Загрузка видео...

Не удалось загрузить видео

На главную

Excited to share LWD: Learning While Deploying. Our robots learn while doing real tasks—restocking groceries, brewing Gongfu tea, making cocktails, making juice, and packing shoes. Deployment is no longer just evaluation; it becomes the training loop. 🧵

jianlanluo's profile picture

Jianlan Luo

2,285 subscribers

623,225 просмотров • 4 месяцев назад •via X (Twitter)

ОбразованиеНаука и технологии
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Our CTO boyuan is demoing our new localization algorithm, which surpasses leading industry solutions. Localization enables a robot to determine its position in an environment, which is essential for navigation. While most systems require robots to start from a predefined location, ours doesn’t. It lets robots boot up anywhere and immediately locate themselves, making deployment far more flexible for real world scenarios.
2:09

Our CTO boyuan is demoing our new localization algorithm, which surpasses leading industry solutions. Localization enables a robot to determine its position in an environment, which is essential for navigation. While most systems require robots to start from a predefined location, ours doesn’t. It lets robots boot up anywhere and immediately locate themselves, making deployment far more flexible for real world scenarios.

OpenMind

109,661 просмотров • 9 месяцев назад

𝗛𝘂𝗺𝘆𝗻𝗹𝗮𝗯𝘀 𝗶𝘀 𝗰𝗿𝗲𝗮𝘁𝗶𝗻𝗴 𝗮 𝗱𝗮𝘁𝗮 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗳𝗼𝗿 𝗽𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗔𝗜. They collect thousands of hours of real human activities like what people see, hear, move, and feel and turn it into clean training data for robots. Unlike language models that learn from the internet, robots need real world experience for messy everyday tasks such as ironing while a child runs past, restocking shelves in a crowded store, or assembling parts on a busy factory floor. Humyn Labs' approach uses four sensory streams captured in actual environments so robots can learn practical skills from human experience.
0:38

𝗛𝘂𝗺𝘆𝗻𝗹𝗮𝗯𝘀 𝗶𝘀 𝗰𝗿𝗲𝗮𝘁𝗶𝗻𝗴 𝗮 𝗱𝗮𝘁𝗮 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝗳𝗼𝗿 𝗽𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗔𝗜. They collect thousands of hours of real human activities like what people see, hear, move, and feel and turn it into clean training data for robots. Unlike language models that learn from the internet, robots need real world experience for messy everyday tasks such as ironing while a child runs past, restocking shelves in a crowded store, or assembling parts on a busy factory floor. Humyn Labs' approach uses four sensory streams captured in actual environments so robots can learn practical skills from human experience.

𝐓𝐡𝐞𝐎𝐧𝐥𝐲𝟏𝐇𝐞𝐫𝐨👑

18,514 просмотров • 17 дней назад

The robot is learning several novel tasks instantly, after just ONE demonstration each... Instant Policy makes it possible: no extra training, no weight updates, just pure in-context learning. It just got accepted at ICLR 2025, and it’s changing how robots learn. With just a single demo, a robot can pick up a new task and start performing it right away. Why this is a big deal: ✅ Learns tasks instantly with just one or a few demonstrations ✅ Improves over time as more demonstrations are given ✅ Uses simulation-based training with “pseudo-demonstrations” for scalability ✅ Can transfer skills across different robots and even follow language-defined tasks It brings in-context learning to robotics, opening up new possibilities for flexible, real-world automation. You can try it yourself: code and weights are available at • • • • Thank you to Edward Johns, Director of the Robot Learning Lab at Imperial College for sharing their work! 🙏
0:50

The robot is learning several novel tasks instantly, after just ONE demonstration each... Instant Policy makes it possible: no extra training, no weight updates, just pure in-context learning. It just got accepted at ICLR 2025, and it’s changing how robots learn. With just a single demo, a robot can pick up a new task and start performing it right away. Why this is a big deal: ✅ Learns tasks instantly with just one or a few demonstrations ✅ Improves over time as more demonstrations are given ✅ Uses simulation-based training with “pseudo-demonstrations” for scalability ✅ Can transfer skills across different robots and even follow language-defined tasks It brings in-context learning to robotics, opening up new possibilities for flexible, real-world automation. You can try it yourself: code and weights are available at • • • • Thank you to Edward Johns, Director of the Robot Learning Lab at Imperial College for sharing their work! 🙏

Ilir Aliu

46,285 просмотров • 1 год назад

So I heard we need more data for robot learning :) Purely real world teleop is expensive and slow, making large scale data collection challenging. I’ve been excited about getting more data into robot learning, going beyond just real-world teleop data. To this end, we’ve been scaling up data generation with RL in realistic simulations generated on the fly from crowdsourced videos. Enables realistic data collection, much more cheaply than purely real world teleop. Importantly, data collection becomes even*cheaper* with more environments, allowing training with over 100x more data. Transfers to real robots for generalizable manipulation. A 🧵 (1/N)
0:41

So I heard we need more data for robot learning :) Purely real world teleop is expensive and slow, making large scale data collection challenging. I’ve been excited about getting more data into robot learning, going beyond just real-world teleop data. To this end, we’ve been scaling up data generation with RL in realistic simulations generated on the fly from crowdsourced videos. Enables realistic data collection, much more cheaply than purely real world teleop. Importantly, data collection becomes even*cheaper* with more environments, allowing training with over 100x more data. Transfers to real robots for generalizable manipulation. A 🧵 (1/N)

Abhishek Gupta

13,358 просмотров • 1 год назад

Our elites have failed. They're profiting from chaos while public services collapse. I just dropped a new episode with William Clouston SDP diving into the "Doom Loop" of government waste, diversity quotas, and the grift that's making everyone poorer. Links to full episode in 🧵
0:41

Our elites have failed. They're profiting from chaos while public services collapse. I just dropped a new episode with William Clouston SDP diving into the "Doom Loop" of government waste, diversity quotas, and the grift that's making everyone poorer. Links to full episode in 🧵

Peter McCormack 🏴‍☠️🇬🇧🇮🇪

17,563 просмотров • 10 месяцев назад

Meet the world at home, where life happens and bots become family 35 days ago, at our “Born to Bot, Bot to Family” launch event, we shared our vision of bringing robots into real homes. Today, we’re very happy to share that our robots are now gradually entering real families. For embodied AI, the real world is everyday life: different routines, different kitchens, and different ways of doing even the simplest tasks. This is where robots meet the world at home, where life happens and bots become family. They are still learning. They may move slowly, hesitate, and sometimes look a little clumsy. But every home they enter helps them understand the world a little better.
5:47

Meet the world at home, where life happens and bots become family 35 days ago, at our “Born to Bot, Bot to Family” launch event, we shared our vision of bringing robots into real homes. Today, we’re very happy to share that our robots are now gradually entering real families. For embodied AI, the real world is everyday life: different routines, different kitchens, and different ways of doing even the simplest tasks. This is where robots meet the world at home, where life happens and bots become family. They are still learning. They may move slowly, hesitate, and sometimes look a little clumsy. But every home they enter helps them understand the world a little better.

X Square Robot

1,204,483 просмотров • 3 месяцев назад

Hard. Fast. But predictable. Great way to limit training risk. To me, shields and mats are about expressing power while managing contact load. What i see most is the runner not making it easy, but not trying to win. Giving just the right weight to it.
0:26

Hard. Fast. But predictable. Great way to limit training risk. To me, shields and mats are about expressing power while managing contact load. What i see most is the runner not making it easy, but not trying to win. Giving just the right weight to it.

Andrew Ryland

82,366 просмотров • 1 год назад

Robots Doing Stuff #61 | Bimanual Nut Pick and Threading This is a real customer task: one step in a longer multi-part sequence we’re training in our labs. Threading requires precise alignment, force control, and fine manipulation. These are foundational capabilities for industrial dexterity applications that are currently inaccessible with traditional automation. We teleoperate these tasks to collect the data that trains our Physical AI. Next stop: autonomous. Visit to learn more about our Physical AI-enabled automation solutions. #Dexterity #DexterousHand #RobotHand #Robotics #RobotsDoingStuff #SanctuaryAI
0:35

Robots Doing Stuff #61 | Bimanual Nut Pick and Threading This is a real customer task: one step in a longer multi-part sequence we’re training in our labs. Threading requires precise alignment, force control, and fine manipulation. These are foundational capabilities for industrial dexterity applications that are currently inaccessible with traditional automation. We teleoperate these tasks to collect the data that trains our Physical AI. Next stop: autonomous. Visit to learn more about our Physical AI-enabled automation solutions. #Dexterity #DexterousHand #RobotHand #Robotics #RobotsDoingStuff #SanctuaryAI

Sanctuary AI

38,039 просмотров • 2 месяцев назад

Austin-based robotics startup Apptronik has unveiled Apollo 2, the latest version of its humanoid robot. Apollo 2 comes in both bipedal and wheeled versions and performs real-world tasks while collecting training data. Apptronik is using this data with Google DeepMind to train Gemini Robotics AI models, making Apollo smarter and more capable.
0:28

Austin-based robotics startup Apptronik has unveiled Apollo 2, the latest version of its humanoid robot. Apollo 2 comes in both bipedal and wheeled versions and performs real-world tasks while collecting training data. Apptronik is using this data with Google DeepMind to train Gemini Robotics AI models, making Apollo smarter and more capable.

Space and Technology

11,540 просмотров • 1 месяц назад

🌟 AZCoiner: The Ultimate AI-Powered Crypto Platform! 🌟 AZCoiner integrates AI into the crypto market, making it easier to earn and optimize opportunities💸 Our AI Agent automates tasks like airdrop hunting, investment strategy optimization, and providing personalized insights. Powered by Reinforcement Learning & GO-AI, we connect the community and support Web3 projects to thrive🌐🚀 💡 Join us, train the AI, and earn rewards while shaping the future of crypto! AZCoiner - Learn, grow, and earn sustainably 🌌 #AZCoiner $AZC
1:19

🌟 AZCoiner: The Ultimate AI-Powered Crypto Platform! 🌟 AZCoiner integrates AI into the crypto market, making it easier to earn and optimize opportunities💸 Our AI Agent automates tasks like airdrop hunting, investment strategy optimization, and providing personalized insights. Powered by Reinforcement Learning & GO-AI, we connect the community and support Web3 projects to thrive🌐🚀 💡 Join us, train the AI, and earn rewards while shaping the future of crypto! AZCoiner - Learn, grow, and earn sustainably 🌌 #AZCoiner $AZC

AZCoiner

345,275 просмотров • 1 год назад

Robots are getting smarter, but most still fail the same way. They don’t learn from their own mistakes. A new paper proposes something different: a way for robots to self-improve directly from their failures in the real world. It’s called PLD (Probe, Learn, Distill). The idea: instead of collecting endless human demos, let the robot figure out where it fails, learn how to recover, and then distill that knowledge back into its main model. Key takeaways from the research: ✅ Uses residual reinforcement learning to recover from policy failures ✅ Achieves 99% success on LIBERO and 100% on real Franka and YAM arms ✅ Runs hour-long manipulation tasks without human resets ✅ Builds a feedback loop between real-world data and model adaptation Unlike supervised fine-tuning, which relies on humans, PLD learns from the robot’s own experience. By training on its own failure distribution, the model becomes both more efficient and more aligned with the real world. It’s not just a technical shift, it’s a step toward robots that improve themselves through real-world practice. Thanks for sharing, Wenli Xiao !! Paper and demos: —- Weekly robotics and AI insights. Subscribe free:
0:47

Robots are getting smarter, but most still fail the same way. They don’t learn from their own mistakes. A new paper proposes something different: a way for robots to self-improve directly from their failures in the real world. It’s called PLD (Probe, Learn, Distill). The idea: instead of collecting endless human demos, let the robot figure out where it fails, learn how to recover, and then distill that knowledge back into its main model. Key takeaways from the research: ✅ Uses residual reinforcement learning to recover from policy failures ✅ Achieves 99% success on LIBERO and 100% on real Franka and YAM arms ✅ Runs hour-long manipulation tasks without human resets ✅ Builds a feedback loop between real-world data and model adaptation Unlike supervised fine-tuning, which relies on humans, PLD learns from the robot’s own experience. By training on its own failure distribution, the model becomes both more efficient and more aligned with the real world. It’s not just a technical shift, it’s a step toward robots that improve themselves through real-world practice. Thanks for sharing, Wenli Xiao !! Paper and demos: —- Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

22,881 просмотров • 10 месяцев назад

After a few illegal aliens died while fleeing ICE, the Trump admin has CUCKED again and is no longer doing traffic stops to deport people. “It's better than ever in terms of resources. They just won't now. They just won't do it.”
2:31

After a few illegal aliens died while fleeing ICE, the Trump admin has CUCKED again and is no longer doing traffic stops to deport people. “It's better than ever in terms of resources. They just won't now. They just won't do it.”

Fuentes Updates

54,798 просмотров • 1 месяц назад

Lonely Mountain Town is out now! Very proud of this collection of songs, and I’m excited to finally share them with y’all. It was a cathartic experience for me to record these songs while on the road last year. I hope you enjoy this EP as much as I enjoyed making it. Listen:
0:30

Lonely Mountain Town is out now! Very proud of this collection of songs, and I’m excited to finally share them with y’all. It was a cathartic experience for me to record these songs while on the road last year. I hope you enjoy this EP as much as I enjoyed making it. Listen:

Charles Wesley Godwin

11,736 просмотров • 1 год назад

📢 We're excited to share that the GraphLinq Intent AI LLM Integration is now live! 🛠️ Try now at: Now automate smart-contract execution, trades, tracking tools, and much more like deploying a token on any EVM blockchain or more complex tasks like making automated trades on any of the DEXs available on our IDE etc., powered by simple conversational prompts. This is a perfect blend of AI & Building on Web3 without having to code and we're just getting started 💪 #Ethereum #Solana #Polygon #DeFi #GLQChain #Avalanche #Web3 #Blockchain #Nocode #Automation
1:21

📢 We're excited to share that the GraphLinq Intent AI LLM Integration is now live! 🛠️ Try now at: Now automate smart-contract execution, trades, tracking tools, and much more like deploying a token on any EVM blockchain or more complex tasks like making automated trades on any of the DEXs available on our IDE etc., powered by simple conversational prompts. This is a perfect blend of AI & Building on Web3 without having to code and we're just getting started 💪 #Ethereum #Solana #Polygon #DeFi #GLQChain #Avalanche #Web3 #Blockchain #Nocode #Automation

GraphLinq Chain

466,184 просмотров • 2 лет назад

Testnet/Beta is officially live! GPUs are live and ready to use, while TPUs and nodes are undergoing a bit of maintenance. This platform has been a long time in the making, and we’re excited to share it with you. 🔗 Try it here: 📖 Tutorial: Stay tuned for the second update, where advertisement banners and revenue sharing will go live. Big things ahead!
0:36

Testnet/Beta is officially live! GPUs are live and ready to use, while TPUs and nodes are undergoing a bit of maintenance. This platform has been a long time in the making, and we’re excited to share it with you. 🔗 Try it here: 📖 Tutorial: Stay tuned for the second update, where advertisement banners and revenue sharing will go live. Big things ahead!

Virtu Network

10,653 просмотров • 1 год назад

Fat loss happens inside your mitochondria, not just on the scale. Stored fat only becomes usable energy when two things happen: • Fatty acids are released from storage. • Your body needs the energy to produce ATP. That's why hormones, training, and nutrition all work together. And while a calorie deficit is important, making it too aggressive can trigger metabolic adaptation, where your body slows calorie burn in an effort to conserve.
1:04

Fat loss happens inside your mitochondria, not just on the scale. Stored fat only becomes usable energy when two things happen: • Fatty acids are released from storage. • Your body needs the energy to produce ATP. That's why hormones, training, and nutrition all work together. And while a calorie deficit is important, making it too aggressive can trigger metabolic adaptation, where your body slows calorie burn in an effort to conserve.

Gary Brecka

26,807 просмотров • 2 месяцев назад

🚨 MIKE LEE JUST DETONATED ON THE SENATE FLOOR While they’re packing their bags for August recess, China is caught hacking our election data — and Trump just exposed it. Lee wasn’t having it: “We’re expected to just shrug our shoulders and do NOTHING when they target the American voter?! When they target that which is most uniquely, sacredly given to American citizens… the exclusive right to vote…” “NO SIR, NOT ON MY WATCH!” “We cannot, consistent with our oaths to uphold, protect and defend the Constitution… simply ignore it!” This is the fight. The right to vote is the sine qua non of American citizenship — and the Senate is sleepwalking while the enemy targets it. Follow me for more intel drops!
1:01

🚨 MIKE LEE JUST DETONATED ON THE SENATE FLOOR While they’re packing their bags for August recess, China is caught hacking our election data — and Trump just exposed it. Lee wasn’t having it: “We’re expected to just shrug our shoulders and do NOTHING when they target the American voter?! When they target that which is most uniquely, sacredly given to American citizens… the exclusive right to vote…” “NO SIR, NOT ON MY WATCH!” “We cannot, consistent with our oaths to uphold, protect and defend the Constitution… simply ignore it!” This is the fight. The right to vote is the sine qua non of American citizenship — and the Senate is sleepwalking while the enemy targets it. Follow me for more intel drops!

QThestorm

24,880 просмотров • 1 месяц назад

Small businesses, listen up! Just in time for the San Francisco “summer,” you no longer have to pay to put tables and chairs on the sidewalk. A quick, free registration is all it takes, and you’re good to go. Through our permit reform initiative, PermitSF, my administration is cutting red tape and making it easier for small businesses to thrive.
0:29

Small businesses, listen up! Just in time for the San Francisco “summer,” you no longer have to pay to put tables and chairs on the sidewalk. A quick, free registration is all it takes, and you’re good to go. Through our permit reform initiative, PermitSF, my administration is cutting red tape and making it easier for small businesses to thrive.

Daniel Lurie 丹尼爾·羅偉

37,733 просмотров • 1 год назад