正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Introduce 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 -- Learning! With 50 demos, our robot can autonomously complete complex mobile manipulation tasks: - cook and serve shrimp🦐 - call and take elevator🛗 - store a 3Ibs pot to a two-door cabinet Open-sourced! Co-led Tony Zhao, Chelsea Finn

Zipeng Fu

14,281 subscribers

1,568,419 次观看 • 2 年前 •via X (Twitter)

健康养生教育科学技术

Anya Rossi• Live Now

Private livecam show

10 条评论

Zipeng Fu 的头像

Zipeng Fu2 年前

Our robot can consistently handle these tasks, succeeding: - 9 times in a row for Wipe Wine - 5 times for Call Elevator - robust against distractors for Use Cabinet - extrapolate to chairs unseen during training

Zipeng Fu 的头像

Zipeng Fu2 年前

How do we achieve this with only 50 demos? The key is to co-train imitation learning algorithms with static ALOHA data. We found this to consistently improve performance, especially for tasks that require precise manipulation.

Zipeng Fu 的头像

Zipeng Fu2 年前

Co-training (1) improves the performance across all tasks, (2) is compatible with ACT, Diffusion Policy and VINN, (3) is robust to different data mixtures.

Zipeng Fu 的头像

Zipeng Fu2 年前

We open-source all the software and data of Mobile ALOHA! Project Website 🛜: Code for Imitation Learning 🖥️: Data 📊:

Zipeng Fu 的头像

Zipeng Fu2 年前

Want to dive deeper into the hardware of Mobile ALOHA? Check out 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 -- Hardware from co-lead @tonyzzhao!

Karol Hausman 的头像

Karol Hausman2 年前

@tonyzzhao @chelseabfinn Awesome project, congrats!

Zipeng Fu 的头像

Zipeng Fu2 年前

@tonyzzhao @chelseabfinn Thanks Karol!

Nick Dobos 的头像

Nick Dobos2 年前

@tonyzzhao @chelseabfinn Did you eat the shrimp?! How did it taste!?

Qualy the lightbulb 的头像

Qualy the lightbulb2 年前

@tonyzzhao @chelseabfinn > cook and serve shrimp Are you trying to provoke me?

Luddite Design 的头像

Luddite Design2 年前

@tonyzzhao @chelseabfinn Can it?

相关视频

Introducing 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 -- Hardware! A low-cost, open-source, mobile manipulator. One of the most high-effort projects in my past 5yrs! Not possible without co-lead Zipeng Fu and Chelsea Finn. At the end, what's better than cooking yourself a meal with the 🤖🧑‍🍳

Introducing 𝐌𝐨𝐛𝐢𝐥𝐞 𝐀𝐋𝐎𝐇𝐀🏄 -- Hardware! A low-cost, open-source, mobile manipulator. One of the most high-effort projects in my past 5yrs! Not possible without co-lead Zipeng Fu and Chelsea Finn. At the end, what's better than cooking yourself a meal with the 🤖🧑‍🍳

Tony Zhao

1,669,194 次观看 • 2 年前

Mobile ALOHA's hardware is very capable. We brought it home yesterday and tried more tasks! It can: - do laundry👔👖 - self-charge⚡️ - use a vacuum - water plants🌳 - load and unload a dishwasher - use a coffee machine☕️ - obtain drinks from the fridge and open a beer🍺 - open doors🚪 - play with pets🐱 - throw away trash - turn on/off a lamp💡 Project website: Co-lead Tony Zhao, advised by Chelsea Finn (amazing photographing from Qingqing Zhao )

Mobile ALOHA's hardware is very capable. We brought it home yesterday and tried more tasks! It can: - do laundry👔👖 - self-charge⚡️ - use a vacuum - water plants🌳 - load and unload a dishwasher - use a coffee machine☕️ - obtain drinks from the fridge and open a beer🍺 - open doors🚪 - play with pets🐱 - throw away trash - turn on/off a lamp💡 Project website: Co-lead Tony Zhao, advised by Chelsea Finn (amazing photographing from Qingqing Zhao )

Zipeng Fu

2,983,821 次观看 • 2 年前

Introducing MARS, the first Personal AI Robot. In technology, getting to the future we want starts with building it together. This is why we designed Mars. Inspired by early PCs, Mars is a powerful, complete, extendable robot. It runs BASIC, an open embodied AI agent we created for you to program with code, demonstrations, and prompts. This allows Mars to perform complex long-horizon tasks involving spatial memory, reasoning, manipulation and navigation. Pre-orders open now ⬇️ Demos & details in the thread

Introducing MARS, the first Personal AI Robot. In technology, getting to the future we want starts with building it together. This is why we designed Mars. Inspired by early PCs, Mars is a powerful, complete, extendable robot. It runs BASIC, an open embodied AI agent we created for you to program with code, demonstrations, and prompts. This allows Mars to perform complex long-horizon tasks involving spatial memory, reasoning, manipulation and navigation. Pre-orders open now ⬇️ Demos & details in the thread

Axel

314,313 次观看 • 1 年前

Excited to introduce SOLE-R1, a video-language reasoning model for zero-shot reward prediction for robot manipulation tasks! SOLE-R1 reasoning can serve as the SOLE signal for learning new tasks (completely from scratch) through online RL - i.e., robots start with random actions and learn previously unseen tasks guided only by SOLE-R1 rewards, without any demonstrations, ground-truth rewards, success indicators, or task-specific tuning. SOLE-R1 significantly outperforms strong baselines (e.g., Robometer, RoboReward, TOPReward, GPT-5, Gemini-3-Pro) in zero-shot online RL when evaluated across 40 tasks - including a real-world tabletop manipulation setting and 4 sim environments (LIBERO, ManiSkill, Meta-World, RoboSuite). We open source all models, training data, and code. Website, demos, and paper at: 🧵 (1/6)

Excited to introduce SOLE-R1, a video-language reasoning model for zero-shot reward prediction for robot manipulation tasks! SOLE-R1 reasoning can serve as the SOLE signal for learning new tasks (completely from scratch) through online RL - i.e., robots start with random actions and learn previously unseen tasks guided only by SOLE-R1 rewards, without any demonstrations, ground-truth rewards, success indicators, or task-specific tuning. SOLE-R1 significantly outperforms strong baselines (e.g., Robometer, RoboReward, TOPReward, GPT-5, Gemini-3-Pro) in zero-shot online RL when evaluated across 40 tasks - including a real-world tabletop manipulation setting and 4 sim environments (LIBERO, ManiSkill, Meta-World, RoboSuite). We open source all models, training data, and code. Website, demos, and paper at: 🧵 (1/6)

Philip Schroeder

19,740 次观看 • 4 个月前

🧐Applying world models to improve real-world policy on challenging manipulation tasks used to be considered out of reach. 😌After sustained effort, we’re now seeing encouraging progress. 🚀Thrilled to introduce RISE: Self-Improving Robot Policy with Compositional World Model RISE is, to our knowledge, the first work to use a world model as an effective learning environment for challenging real-world manipulation, enabling policy improvement on tasks that demand high dynamics, dexterity, and precision. Incredible teamwork with KL Hongyang Li Yue Xiangyu Hao Zhao Chonghao Sima

🧐Applying world models to improve real-world policy on challenging manipulation tasks used to be considered out of reach. 😌After sustained effort, we’re now seeing encouraging progress. 🚀Thrilled to introduce RISE: Self-Improving Robot Policy with Compositional World Model RISE is, to our knowledge, the first work to use a world model as an effective learning environment for challenging real-world manipulation, enabling policy improvement on tasks that demand high dynamics, dexterity, and precision. Incredible teamwork with KL Hongyang Li Yue Xiangyu Hao Zhao Chonghao Sima

Jiazhi Yang

77,124 次观看 • 7 个月前

Sunday Robotics unveiled its home robot, Memo, a wheeled robot with two arms and pincerlike hands. Training method: Sunday pays remote workers to perform household tasks wearing gloves that resemble Memo's hands. The Mountain View-based company is focused on a full-stack approach, building both hardware and AI models. Beta testing is scheduled to begin next year. Co-founders: - CEO Tony Zhao, ex-Stanford Ph.D. dropout and ex-DeepMind/Tesla, co-created ALOHA project (imitation learning). - CTO Cheng Chi, a Columbia Ph.D., invented the highly-cited work, Diffusion Policy.

Sunday Robotics unveiled its home robot, Memo, a wheeled robot with two arms and pincerlike hands. Training method: Sunday pays remote workers to perform household tasks wearing gloves that resemble Memo's hands. The Mountain View-based company is focused on a full-stack approach, building both hardware and AI models. Beta testing is scheduled to begin next year. Co-founders: - CEO Tony Zhao, ex-Stanford Ph.D. dropout and ex-DeepMind/Tesla, co-created ALOHA project (imitation learning). - CTO Cheng Chi, a Columbia Ph.D., invented the highly-cited work, Diffusion Policy.

The Humanoid Hub

140,597 次观看 • 10 个月前

Enough with LLMs - exciting things are happening in the world of atoms. This is Stanford ALOHA, a low-cost and agile robot platform. The whole system is open-source (!!): hardware design, CAD models for 3D printing, simulator, and training code. Time to graft a physical arm onto GPTs 🦾 Led by my friend Tony Zhao at Stanford, and advised by Chelsea Finn Sergey Levine Vikash Kumar ✈️IROS2026. Project page: If you don't want to 3D print your own components, you can buy the setup at

Enough with LLMs - exciting things are happening in the world of atoms. This is Stanford ALOHA, a low-cost and agile robot platform. The whole system is open-source (!!): hardware design, CAD models for 3D printing, simulator, and training code. Time to graft a physical arm onto GPTs 🦾 Led by my friend Tony Zhao at Stanford, and advised by Chelsea Finn Sergey Levine Vikash Kumar ✈️IROS2026. Project page: If you don't want to 3D print your own components, you can buy the setup at

Jim Fan

818,862 次观看 • 3 年前

Today we are excited to open up Neuracore to the academic community! Neuracore is a new data foundation built to accelerate robot learning by removing one of the field’s biggest bottlenecks: capturing and working with high-fidelity multimodal robotics data. For the first time, researchers can store, view, and work with robotics data in a cloud-native system built specifically for large-scale learning, and we are making this core platform completely free for academia. The platform lets teams capture every sensor at its native rate, store and visualize data without loss, and then train and deploy models locally using our open-source code (Link in the comments). We are rolling out access to select academic institutions first. Anyone with an academic email can sign up, and if your institution is not part of the initial rollout, you will be able to join the waitlist directly. Beyond providing this infrastructure, we see an opportunity to build a global community where engineers and researchers can share, collaborate, and advance the frontier of robot learning together. Supported by our recent $3M pre-seed round led by Earlybird VC, we are excited to take this mission even further. Our long-term goal is for Neuracore to become the natural home for cutting-edge robot learning algorithms and real-world robotics experimentation, helping accelerate the next wave of Physical AI.

Today we are excited to open up Neuracore to the academic community! Neuracore is a new data foundation built to accelerate robot learning by removing one of the field’s biggest bottlenecks: capturing and working with high-fidelity multimodal robotics data. For the first time, researchers can store, view, and work with robotics data in a cloud-native system built specifically for large-scale learning, and we are making this core platform completely free for academia. The platform lets teams capture every sensor at its native rate, store and visualize data without loss, and then train and deploy models locally using our open-source code (Link in the comments). We are rolling out access to select academic institutions first. Anyone with an academic email can sign up, and if your institution is not part of the initial rollout, you will be able to join the waitlist directly. Beyond providing this infrastructure, we see an opportunity to build a global community where engineers and researchers can share, collaborate, and advance the frontier of robot learning together. Supported by our recent $3M pre-seed round led by Earlybird VC, we are excited to take this mission even further. Our long-term goal is for Neuracore to become the natural home for cutting-edge robot learning algorithms and real-world robotics experimentation, helping accelerate the next wave of Physical AI.

Neuracore

40,810 次观看 • 10 个月前

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

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

186,783 次观看 • 6 个月前

🚀 We just shipped a big upgrade to our imitation-learning-in-simulation playground in LeRobot, built together with Lightwheel ! You can now teleoperate robots in sim (keyboard or real robot) and collect training demos instantly. This makes it possible to run real IL research on harder, more realistic manipulation tasks, even if you don’t have hardware. New tasks: 🟠 pick orange to the plate 🧺 fold cloth (yes!) 📦 pick 2 “e” toys to the box 🔴 lift red cube Through our partnership with Lightwheel, LeIsaac was integrated into EnvHub on day one, a best-practice integration that strengthens both ecosystems and pushes simulation-first robotics forward. 👉 Load a task via EnvHub, start teleop-ing, record your demo, upload the data on the hub and you’re ready to train or test.

🚀 We just shipped a big upgrade to our imitation-learning-in-simulation playground in LeRobot, built together with Lightwheel ! You can now teleoperate robots in sim (keyboard or real robot) and collect training demos instantly. This makes it possible to run real IL research on harder, more realistic manipulation tasks, even if you don’t have hardware. New tasks: 🟠 pick orange to the plate 🧺 fold cloth (yes!) 📦 pick 2 “e” toys to the box 🔴 lift red cube Through our partnership with Lightwheel, LeIsaac was integrated into EnvHub on day one, a best-practice integration that strengthens both ecosystems and pushes simulation-first robotics forward. 👉 Load a task via EnvHub, start teleop-ing, record your demo, upload the data on the hub and you’re ready to train or test.

LeRobot

44,856 次观看 • 10 个月前

For robots to be useful, they must be able to interact with a wide variety of environments; and yet, scaling interaction data is difficult, expensive, and time consuming. Instead, much research revolves around sim-to-real manipulation — but mostly this has not been mobile manipulation. Recently, though, this has begun to change. Two recent papers from Tairan He and Haoru Xue show us how to unlock the potential of this technique, building policies which, without any real data at all, can move objects around in the world and open doors in the real world with a humanoid robot. Watch Episode #60 of RoboPapers now to learn more, hosted by Chris Paxton and Jiafei Duan. In this episode, we cover two papers:. First is VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation; and second is DoorMan: Opening the Sim-to-Real Door for Humanoid Pixel-to-Action Policy Transfer.

For robots to be useful, they must be able to interact with a wide variety of environments; and yet, scaling interaction data is difficult, expensive, and time consuming. Instead, much research revolves around sim-to-real manipulation — but mostly this has not been mobile manipulation. Recently, though, this has begun to change. Two recent papers from Tairan He and Haoru Xue show us how to unlock the potential of this technique, building policies which, without any real data at all, can move objects around in the world and open doors in the real world with a humanoid robot. Watch Episode #60 of RoboPapers now to learn more, hosted by Chris Paxton and Jiafei Duan. In this episode, we cover two papers:. First is VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation; and second is DoorMan: Opening the Sim-to-Real Door for Humanoid Pixel-to-Action Policy Transfer.

RoboPapers

30,767 次观看 • 8 个月前

In my past research experience, finding or developing an appropriate simulation environment, dataset, and benchmark has always been a challenge. Missing features, limited support, or unexpected bugs often occupied my days and nights. Moreover, current simulation platforms are relatively fragmented—making it challenging to replicate the success of the RT-X dataset in unifying community efforts. Introducing RoboVerse, we provide a unified platform, dataset, and benchmark for scalable and generalizable robot learning. We hope to build a shared foundation to combine the community efforts. RoboVerse includes: MetaSim: We carefully designed a configuration system and a universal interface to align current robotic simulators. With MetaSim, you can use any simulator with the same code—bringing together the community’s diverse efforts under one framework! RoboVerse Dataset and Benchmark: We unify popular simulation environments and benchmarks into a single cohesive system and introduce the RoboVerse dataset—a large-scale, high-quality synthetic dataset. Additionally, we propose a standardized benchmark across both imitation learning and reinforcement learning. A cool feature enabled by our unified framework: Hybrid Simulation! You can now integrate physics engines and renderers from different simulators—e.g., using MuJoCo precise physics with Isaac photorealistic rendering. This not only elevates simulation fidelity but also significantly enhances real-world transfer performance across complex robotic applications. Hopefully, our team’s efforts could serve the robotic community to thrive vibrantly in the years to come. RoboVerse is open-sourced🥳!!! Project Page: Documentation: Github Repo: Paper:

In my past research experience, finding or developing an appropriate simulation environment, dataset, and benchmark has always been a challenge. Missing features, limited support, or unexpected bugs often occupied my days and nights. Moreover, current simulation platforms are relatively fragmented—making it challenging to replicate the success of the RT-X dataset in unifying community efforts. Introducing RoboVerse, we provide a unified platform, dataset, and benchmark for scalable and generalizable robot learning. We hope to build a shared foundation to combine the community efforts. RoboVerse includes: MetaSim: We carefully designed a configuration system and a universal interface to align current robotic simulators. With MetaSim, you can use any simulator with the same code—bringing together the community’s diverse efforts under one framework! RoboVerse Dataset and Benchmark: We unify popular simulation environments and benchmarks into a single cohesive system and introduce the RoboVerse dataset—a large-scale, high-quality synthetic dataset. Additionally, we propose a standardized benchmark across both imitation learning and reinforcement learning. A cool feature enabled by our unified framework: Hybrid Simulation! You can now integrate physics engines and renderers from different simulators—e.g., using MuJoCo precise physics with Isaac photorealistic rendering. This not only elevates simulation fidelity but also significantly enhances real-world transfer performance across complex robotic applications. Hopefully, our team’s efforts could serve the robotic community to thrive vibrantly in the years to come. RoboVerse is open-sourced🥳!!! Project Page: Documentation: Github Repo: Paper:

Haoran Geng

84,421 次观看 • 1 年前

⚡SHARPA LEAVES THE LAB, HEADS TO DAIRY QUEEN Sharpa is taking its 22-DoF Wave hand beyond demos with Project Edison, preparing its North robot for a real Dairy Queen deployment in Shanghai. Humanoid Scott got an early look at the project. A few things that really stood out from our conversation: > Sharpa is building the full stack around the hand, including the robot arms, mobile base and AI > North has to work in the existing kitchen, with no redesign of the workspace around the robot > Scott Walter says making a Blizzard involves roughly 50–60 steps through an existing human workflow > The preview footage is teleoperated and sped up. Sharpa expects the released system to be fully autonomous But the bigger story here is Sharpa taking dexterous robot hands from impressive demos to complete systems capable of useful commercial work. And Sharpa founder David Li has offered to join us on Over The Horizon to take us deeper inside Project Edison. Great discussion with Humanoid Scott and Mehrdad Farimani. Looking forward to hopefully having Sharpa and David Li join us for a deeper dive into their work! ⚡Full episode linked below in the thread!👇🏽

⚡SHARPA LEAVES THE LAB, HEADS TO DAIRY QUEEN Sharpa is taking its 22-DoF Wave hand beyond demos with Project Edison, preparing its North robot for a real Dairy Queen deployment in Shanghai. Humanoid Scott got an early look at the project. A few things that really stood out from our conversation: > Sharpa is building the full stack around the hand, including the robot arms, mobile base and AI > North has to work in the existing kitchen, with no redesign of the workspace around the robot > Scott Walter says making a Blizzard involves roughly 50–60 steps through an existing human workflow > The preview footage is teleoperated and sped up. Sharpa expects the released system to be fully autonomous But the bigger story here is Sharpa taking dexterous robot hands from impressive demos to complete systems capable of useful commercial work. And Sharpa founder David Li has offered to join us on Over The Horizon to take us deeper inside Project Edison. Great discussion with Humanoid Scott and Mehrdad Farimani. Looking forward to hopefully having Sharpa and David Li join us for a deeper dive into their work! ⚡Full episode linked below in the thread!👇🏽

Royden D'Souza

369,693 次观看 • 2 个月前

For decades, we’ve dreamed of robots that can seamlessly step into our world and lend a hand. Today, we take a major stride toward making that dream a reality: Introducing Gemini Robotics 2 from Google DeepMind, the intelligence layer powering the next generation of truly adaptable robots. This major advance unlocks intelligent whole-body control, advanced dexterity, and even multi-robot collaboration 🤯. Ok but... how does a robot actually "think"? Real-world tasks take time and planning. To manage that complexity, our new embodied reasoning model, Gemini Robotics ER 2, acts as the robot’s high-level brain, enhancing the robot’s capabilities to: — Observe the environment — Reason about the actions needed to complete the task — Coordinate with the vision-language-action model to carry out actions — Track progress until the job is done This setup allows robots to execute complex multi-step workflows, self-correct if a step fails, and adapt to completely novel situations. Learn more about Gemini Robotics ER 2 (and our two other brand new models) here:

Google AI

787,497 次观看 • 2 个月前

Synchronize Dual Hands for Physics-Based Dexterous Guitar Playing discuss: We present a novel approach to synthesize dexterous motions for physically simulated hands in tasks that require coordination between the control of two hands with high temporal precision. Instead of directly learning a joint policy to control two hands, our approach performs bimanual control through cooperative learning where each hand is treated as an individual agent. The individual policies for each hand are first trained separately, and then synchronized through latent space manipulation in a centralized environment to serve as a joint policy for two-hand control. By doing so, we avoid directly performing policy learning in the joint state-action space of two hands with higher dimensions, greatly improving the overall training efficiency. We demonstrate the effectiveness of our proposed approach in the challenging guitar-playing task. The virtual guitarist trained by our approach can synthesize motions from unstructured reference data of general guitar-playing practice motions, and accurately play diverse rhythms with complex chord pressing and string picking patterns based on the input guitar tabs that do not exist in the references. Along with this paper, we provide the motion capture data that we collected as the reference for policy training.

Synchronize Dual Hands for Physics-Based Dexterous Guitar Playing discuss: We present a novel approach to synthesize dexterous motions for physically simulated hands in tasks that require coordination between the control of two hands with high temporal precision. Instead of directly learning a joint policy to control two hands, our approach performs bimanual control through cooperative learning where each hand is treated as an individual agent. The individual policies for each hand are first trained separately, and then synchronized through latent space manipulation in a centralized environment to serve as a joint policy for two-hand control. By doing so, we avoid directly performing policy learning in the joint state-action space of two hands with higher dimensions, greatly improving the overall training efficiency. We demonstrate the effectiveness of our proposed approach in the challenging guitar-playing task. The virtual guitarist trained by our approach can synthesize motions from unstructured reference data of general guitar-playing practice motions, and accurately play diverse rhythms with complex chord pressing and string picking patterns based on the input guitar tabs that do not exist in the references. Along with this paper, we provide the motion capture data that we collected as the reference for policy training.

AK

26,917 次观看 • 2 年前

Can reinforcement learning from AI feedback unlock new capabilities in AI agents? Introducing Motif, an LLM-powered method for intrinsic motivation from AI feedback. Motif extracts reward functions from Llama 2's preferences and uses them to train agents with reinforcement learning. On the complex NetHack game, Motif solves previously unsolved tasks without needing any expert demonstrations. Surprisingly, Motif's reward leads to better game score than the one obtained by using the score itself as a reward. Given access to an event captioning mechanism, a few properties make Motif a general method: • it is entirely based on open models • the LLM doesn't need direct access to the environment dynamics (e.g., its source code) • the LLM doesn't need to understand observation and action spaces The best part? You can start using Motif right now, even on a small compute budget: the whole pipeline can take less than two GPU-days. Feel free to read our paper and try our code out. Paper: Code: Blog post: Work co-lead by Martin Klissarov and myself, with Shagun Sodhani Roberta Raileanu Pierre-Luc Bacon Pascal Vincent Amy Zhang Mikael Henaff Learn more in the thread 🧵

Can reinforcement learning from AI feedback unlock new capabilities in AI agents? Introducing Motif, an LLM-powered method for intrinsic motivation from AI feedback. Motif extracts reward functions from Llama 2's preferences and uses them to train agents with reinforcement learning. On the complex NetHack game, Motif solves previously unsolved tasks without needing any expert demonstrations. Surprisingly, Motif's reward leads to better game score than the one obtained by using the score itself as a reward. Given access to an event captioning mechanism, a few properties make Motif a general method: • it is entirely based on open models • the LLM doesn't need direct access to the environment dynamics (e.g., its source code) • the LLM doesn't need to understand observation and action spaces The best part? You can start using Motif right now, even on a small compute budget: the whole pipeline can take less than two GPU-days. Feel free to read our paper and try our code out. Paper: Code: Blog post: Work co-lead by Martin Klissarov and myself, with Shagun Sodhani Roberta Raileanu Pierre-Luc Bacon Pascal Vincent Amy Zhang Mikael Henaff Learn more in the thread 🧵

Pierluca D'Oro

311,942 次观看 • 2 年前

Planning with the views: Can VLMs predict how each camera move changes the view, and plan many such moves ahead? We introduce ViewSuite with 6 DoF camera control and ~165K task instances, testing: Path-to-View View-to-Path Interactive View Planning A sharp Planning Gap emerges: + can roughly "track" how camera action changes views - cannot "compose" a plan towards a target view at all We then try to teach VLMs with Reinforcement Learning. - RL cannot teach VLMs such planning ability, only 2.5% success rate with Qwen2.5-VL-7B. + With View Graph Distillation (our RL-Graph-SFT framework), 2.5% → 47.8% Below, we answer these questions: Q1. What are the failure modes? Q2. How can we make RL work? Q3. What has the model learned? Can we open up the model to see before/after? Can such spatial priors transfer to other view related tasks? Led by Kangrui Wang, great to work with Linjie (Lindsey) Li Zhengyuan Yang Shiqi Chen Zihan "Zenus" Wang Fei-Fei Li Jiajun Wu Leonidas Guibas, Lijuan Wang. A joint efforts with Stanford AI Lab Stanford Vision and Learning Lab Microsoft Research.

Planning with the views: Can VLMs predict how each camera move changes the view, and plan many such moves ahead? We introduce ViewSuite with 6 DoF camera control and ~165K task instances, testing: Path-to-View View-to-Path Interactive View Planning A sharp Planning Gap emerges: + can roughly "track" how camera action changes views - cannot "compose" a plan towards a target view at all We then try to teach VLMs with Reinforcement Learning. - RL cannot teach VLMs such planning ability, only 2.5% success rate with Qwen2.5-VL-7B. + With View Graph Distillation (our RL-Graph-SFT framework), 2.5% → 47.8% Below, we answer these questions: Q1. What are the failure modes? Q2. How can we make RL work? Q3. What has the model learned? Can we open up the model to see before/after? Can such spatial priors transfer to other view related tasks? Led by Kangrui Wang, great to work with Linjie (Lindsey) Li Zhengyuan Yang Shiqi Chen Zihan "Zenus" Wang Fei-Fei Li Jiajun Wu Leonidas Guibas, Lijuan Wang. A joint efforts with Stanford AI Lab Stanford Vision and Learning Lab Microsoft Research.

Manling Li

64,183 次观看 • 3 个月前

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵

Jim Fan

2,677,730 次观看 • 3 年前