#Robots

NVIDIA announces new Isaac tools for agents as Agile Robots unveils Agile ONE SN, a reference humanoid for robotics research. 🦾 Built on NVIDIA Isaac GR00T and powered by Jetson Thor, Agile ONE SN gives researchers and developers an integrated platform to develop, train and validate humanoid skills. The Isaac expansion also introduces new datasets, a runtime, and robot learning framework updates to accelerate development from data to deployment. Read the #NVIDIAGTC announcement:

We’re excited to announce the official release of NVIDIA Isaac Lab 3.0. 🎉 Isaac Lab 3.0 brings a new modular and lightweight architecture along with major new capabilities across the robot learning workflow: multi-physics and multi-rendering support, training at scale, and streamlined deployment on real hardware. With Newton integration, the new OVRTX renderer, RL post-training, and LEAPP deployment, there’s a lot to explore. Here’s a look at what’s new 🧵 (1/7)

Toyota takes on Tesla and China in humanoid robot race

Has GPT-6-Astra Achieved Embodied Intelligence? Across 12 benchmarks, access to training demonstrations and up to 10× the execution budget push several tasks close to 100% success. How does this translate to real robots? Watch our demos across 8 robot configurations to see its capabilities and remaining challenges. Paper: Project & demos: GitHub:

Austin’s robot population is growing

Interesting So they are assuming annual production of 550k humanoids and total DRAM demand of 21.5 PB Even if annual production was 20X (11M), total DRAM demand will be 430 PB or 0.43 EB DRAM production in 2026 will be around 48EB So, 21.5PB is around 0.04% of 2026 DRAM production

Counterpoint Research expects DRAM demand from humanoid robots to surge 20x by 2030, from around 8,300 terabits in 2026 to 172,000 terabits. The firm also expects memory per robot to more than double, from 19GB today to 39GB by 2030.

@zephyr_z9 Silly analysis. Robot RAM needs will exceed everything else on Earth, as you need the robots to be on physical substates in order to shape atoms. Super intelligence needs will vastly exceed that, but ~100% of SI will be deep space compute. Obviously.

Here are the most important highlights from @Tesla's Q3 2026 earnings call: • Tesla's 2027 Megapack production capacity is nearly sold out. • Tesla is aiming to produce 5 million vehicles annually by 2030. • Customers are increasingly visiting Tesla showrooms specifically to experience FSD rather than shopping for a particular vehicle model. • Tesla said more than 75% of new North American vehicle deliveries included a paid FSD subscription signup, despite every vehicle already coming with a free one-month trial. • FSD financial disclosure: Tesla plans to disclose additional FSD financial and operating metrics as the business scales, including subscription revenue and churn rates. • Elon said Tesla is hoping for EU-wide FSD approval within a few months and would be surprised if approval takes longer than Q1 2027, although the timing depends on regulators. • Unsupervised FSD for customer vehicles: Tesla expects to enable some customer-owned vehicles to participate in its autonomous Robotaxi network around Q2 2027, subject to fleet-management and geofencing requirements. • Tesla is developing its own AI-based mapping and navigation system that will rely on intelligence gathered from its vehicle fleet rather than traditional third-party mapping systems. The system will support conversational voice commands, personalized routing preferences, improved parking, and more intelligent destination handling. •  Tesla plans to launch Cybercab service in Dallas in November 2026, followed by additional cities. • Tesla plans to progressively extend robotaxi operating hours, with the goal of reaching 24/7 service by the end of 2026. Tesla expects to expand service to highways, higher-speed roads, additional weather conditions, and nighttime operations by the end of 2026. • Elon expects Robotaxi to begin making a noticeable contribution to Tesla's financial results around Q2 2027, potentially becoming a significant business by Q3 or Q4 2027. • Tesla is targeting the first deliveries of Cybercabs to fleet owner customers by the end of 2026, but Elon said the vast majority of Cybercabs will be Tesla-owned and operated as transportation-as-a-service vehicles rather than sold directly to consumers. • Tesla has started early Optimus production. Elon said: "The growth of Optimus is an S curve, but the initial curve is a shallow slope. There's still some things we need to tweak about the design. There's still quite a lot of work to do with the AI in the robot. I would expect that the AI, the electromechanical design, reliability, usage life, expect these issues to be solved by the end of next year." • Elon said most of the equipment for an Optimus production line designed to manufacture approximately 1 million robots annually is already at Fremont. Elon expects it will take at least another year, possibly longer, to reach an annualized production rate of approximately 1 million Optimus robots, equivalent to roughly 20,000 per week. • Elon said Tesla may release a new Optimus V3 demonstration video within one or two months, saying that it should show genuinely AI-driven tasks rather than preprogrammed movements. • Tesla is targeting mid-2027 to integrate its AI5 chip into Optimus. • Management said Optimus production costs are trending toward the expected manufacturing cost of Cybercab, although the robot's design is still being refined. • Tesla hopes to manufacture its first experimental semiconductor chips around May or June 2027. Tesla is targeting useful chip production at volume in Terafab around Q3 2028. • Management expects higher Megapack average selling prices and further improvements in gross margins (currently 33%) as production expands.

If you’re looking at the new Roadster and all you can talk about is sales and margins, you’re missing the whole fuckin point. As a long-term Tesla investor, I’m looking at what Tesla and SpaceX can build together. Now imagine what could happen if the two companies actually merged. This car is a glimpse of that potential. Some of you are so busy counting cars that you can’t see what’s right in front of you. Think bigger.

Optimus and Cybercab will be the vast majority of Tesla’s value. The only things that will scale to incredible size (K2 and beyond) economically are super intelligence, robots, solar power and mass to space. And, of course, if you have the first two, the rest will follow. But Roadster is super cool, so we’re doing it anyway.

Tank Man can hardly contain his rage as this robot oil manager spins out of control in Mega Man: Dual Override! After reaching his boiling point from the lack of being acknowledged, he'll have plenty in store for this slippery skirmish! #MegaManDO 🛢️

Unitree Introducing: H2-X Kung Fu Legend Guardian🤖 Powerful performance to serve and guard you, with a maximum endurance of approximately 8 hours. Please use the robot safely and in a friendly manner.

Where Automation comes to life.. #futureofautomation #Robots #warehousing

How to learn Physical Intelligence's robot brain without working at Physical Intelligence: PI was in talks this year to raise $1B at an $11B valuation, and its brain, π0, is open source three of its co-founders teach the ideas inside it on YouTube for $0: Sergey Levine, Chelsea Finn and Quan Vuong here's the order I'd watch them in: 1: copy a human first Levine's Berkeley CS 285, lecture 2 (Spring 2026), is supervised learning of behaviors: a robot copies human demos, then drifts off course after its first small mistake. that drift is why ALOHA and π0 predict a whole chunk of motion at once instead of one step at a time 2: learn the math π0 moves with Stanford AA203, lecture 15 (Spring 2026). an hour in, it explains flow matching, the same math π0 uses to turn a decision into smooth arm motion 3: make it better after the demos run out Chelsea Finn's Stanford CS224R: lecture 16 is RL for real robots, and lecture 17 is a guest lecture from Ashish Kumar, who led AI for Tesla Optimus 4: see the whole brain in one talk Quan Vuong's guest talk in ETH's Robot Learning course is about π0.7. 13 minutes in, it folds a shirt on a robot arm that had no folding data of its own then stop watching and start running it: - openpi, PI's own code and π0 weights, 14.4K stars on GitHub - LeRobot by Hugging Face, 28.5K stars, which fine-tunes π0 on your own demos - a $122 SO-101 arm to record those demos on main insight: the lab in talks at $11 billion is built on three ideas stacked on top of each other: copy a human, move smoothly with flow matching, and let a vision-language model decide what to do next all three are free lectures now. the order is the only part nobody hands you P.S. my Optimus-style robot brain guide below turns this order into an 8-week plan

Tesla won't sell you Optimus or show you its brain. but every humanoid brain built in public uses the same 3 layers, and all 3 are free to build at home for $0 to $230. last night I got a humanoid waving on my laptop. here's the 8-week path to its brain

A lighter and more flexible German industrial humanoid robot Germany’s NEURA Robotics has unveiled its 4NE1 Gen 4, featuring a slimmer design, a black textile outer layer and a 20% weight reduction over its predecessor. Based on the previous generation’s 80 kg weight, Gen 4 would weigh around 64 kg, though its exact weight remains unconfirmed. Its brain is Neuron AI, NEURA’s own Physical AI foundation model. It combines vision, touch, force, sound and language to help robots understand their surroundings and execute tasks. Through Neuraverse, learned skills can be shared across connected robots. NEURA is also building a 2,300-square-meter RoboGym with TUM to bridge the gap between training and deployment. Robots learn industrial skills through real-world data, human demonstrations and simulation... while NEURA advances real-world projects with partners such as Schaeffler,which plans to deploy thousands of its humanoids by 2035, and HD Hyundai Samho, where robots are being tested for shipbuilding.

100% open-source robotic arm! 🦾 @seeedstudio released reBot-DevArm, a robotic arm project lowering the barrier to learning robotics... or how people call it these days: physical AI. Everything is open-sourced. Hardware blueprints include sheet metal and 3D printed parts. Detailed BOM covers every screw with purchase links. Software and algorithms include Python SDK, ROS1/2, Isaac Sim, and LeRobot. Let's have a look at robot specs. So it has 1.5 kg payload, 650 mm max reach, 4.5 kg weight, and less than 0.2 mm repeatability with 6 DoF plus gripper. This is true open source for robotics. When every screw, CAD file, motor driver, and algorithm is freely available, desktop robotic arms become accessible to students, researchers, and developers worldwide. ‼️ Start your career in robotics today: We need more roboticists than ever! ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

OpenAI-Backed 1X Launches U.S. Factory to Build 10,000 Home #Robots a Year by @IntEngineering #Robotics #FutureTech #Tech #Technology

20 Open-Source Humanoid Robots You Can Build Yourself | From $50 to $900 What if you could build your own humanoid robot without spending thousands of dollars? In this video, I explore 20 open-source and DIY humanoid robotics projects, ranked from 1 to 20. The list covers a wide range of budgets, from around $50 to $900 for an upper-body build. Some projects provide official 3D models, CAD files or assembled STEP models. Others do not have a complete, ready-to-use 3D model matching the robot shown in the video. For some entries, I arranged available components or created my own visual representations. The 3D models for projects 2, 8, 10 and 6 require verification. Some visuals may be reconstructions or assemblies rather than official models of the original robots. Please check the project documentation before treating any visual as an exact representation of the hardware. The 20 projects featured 1. RoboPrime | Around $60–100 A low-cost robotics project with an open-source design. 2. Open-source DIY humanoid builds A general category covering custom humanoid builds assembled from different hardware components. The humanoid shown is a generic 3D model created by me, not an official robot model. 3. Qmini | Price unconfirmed A compact robot project from Unitree Robotics. 4. Zowi | Around €100 A small educational robot associated with BQ. 5. Otto Humanoid / Otto DIY | From around $50 A DIY robotics platform. The final cost depends on the selected components and configuration. 6. Kayra | Price unconfirmed The visual combines official leg designs with additional components positioned by me. 7. PLEN2 | Around $499 for the kit A compact humanoid robot with published CAD files and walking-related software. 8. Nova Desktop Humanoid | Around $150 A desktop humanoid project by Daniel Hany. I positioned the available head and arm components myself in the 3D scene. 9. Bambot | Around $300 A low-cost robotics project with open-source design resources. 10. Rapiro | Around $390 A small humanoid robot designed by Shota Ishiwatari. The 3D reconstruction shown in the video requires verification against the original model. 11. Valera | Around $400 An open-source humanoid robotics project. 12. Reachy Mini Lite | Around $399 A compact robotics platform developed by Pollen Robotics in collaboration with Hugging Face. 13. Open Duck Mini | Under $400 An open-source bipedal robot project for robotics enthusiasts and developers. 14. MicroDuck | Around $399 A Pollen Robotics project with reinforcement-learning resources. 15. ROBOTIS MINI | Around $540 A compact humanoid platform associated with ROBOTIS components. 16. Halley | Around $500–800 An open-source robot project with 3D design resources. 17. Microban | Around $550–600 A robotics project associated with the Rhoban team at the University of Bordeaux. 18. Endeavor II | Around $600 A humanoid project with an assembled STEP model. 19. XLeRobot | Around $660 An open-source robotics project with accessible hardware designs. 20. InMoov | Around $900 for the upper body Created by Gaël Langevin, InMoov is a well-known 3D-printed humanoid robotics project. What I noticed while preparing this video The projects vary in price and in the amount of design information available. Some provide official 3D models and detailed mechanical files, while others need additional work to create a visual representation. The models shown for projects 2, 6, 8 and 10 need to be distinguished from verified official 3D assets. Check the original repositories and documentation for confirmation.

This completely changes how robots can learn in simulation. NVIDIA just launched SimReady Central with 3,500+ simulation ready OpenUSD assets free and ungated on Hugging Face. ➝ Robot arms grippers hospital equipment warehouse gear kitchens. Lightwheel is the largest contributor to the library. Its assets are physics validated with measured mass and inertia tuned collision geometry and joint limits that hold under contact. These are more than pretty meshes. ➝ A robot in simulation learns from how objects collide how heavy they are what moves and how far. If an object looks right but behaves wrong it teaches the policy the wrong lesson.

Elon Musk just put a timeline on how long working humans have before the robots take over. "AI will probably increase the global economy by 20 to 30%. That's my rough estimate. Meaning, on the order of 20 to 30 trillion per year." That is like a second US economy showing up from nowhere, every year. And his whole point is that machines make it, not people. He also gave a date for the digital side: "AI will be able to do anything digital, anything that does not require the shaping of atoms by hand, by the end of next year." The end of next year. On software in particular: "It's going to be impossible for a human to compete in writing software with AI." So every coder, every analyst and every job done on a screen has a 12 to 18 month clock running, according to the richest person on Earth. And that only covers what runs on electricity. The physical side is bigger: "There will be at least a billion robots in 10 years, and each will produce at least five times the output of a human." "The billion humanoid robots will be more productive than all humans combined." A billion machines outworking all 8 billion of us, with a 10-year deadline. He even called that number conservative, the one he would put serious money on. So how do you build a billion of anything that quickly? Robots that build robots that build robots. Slow at first, then straight up. That is how a few thousand turns into a billion. He also shared his formula for how useful a robot is: the quality of the AI software, multiplied by the quality of the AI chip, multiplied by how dexterous the hands are. Software and chips are what Musk and Nvidia have spent ten years making exponential. The hand is the third factor, and robotics has been stuck on it for four decades. A machine can beat the best chess player alive and still fail to grab an unfamiliar object off a cluttered table the way a toddler does. Grip, pressure, touch, how hard to squeeze. That barely moved while everything digital took off. So the part he calls hard is the part his entire 10-year number rests on. It comes down to one variable, the slowest one in the whole system: the robot hand. The brain is close to solved. The fingers need to catch up this decade, at planetary scale. To be fair, this is where the biggest pile of money on Earth is aimed right now. Tesla and a dozen rivals are pouring everything into it. His software predictions tend to land. His predictions about atoms tend to slip, because physical things always take longer than digital ones. He has also made this robot call before with very different numbers: 10 billion by 2040 in one talk, five per person in another. So the real question: do you think robot hands get solved within 10 years? That single bet is the heart of his prediction. The rest has already come true. Interested in the latest with ai/tech and business? Follow @GeniusGTX and never fall behind. I track ChatGPT, Claude, and how every new agent is changing how we work and think, then hand you the pure signals.

Oh god f*ck, is this even possible? Watch what happens when a physical robot running the Grok Brain Engine gets brutally punted across the room, falls, and instantly recalibrates its entire existence in real time. Look at the live telemetry feed as the hardware recovers from absolute chaos: • Instant Fall Detection: The moment the robot goes down, the neural graph recalculates vector telemetry, friction indexes, and balance locks in milliseconds. • Autonomous Motor Recovery: Instead of a hardcoded reset, thousands of nodes fire simultaneously to execute quick recovery steps and stabilize the chassis. • Complex Object Manipulation: The same brain seamlessly transitions from emergency recovery to tracking, approaching, kneeling, and precisely picking up a tiny plush toy off the floor. We are watching mechanical units develop a real-time instinct to survive, adapt, and execute complex physical tasks without a single line of hardcoded human instruction. Bookmark this before physical AI becomes completely normalized across the real world.

I see how some people can watch flashy robot demos on the timeline and think with a shiver that the robot apocalypse is around the corner, but reality check: There is no robot in the world that can change the batteries in a jack-o-lantern. If you have a robot friendly door handle it might be able to walk outside, pick it up, and carry it back inside. The robot could open the drawer with your batteries in it, but then things get tough. Opening the package of AA batteries? Extracting a precision screwdriver from its case? Unscrewing four tiny screws that require a hard press and twist? Popping batteries out of the case? Yes, you could build and train a specialist robot to do that today, but there is a very long way to go to replace the full range of human capabilities. Near term doomer scenarios should stop at “secretly controlling humanity” instead of “wiping humanity out” 😀

@ID_AA_Carmack These are real benchmarks

@ID_AA_Carmack How about the robot is looking down while assembling a Swiss watch with one hand and then, without looking up, catches an oblique 100mph baseball with the other hand? I also like the idea of juggling chainsaws on fire while riding a unicycle.

The 9 Rare Earth Elements Behind Modern Robotics Robot motors and actuators run on Nd and Pr magnets, with Dy and Tb added for heat resistance. Er and Yb enable LiDAR, and Eu, Tb and Y are used in displays. A humanoid robot can use dozens of these motors, and over 97% of humanoid shipments in the first half of 2026 came from Chinese manufacturers. Created in collaboration with @sprott. #RareEarths #CriticalMinerals #Robotics #HumanoidRobots