Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

Robotics is almost solved! Pun intended. Sunday published a "Solve" standard. They declared the scope (garment types, lighting, surfaces, starting configurations) and adaptation cost (zero—no fine-tuning per home) before evaluating. The same model weights run everywhere. This might be the differentiation between a good demo and reliable work of...

lukas_m_ziegler's profile picture

Lukas Ziegler

57,908 subscribers

36,518 görüntüleme • 17 gün önce •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

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 →
0:56

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

20,898 görüntüleme • 6 ay önce

Are you kidding me??? It grasps multiple objects with different ways, all at once with… a single hand??? No pauses. 1x speed. GENE-26.5 is Genesis AI’s robotics-native multimodal foundation model. It’s trained on 200,000+ hours of real human hand data (motion, force, touch) and runs on a 54-DoF bimanual system: Scaling that human data 4x lifted real-robot success rates from 16.6% to 65.6% on long-horizon dexterous tasks!! Same model weights, zero fine-tuning for this exact sequence. You know those tiny coordinated movements you do without thinking…? Robots couldn’t reliably do that before. Now they can. Today. This is the video you’ll send to my friends, outside of our bubble, when they say “robots are still just demos.” Congrats to the entire team around Zhou Xian! Credit: Seen at Zu Wang (Zu Wang) Genesis official announcement for the full story + longer demo: ——— Weekly robotics and AI insights. Subscribe free:
0:20

Are you kidding me??? It grasps multiple objects with different ways, all at once with… a single hand??? No pauses. 1x speed. GENE-26.5 is Genesis AI’s robotics-native multimodal foundation model. It’s trained on 200,000+ hours of real human hand data (motion, force, touch) and runs on a 54-DoF bimanual system: Scaling that human data 4x lifted real-robot success rates from 16.6% to 65.6% on long-horizon dexterous tasks!! Same model weights, zero fine-tuning for this exact sequence. You know those tiny coordinated movements you do without thinking…? Robots couldn’t reliably do that before. Now they can. Today. This is the video you’ll send to my friends, outside of our bubble, when they say “robots are still just demos.” Congrats to the entire team around Zhou Xian! Credit: Seen at Zu Wang (Zu Wang) Genesis official announcement for the full story + longer demo: ——— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

204,299 görüntüleme • 2 ay önce

This is not a 2050 warehouse. This is a modern air hub of Amazon. Amazon uses robots to help sort and ship billions of packages each year. At Amazon Air, the company's main air hub in Kentucky, hundreds of self-charging robots zoom across the warehouse floor to organize orders. These robots, called “drives,” follow small stickers on the floor to know where to go and when to speed up or slow down. Amazon has more than ONE MILLION robots in its facilities, and it’s even deploying more new types of robots that can sort items of different shapes and sizes. P.S. Can you guess how many robots they had back in 2013? ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:23

This is not a 2050 warehouse. This is a modern air hub of Amazon. Amazon uses robots to help sort and ship billions of packages each year. At Amazon Air, the company's main air hub in Kentucky, hundreds of self-charging robots zoom across the warehouse floor to organize orders. These robots, called “drives,” follow small stickers on the floor to know where to go and when to speed up or slow down. Amazon has more than ONE MILLION robots in its facilities, and it’s even deploying more new types of robots that can sort items of different shapes and sizes. P.S. Can you guess how many robots they had back in 2013? ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

178,090 görüntüleme • 6 gün önce

Sunday Robotics CEO Tony Zhao believes home robots only need to master a handful of chores before they become indispensable. "What is the minimum number of tasks that a robot needs to do to justify its own existence so that people love having it in their homes? I actually think the list is pretty short. The tasks are pretty repetitive. You just need to learn it and just keep doing that over and over again, like folding shirts or loading dishwashers."
1:39

Sunday Robotics CEO Tony Zhao believes home robots only need to master a handful of chores before they become indispensable. "What is the minimum number of tasks that a robot needs to do to justify its own existence so that people love having it in their homes? I actually think the list is pretty short. The tasks are pretty repetitive. You just need to learn it and just keep doing that over and over again, like folding shirts or loading dishwashers."

TBPN

64,079 görüntüleme • 13 gün önce

Robot simulation is getting a big upgrade! 🔥 Normally, building a simulation environment, a kitchen, warehouse, office takes days or weeks. You have to model the geometry, add textures, set lighting, generate collision meshes… all before a robot can even interact with it. Marble by World Labs removes most of that work. You type a text prompt or give it an image, and it generates a full 3D scene with depth, lighting, and a collider mesh you can export into engines like MuJoCo. That means researchers can quickly create lots of different environments to train robots on, which is exactly what domain randomization needs. If we want robots to learn faster, we need to create worlds faster. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:59

Robot simulation is getting a big upgrade! 🔥 Normally, building a simulation environment, a kitchen, warehouse, office takes days or weeks. You have to model the geometry, add textures, set lighting, generate collision meshes… all before a robot can even interact with it. Marble by World Labs removes most of that work. You type a text prompt or give it an image, and it generates a full 3D scene with depth, lighting, and a collider mesh you can export into engines like MuJoCo. That means researchers can quickly create lots of different environments to train robots on, which is exactly what domain randomization needs. If we want robots to learn faster, we need to create worlds faster. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

15,498 görüntüleme • 8 ay önce

Foundation models are enough to solve robotics! Unfortunately, this is not true. We keep hearing that Vision-Language-Action (VLA) models struggle because of the gap between static training and the dynamic real world. A German startup (Sereact) just released a solution that bridges this gap perfectly. They are introducing a new paradigm called Interactive RL Policy Patching. It's a distributed framework that allows robots to self-learn from human corrections without needing full retraining. When a robot fails, a human operator provides a brief "patch" or demonstration. The system then uses online off-policy reinforcement learning to update the behavior instantly. This is powered by a massive foundation model trained on hundreds of millions of interactions from over 100 deployed robot stations. The best part is the distributed parameter synchronization... When one robot learns a fix, the update is published fleet-wide... meaning the entire swarm gets smarter from a single human intervention. They are already proving this on complex manipulation tasks like shoe unboxing and screw sorting, drastically reducing the data needed to handle edge cases. Real-world environments are unforgiving, and I love seeing systems that can actually adapt on the fly! 📍 More info:
0:38

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

Ilir Aliu

18,210 görüntüleme • 6 ay önce

Robot trained entirely in simulation! 🤯 Sudo AI just released #sudo R1. This could be the moment the data bottleneck in robotics finally breaks, a major step forward for getting simulation-trained robots to actually work in the real world. #sudo R1 is a manipulation foundation model that reliably picks objects it has never seen: glass, fabric, metal, irregular shapes. All with zero real-world training data. No tele-op. No demos. No manual labeling. Pure simulation. And picking isn't just any skill, it's the foundation. Almost everything we want robots to do starts with being able to reliably pick something up. Data collection is the bottleneck holding the whole field back. Every new task means hours of tele-op and armies of labelers. If simulation alone can actually deliver production-grade picking, that bottleneck starts to disappear. Really exciting to see! If the results hold, the path to general-purpose robots just got a whole lot shorter. 👇 Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:59

Robot trained entirely in simulation! 🤯 Sudo AI just released #sudo R1. This could be the moment the data bottleneck in robotics finally breaks, a major step forward for getting simulation-trained robots to actually work in the real world. #sudo R1 is a manipulation foundation model that reliably picks objects it has never seen: glass, fabric, metal, irregular shapes. All with zero real-world training data. No tele-op. No demos. No manual labeling. Pure simulation. And picking isn't just any skill, it's the foundation. Almost everything we want robots to do starts with being able to reliably pick something up. Data collection is the bottleneck holding the whole field back. Every new task means hours of tele-op and armies of labelers. If simulation alone can actually deliver production-grade picking, that bottleneck starts to disappear. Really exciting to see! If the results hold, the path to general-purpose robots just got a whole lot shorter. 👇 Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

25,901 görüntüleme • 3 ay önce

Sunday Robotics raised $165M to deploy their home robot Memo, but the more interesting story is how they got here. Instead of starting with a robot, they started by putting sensor-equipped gloves into real homes to capture how people actually do chores like dishes, laundry, and cleaning. That data is now what powers Memo. Here’s my conversation with Co-founder & CEO Tony Zhao. 1:34 Why does the world need Sunday? 2:07 What does Sunday get right? 2:37 Hardest part of building a home robot? 3:12 What does the future of robotics look like? 4:54 What is Memo? 5:23 Biggest misconception about robotics? 5:54 What does a successful launch look like? 6:29 $1B valuation pre-launch. How do you handle it? 8:12 What do people underestimate about Sunday? 10:34 Favorite consumer hardware product Tony Zhao Sunday
14:39

Sunday Robotics raised $165M to deploy their home robot Memo, but the more interesting story is how they got here. Instead of starting with a robot, they started by putting sensor-equipped gloves into real homes to capture how people actually do chores like dishes, laundry, and cleaning. That data is now what powers Memo. Here’s my conversation with Co-founder & CEO Tony Zhao. 1:34 Why does the world need Sunday? 2:07 What does Sunday get right? 2:37 Hardest part of building a home robot? 3:12 What does the future of robotics look like? 4:54 What is Memo? 5:23 Biggest misconception about robotics? 5:54 What does a successful launch look like? 6:29 $1B valuation pre-launch. How do you handle it? 8:12 What do people underestimate about Sunday? 10:34 Favorite consumer hardware product Tony Zhao Sunday

Ritwik Pavan

25,691 görüntüleme • 3 ay önce

Robots might learn better from video than from language! 📼 Most Vision-Language-Action (VLA) models learn what to do from text, but still struggle with how things move in the real world. That makes them data-hungry and slow to train. mimic video takes a different route. Instead of grounding robot control in text, it grounds it in video, using large pre-trained video models that already capture physical motion and dynamics. The idea is straightforward: let the video model handle “what will happen next,” and let a smaller control model focus only on turning that visual plan into robot actions. The result is big gains in practice. Robots trained this way need 10× less data, converge twice as fast, and perform better on both simulated benchmarks and real bimanual manipulation tasks. If robots can “imagine” motion using video, control becomes a much simpler problem. Shoutout to Jonas Pai, Liam Achenbach, Oier Mees, Elvis Nava and the rest of the team! Here's the project page: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:47

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

Lukas Ziegler

50,063 görüntüleme • 7 ay önce

We might be solving the wrong problem in robotics. That’s what this makes clear. UMI → Universal Manipulation Interface A simple $400 gripper that lets you teach robots by demonstration. You hold it like a tool. Show the task. The robot learns. No teleoperation. No expensive hardware. No robot-specific data. Stanford open-sourced everything → hardware, code, datasets. What stands out to me is the bottleneck. Not algorithms. Data. Teleoperation → ~35 demos/hour UMI → ~111 demos/hour And the data transfers across robots → UR5, Franka, others. The design is surprisingly practical: → GoPro fisheye lens (155° FOV) + mirrors for depth → SLAM + IMU for precise 6DoF tracking → latency matching for dynamic tasks → diffusion policies for multimodal actions Then it scales. Cheng Chi takes this further with Sunday Robotics (with Tony Zhao). A $200 glove → deployed in 500+ homes → ~10 million real-world interactions. Not lab data. Real human behavior. Their robot learns dishes, laundry, espresso → with zero robot-specific data. This is where the shift becomes obvious. From training robots in controlled environments → to learning directly from humans at scale So here’s the real question: Will robotics be unlocked by better models… or by unlocking data? #ArtificialIntelligence #Robotics #AI #Innovation #FutureOfWork
0:20

We might be solving the wrong problem in robotics. That’s what this makes clear. UMI → Universal Manipulation Interface A simple $400 gripper that lets you teach robots by demonstration. You hold it like a tool. Show the task. The robot learns. No teleoperation. No expensive hardware. No robot-specific data. Stanford open-sourced everything → hardware, code, datasets. What stands out to me is the bottleneck. Not algorithms. Data. Teleoperation → ~35 demos/hour UMI → ~111 demos/hour And the data transfers across robots → UR5, Franka, others. The design is surprisingly practical: → GoPro fisheye lens (155° FOV) + mirrors for depth → SLAM + IMU for precise 6DoF tracking → latency matching for dynamic tasks → diffusion policies for multimodal actions Then it scales. Cheng Chi takes this further with Sunday Robotics (with Tony Zhao). A $200 glove → deployed in 500+ homes → ~10 million real-world interactions. Not lab data. Real human behavior. Their robot learns dishes, laundry, espresso → with zero robot-specific data. This is where the shift becomes obvious. From training robots in controlled environments → to learning directly from humans at scale So here’s the real question: Will robotics be unlocked by better models… or by unlocking data? #ArtificialIntelligence #Robotics #AI #Innovation #FutureOfWork

Pascal Bornet

186,223 görüntüleme • 3 ay önce

This guy connected a computer vision model to dual robotic manipulators on his desk and the system now folds shirts in 47 seconds per garment without any human intervention after loading Automated laundry folding is one of those problems that sounds trivial until you realize fabric has no rigid structure and every wrinkle changes the optimal fold path You need the robot to detect garment boundaries through visual segmentation, identify sleeve edges and collar positions on randomly oriented fabric, generate dynamic reference coordinates that shift with garment size, synchronize two independent robotic arms to pull opposing fabric edges without tearing, and execute all of this without a conveyor belt or fixed staging area Most people assume you need a commercial folding machine or at least a rigid frame to hold clothes in place This guy just bolted two robot arms to a workbench, ran a Flask server with a Laundrobot vision library, and built a preset selection interface that handles nine garment types The setup was minimal: a Python backend processing camera frames, a segmentation model running inference locally, two manipulators with soft grippers, and a heads-up display showing red and blue anchor points overlaid on live fabric The system scans the garment, the vision pipeline outputs coordinates like 284.262 and 965.262, the dashboard waits for a RUN command, and the arms fold the item in two geometric steps The robot picks up shirts, pants, towels, and socks from any position on the desk with zero calibration and zero pre-staging It is the same principle robotic pick-and-place systems use in factories but instead of metal parts it is handling deformable textiles that compress and slide unpredictably The arms have no concept of what clean laundry means to a human They think they are executing waypoint trajectories but the output is getting transformed into neatly stacked garments that take zero cognitive load from the operator If a household generates 14 loads of laundry per month and folding takes eleven minutes per load this is how you reclaim 154 minutes without outsourcing or spending four figures on hardware This is the cleanest domestic automation I have seen: one desk, two arms, one camera, and between them a folding operation that runs while you do anything else
0:51

This guy connected a computer vision model to dual robotic manipulators on his desk and the system now folds shirts in 47 seconds per garment without any human intervention after loading Automated laundry folding is one of those problems that sounds trivial until you realize fabric has no rigid structure and every wrinkle changes the optimal fold path You need the robot to detect garment boundaries through visual segmentation, identify sleeve edges and collar positions on randomly oriented fabric, generate dynamic reference coordinates that shift with garment size, synchronize two independent robotic arms to pull opposing fabric edges without tearing, and execute all of this without a conveyor belt or fixed staging area Most people assume you need a commercial folding machine or at least a rigid frame to hold clothes in place This guy just bolted two robot arms to a workbench, ran a Flask server with a Laundrobot vision library, and built a preset selection interface that handles nine garment types The setup was minimal: a Python backend processing camera frames, a segmentation model running inference locally, two manipulators with soft grippers, and a heads-up display showing red and blue anchor points overlaid on live fabric The system scans the garment, the vision pipeline outputs coordinates like 284.262 and 965.262, the dashboard waits for a RUN command, and the arms fold the item in two geometric steps The robot picks up shirts, pants, towels, and socks from any position on the desk with zero calibration and zero pre-staging It is the same principle robotic pick-and-place systems use in factories but instead of metal parts it is handling deformable textiles that compress and slide unpredictably The arms have no concept of what clean laundry means to a human They think they are executing waypoint trajectories but the output is getting transformed into neatly stacked garments that take zero cognitive load from the operator If a household generates 14 loads of laundry per month and folding takes eleven minutes per load this is how you reclaim 154 minutes without outsourcing or spending four figures on hardware This is the cleanest domestic automation I have seen: one desk, two arms, one camera, and between them a folding operation that runs while you do anything else

Blaze

24,799 görüntüleme • 2 ay önce

This man does house inspections. But, he is not an inspector for the Builders, he is an inspector that the homebuyer hires to be thorough to document damages and deficiencies. He provides the buyer with a detailed report, photographs, and then the buyer forwards it to the builder to have them make the corrections. Check out this house. 🤯 These houses are in Florida. They are probably priced between $400K-$500k and you know they’re going to advertise them as luxury, brand new homes. What a waste of resources and materials. Crazy how terrible new home construction quality is in this country. It’s worth the money to hire a quality inspector of your own before you purchase a house. Have you ever had any nightmare stories when buying a new home? Something you didn’t catch before you bought it?
3:45

This man does house inspections. But, he is not an inspector for the Builders, he is an inspector that the homebuyer hires to be thorough to document damages and deficiencies. He provides the buyer with a detailed report, photographs, and then the buyer forwards it to the builder to have them make the corrections. Check out this house. 🤯 These houses are in Florida. They are probably priced between $400K-$500k and you know they’re going to advertise them as luxury, brand new homes. What a waste of resources and materials. Crazy how terrible new home construction quality is in this country. It’s worth the money to hire a quality inspector of your own before you purchase a house. Have you ever had any nightmare stories when buying a new home? Something you didn’t catch before you bought it?

👉M-Û-R-Č-H👈

210,506 görüntüleme • 3 ay önce

Robotics is entering the mainstream! 👀 Let's accept this fact. Consumptionism – stonks. Possibilities (HW + SW) – stonks. Demographics - not stonks. Modern robots are often designed to interact with our legacy world, whether it's a household, a factory, or a shipyard. That means they often need to work with existing machines, tools, and spaces. In short: retrofitting. Improving old machines with new technologies without major interference in their operation. Modern way of integrating things. You don't have to change the entire environment to deploy a robot. The robot will adapt, just as human would when freshly hired. The promise is to have an army of robots to help with the world designed for us. I also believe that general purpose robots, are not a direct competition for single purpose machines (like a highly efficient specialized packaging machine, for example). It's like adding a robot to this carriage. Specialized machine could be a car ;) ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:59

Robotics is entering the mainstream! 👀 Let's accept this fact. Consumptionism – stonks. Possibilities (HW + SW) – stonks. Demographics - not stonks. Modern robots are often designed to interact with our legacy world, whether it's a household, a factory, or a shipyard. That means they often need to work with existing machines, tools, and spaces. In short: retrofitting. Improving old machines with new technologies without major interference in their operation. Modern way of integrating things. You don't have to change the entire environment to deploy a robot. The robot will adapt, just as human would when freshly hired. The promise is to have an army of robots to help with the world designed for us. I also believe that general purpose robots, are not a direct competition for single purpose machines (like a highly efficient specialized packaging machine, for example). It's like adding a robot to this carriage. Specialized machine could be a car ;) ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

14,396 görüntüleme • 5 ay önce

A new robot policy just cleaned up a kitchen it had never seen before [watch what happens. paper included ⬇️] Pi-0.5 builds on top of Pi-0 and shows how smart co-training with diverse data can unlock real generalization in the home. It doesn’t just learn from one setup but adapts to many, including homes it’s never seen. What it does ✅ Handles new homes without training in them ✅ Follows complex language instructions ✅ Cleans, places dishes, handles spills ✅ Matches in-home training models using cross-embodiment and web data Robots that understand tasks and adapt to new spaces are finally within reach. More in the blog: Read the paper ⬇️ Physical Intelligence, co-founded by UC Berkeley professor Sergey Levine, is a robotics startup developing general-purpose AI foundation models that enable robots to perform a wide variety of real-world tasks with human-like adaptability, recently raising $400 million to advance this vision.
1:22

A new robot policy just cleaned up a kitchen it had never seen before [watch what happens. paper included ⬇️] Pi-0.5 builds on top of Pi-0 and shows how smart co-training with diverse data can unlock real generalization in the home. It doesn’t just learn from one setup but adapts to many, including homes it’s never seen. What it does ✅ Handles new homes without training in them ✅ Follows complex language instructions ✅ Cleans, places dishes, handles spills ✅ Matches in-home training models using cross-embodiment and web data Robots that understand tasks and adapt to new spaces are finally within reach. More in the blog: Read the paper ⬇️ Physical Intelligence, co-founded by UC Berkeley professor Sergey Levine, is a robotics startup developing general-purpose AI foundation models that enable robots to perform a wide variety of real-world tasks with human-like adaptability, recently raising $400 million to advance this vision.

Ilir Aliu - eu/acc

18,120 görüntüleme • 1 yıl önce

🚨 BREAKING: Microsoft's first robotics foundation model! 🤯 Microsoft just announced Rho-alpha (ρα), their first robotics model derived from the Phi series of vision-language models. Rho-alpha translates natural language commands into control signals for robotic systems performing bimanual manipulation tasks. Commands like "push the green button with the right gripper," "pull out the red wire," "flip the top switch on," or "turn the knob to position 5" get executed directly by dual-arm robots. What makes this different from standard vision-language-action (VLA) models is the additional modalities. Rho-alpha is a VLA+ model that adds tactile sensing to the perceptual mix, with plans to incorporate force feedback. On the learning side, the model is designed to continually improve during deployment by learning from human feedback. The training approach combines trajectories from physical demonstrations and simulated tasks with web-scale visual question answering data. Since teleoperation data is scarce and expensive, Microsoft is using NVIDIA Isaac Sim on Azure to generate physically accurate synthetic datasets via reinforcement learning. These simulated trajectories get combined with commercial and open physical demonstration datasets. The model is currently under evaluation on dual-arm setups and humanoid robots. Microsoft is opening an Early Access Program for organizations interested in evaluating Rho-alpha. Robots that can adapt to dynamic situations and human preferences are more useful in real environments and more trusted by the people operating them. Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:34

🚨 BREAKING: Microsoft's first robotics foundation model! 🤯 Microsoft just announced Rho-alpha (ρα), their first robotics model derived from the Phi series of vision-language models. Rho-alpha translates natural language commands into control signals for robotic systems performing bimanual manipulation tasks. Commands like "push the green button with the right gripper," "pull out the red wire," "flip the top switch on," or "turn the knob to position 5" get executed directly by dual-arm robots. What makes this different from standard vision-language-action (VLA) models is the additional modalities. Rho-alpha is a VLA+ model that adds tactile sensing to the perceptual mix, with plans to incorporate force feedback. On the learning side, the model is designed to continually improve during deployment by learning from human feedback. The training approach combines trajectories from physical demonstrations and simulated tasks with web-scale visual question answering data. Since teleoperation data is scarce and expensive, Microsoft is using NVIDIA Isaac Sim on Azure to generate physically accurate synthetic datasets via reinforcement learning. These simulated trajectories get combined with commercial and open physical demonstration datasets. The model is currently under evaluation on dual-arm setups and humanoid robots. Microsoft is opening an Early Access Program for organizations interested in evaluating Rho-alpha. Robots that can adapt to dynamic situations and human preferences are more useful in real environments and more trusted by the people operating them. Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

60,912 görüntüleme • 6 ay önce

MIT PhD student Alex Zhang reveals the scaling result where a model trained on short tasks generalizes to problems 100x longer for free: "If you're very clever about the design of your harness or how you use the language model, you can almost get scaling gains for free." "If you train a model naively, there's no tricks. It's just the same way you train a model on these RL environments. You just roll it out, and then you just get some reward." "If you train it on only short tasks, like only tasks that are 10,000 tokens long, and then you were to run it on a similar domain, but at a million tokens, or 10 million tokens, or 100,000 tokens, it generalizes really, really well. If you look at it compared to even the base transformer, you get way better generalization properties." "When the model uses an RLM (Recursive Language Model) after it's trained on these short tasks, it will see some kind of trajectory of actions that it does. Between these two problems of different lengths, the RLM learns to see them as almost the same problem." "Token for token, they're almost the same. You can describe it in code. In one code setting, maybe the for loop is a little bigger, but it's the same kind of code and it derives the constants from the data. There's no hard coding, so they literally look the same." alex zhang
2:41

MIT PhD student Alex Zhang reveals the scaling result where a model trained on short tasks generalizes to problems 100x longer for free: "If you're very clever about the design of your harness or how you use the language model, you can almost get scaling gains for free." "If you train a model naively, there's no tricks. It's just the same way you train a model on these RL environments. You just roll it out, and then you just get some reward." "If you train it on only short tasks, like only tasks that are 10,000 tokens long, and then you were to run it on a similar domain, but at a million tokens, or 10 million tokens, or 100,000 tokens, it generalizes really, really well. If you look at it compared to even the base transformer, you get way better generalization properties." "When the model uses an RLM (Recursive Language Model) after it's trained on these short tasks, it will see some kind of trajectory of actions that it does. Between these two problems of different lengths, the RLM learns to see them as almost the same problem." "Token for token, they're almost the same. You can describe it in code. In one code setting, maybe the for loop is a little bigger, but it's the same kind of code and it derives the constants from the data. There's no hard coding, so they literally look the same." alex zhang

MTS

99,784 görüntüleme • 13 gün önce

"One of the very confusing things about the models right now: how to reconcile the fact that they are doing so well on evals. And you look at the evals and you go, 'Those are pretty hard evals.' But the economic impact seems to be dramatically behind. There is [a possible] explanation. Back when people were doing pre-training, the question of what data to train on was answered, because that answer was everything. So you don't have to think if it's going to be this data or that data. When people do RL training, they say, 'Okay, we want to have this kind of RL training for this thing and that kind of RL training for that thing.' You say, 'Hey, I would love our model to do really well when we release it. I want the evals to look great. What would be RL training that could help on this task?' If you combine this with generalization of the models actually being inadequate, that has the potential to explain a lot of what we are seeing, this disconnect between eval performance and actual real-world performance"
2:21

"One of the very confusing things about the models right now: how to reconcile the fact that they are doing so well on evals. And you look at the evals and you go, 'Those are pretty hard evals.' But the economic impact seems to be dramatically behind. There is [a possible] explanation. Back when people were doing pre-training, the question of what data to train on was answered, because that answer was everything. So you don't have to think if it's going to be this data or that data. When people do RL training, they say, 'Okay, we want to have this kind of RL training for this thing and that kind of RL training for that thing.' You say, 'Hey, I would love our model to do really well when we release it. I want the evals to look great. What would be RL training that could help on this task?' If you combine this with generalization of the models actually being inadequate, that has the potential to explain a lot of what we are seeing, this disconnect between eval performance and actual real-world performance"

Dwarkesh Patel

502,249 görüntüleme • 8 ay önce

.Michael Dell 🇺🇸 on what a startup has to do: “When you're creating a new business with new technology and new business model there’s no playbook. There’s no book you can read that says here’s how you do it. And if you hire the people from the existing industry companies and ask them to go do it, they're just going to do it the way they were doing it before. That’s not going to work. So you need a lot of creativity. You need to intuit your way to the answer by experimenting. You come up with a theory and you try it. Did it work? Great. Let’s do 2x that. Let’s do 10x that. Let’s do 100x that. You keep experimenting until you find the answer.” — Peter Thiel has a great line on this in his book Zero to One: “This is what startup has to do: question received ideas and rethink business from scratch.”
1:23

.Michael Dell 🇺🇸 on what a startup has to do: “When you're creating a new business with new technology and new business model there’s no playbook. There’s no book you can read that says here’s how you do it. And if you hire the people from the existing industry companies and ask them to go do it, they're just going to do it the way they were doing it before. That’s not going to work. So you need a lot of creativity. You need to intuit your way to the answer by experimenting. You come up with a theory and you try it. Did it work? Great. Let’s do 2x that. Let’s do 10x that. Let’s do 100x that. You keep experimenting until you find the answer.” — Peter Thiel has a great line on this in his book Zero to One: “This is what startup has to do: question received ideas and rethink business from scratch.”

David Senra

26,343 görüntüleme • 16 gün önce

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.
4:50

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.

Santiago

384,510 görüntüleme • 3 yıl önce

Physical Intelligence (Physical Intelligence) is building a foundation model that can control any robot to do any task — what the team describes as the GPT moment for robotics. The company's cross-embodiment approach trains across many different robot platforms, and recent results show tasks being performed zero-shot that last year required hundreds of hours of data collection. In this episode of the Lightcone Podcast , co-founder Quan Vuong (Quan Vuong) sat down with Garry Tan, Jared Friedman, Diana, and Harj Taggar to talk about why robotics is finally ready for its scaling moment, how PI runs its models in the cloud rather than on-device, and the playbook for what Quan sees as a Cambrian explosion of vertical robotics companies. 00:00 — Robotics just got cheaper 00:41 — The GPT moment for robotics 02:24 — Why robots didn’t work before 05:30 — The breakthrough that changed everything 09:12 — The data problem 13:33 — Robots learning without data 15:05 — Robots folding laundry (for real) 22:18 — From engineering problem → ops problem 29:12 — The startup playbook 38:46 — Thousands of robotics startups are coming
49:26

Physical Intelligence (Physical Intelligence) is building a foundation model that can control any robot to do any task — what the team describes as the GPT moment for robotics. The company's cross-embodiment approach trains across many different robot platforms, and recent results show tasks being performed zero-shot that last year required hundreds of hours of data collection. In this episode of the Lightcone Podcast , co-founder Quan Vuong (Quan Vuong) sat down with Garry Tan, Jared Friedman, Diana, and Harj Taggar to talk about why robotics is finally ready for its scaling moment, how PI runs its models in the cloud rather than on-device, and the playbook for what Quan sees as a Cambrian explosion of vertical robotics companies. 00:00 — Robotics just got cheaper 00:41 — The GPT moment for robotics 02:24 — Why robots didn’t work before 05:30 — The breakthrough that changed everything 09:12 — The data problem 13:33 — Robots learning without data 15:05 — Robots folding laundry (for real) 22:18 — From engineering problem → ops problem 29:12 — The startup playbook 38:46 — Thousands of robotics startups are coming

Y Combinator

115,590 görüntüleme • 3 ay önce