, human-like skills such as –– Picking & lifting objects, Opening drawers & doors, Precise punching, kicking, and jumping, Bimanual manipulation tasks Our key innovation: a diffusion prior over human motion that guides RL, eliminating the need for massive teleoperation datasets, and producing motions that look human while transferring to real hardware. Trained purely in simulation, deployed on the Unitree G1 humanoid, DreamControl policies run in real time, bridging sim-to-real with unprecedented naturalness. We leverage a novel hybrid edge + cloud infrastructure that runs RL-trained policies on the edge backed by powerful AI models running in the cloud This is the next step in General Robotics’ journey toward general-purpose humanoid assistants that interact, adapt, and assist autonomously. Paper: Blog: 1/n

The power of generative models — now embodied in humanoids. Announcing DreamControl –– After a year-long research effort at General Robotics — we present a scalable framework for whole-body humanoid control that fuses diffusion priors with reinforcement learning to unlock real-world scene interaction. Diffusion + RL → natural whole-body skills on real robots. DreamControl enables humanoids to move beyond locomotion demos → performing natural, human-like skills such as –– Picking & lifting objects, Opening drawers & doors, Precise punching, kicking, and jumping, Bimanual manipulation tasks Our key innovation: a diffusion prior over human motion that guides RL, eliminating the need for massive teleoperation datasets, and producing motions that look human while transferring to real hardware. Trained purely in simulation, deployed on the Unitree G1 humanoid, DreamControl policies run in real time, bridging sim-to-real with unprecedented naturalness. We leverage a novel hybrid edge + cloud infrastructure that runs RL-trained policies on the edge backed by powerful AI models running in the cloud This is the next step in General Robotics’ journey toward general-purpose humanoid assistants that interact, adapt, and assist autonomously. Paper: Blog: 1/n

Ashish Kapoor

118,321 views • 11 months ago

How humanoids move from "Kung Fu" mode to "Blue-collar" work 🤖🦾 Unitree founder Xingxing Wang just dropped some major truth bombs about the reality of robot labor at the HEIS meeting in Beijing Extreme mobility like the "Kung Fu" bot at the Gala is the prerequisite—stable movement is the foundation for robots to actually start working Unitree has been deploying humanoids at NIO and Geely since 2024 with a 100% success rate on single tasks like simple parts assembly The real challenge is long-sequence complex tasks with 20+ steps—success rates struggle here especially when handling tiny parts Whole-body teleoperation is the key technical breakthrough for capturing high-quality physical world data to solve the industry's data scarcity Standardization is now the priority—setting unified task metrics, evaluation systems, and safety baselines to move beyond one-off lab innovations In 2025 Unitree shipped over 5,500 humanoids with more than 6,500 units coming off the production line leading the global market in volume The end goal is clear—Unitree wants humanoids working in its own factories to manufacture even more robots, closing the automation loop Source: HEIS Annual Meeting #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #Unitree #SmartManufacturing #Automation

How humanoids move from "Kung Fu" mode to "Blue-collar" work 🤖🦾 Unitree founder Xingxing Wang just dropped some major truth bombs about the reality of robot labor at the HEIS meeting in Beijing Extreme mobility like the "Kung Fu" bot at the Gala is the prerequisite—stable movement is the foundation for robots to actually start working Unitree has been deploying humanoids at NIO and Geely since 2024 with a 100% success rate on single tasks like simple parts assembly The real challenge is long-sequence complex tasks with 20+ steps—success rates struggle here especially when handling tiny parts Whole-body teleoperation is the key technical breakthrough for capturing high-quality physical world data to solve the industry's data scarcity Standardization is now the priority—setting unified task metrics, evaluation systems, and safety baselines to move beyond one-off lab innovations In 2025 Unitree shipped over 5,500 humanoids with more than 6,500 units coming off the production line leading the global market in volume The end goal is clear—Unitree wants humanoids working in its own factories to manufacture even more robots, closing the automation loop Source: HEIS Annual Meeting #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #Unitree #SmartManufacturing #Automation

RoboHub🤖

31,755 views • 5 months ago

How well do egocentric human videos scale? Apparently, really well. Dyna Robotics has introduced Dyna-2, a world-action model (WAM) pre-trained on more than 1 million hours of real egocentric human video. It learns by jointly predicting the next video frames and the actions, and that video prediction (world modeling) is what carries learning from humans to robots. • More human video predictably improves action prediction. • Task accuracy improves across embodiments (bimanual arms and a humanoid) with zero robot data in pre-training. • Post-trained on just a few hours of robot data, mean task performance rose 20% to 53% as human pre-training scaled 1k to 1M hours. WAM vs VLA, matched apple-to-apple: the WAM hit 1.55x the VLA's success rate and won 65% of head-to-heads. Deployed zero-shot at unseen customer sites, it passed 87% vs the VLA's 46%.

How well do egocentric human videos scale? Apparently, really well. Dyna Robotics has introduced Dyna-2, a world-action model (WAM) pre-trained on more than 1 million hours of real egocentric human video. It learns by jointly predicting the next video frames and the actions, and that video prediction (world modeling) is what carries learning from humans to robots. • More human video predictably improves action prediction. • Task accuracy improves across embodiments (bimanual arms and a humanoid) with zero robot data in pre-training. • Post-trained on just a few hours of robot data, mean task performance rose 20% to 53% as human pre-training scaled 1k to 1M hours. WAM vs VLA, matched apple-to-apple: the WAM hit 1.55x the VLA's success rate and won 65% of head-to-heads. Deployed zero-shot at unseen customer sites, it passed 87% vs the VLA's 46%.

The Humanoid Hub

15,034 views • 16 days ago