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。

🤔 Can we train one VLA policy to control multi-robot teams without any explicit communication? ✨ Introducing CHORUS: a single policy for decentralized, multi-embodiment collaboration 🧵⬇️

Ria Doshi

1,030 subscribers

61,245 次观看 • 3 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

9 条评论

Ria Doshi 的头像

Ria Doshi3 个月前

What makes decentralized, multi-robot collaboration hard? 🔹each robot only sees its own view of the scene 🔹teams can have diverse embodiments 🔹training cost balloons with team size

Ria Doshi 的头像

Ria Doshi3 个月前

So how do we train CHORUS? (1) Split team-wide demos into per-robot episodes (robot obs + identity prompt → actions) (2) Finetune one, shared π0.5 backbone on these episodes = one policy that can control multi-embodied teams, with a learned model of every teammate's actions

Ria Doshi 的头像

Ria Doshi3 个月前

At deployment, each robot runs its own copy of CHORUS, conditioned only on its own observations. We adjust chunk sizes w.r.t. robot control frequency. ✅ No inference time comms, shared camera views, or online alignment: coordination emerges entirely from the learned policy.

Ria Doshi 的头像

Ria Doshi3 个月前

CHORUS outperforms from-scratch diffusion by 64% points across basket lifting, tape measurement, book handovers, and more! ✨ Pretrained visuomotor priors transfer surprisingly well, even though multi-robot collaboration never appears in pretraining data. [Result 1/4]

Ria Doshi 的头像

Ria Doshi3 个月前

Sharing one policy across the team is not just more efficient; it makes robots more reactive to each other. We find that CHORUS nearly 2x’s teammate reactivity: robots adapt to each other’s behavior instead of blindly plowing through the task ⬇️ [2/4]

Ria Doshi 的头像

Ria Doshi3 个月前

We also compare to a centralized VLA that sees a team-wide obs and outputs all robot actions at once. In theory, an upper bound. Surprisingly, CHORUS beats it. Centralization inflates the input+action space and drifts from the VLA's pre-training distribution. Decentralization keeps those priors intact. [3/4]

Ria Doshi 的头像

Ria Doshi3 个月前

And last but not least, CHORUS scales seamlessly to 3-robot teams. Check out three mobile manipulators transporting laundry… in my apartment! 🥹 [4/4]

Ria Doshi 的头像

Ria Doshi3 个月前

Check out our website for more! 🌐 📄 Grateful for my amazing collaborators: @TianGao_19, @_anniechen_ , @chelseabfinn, @leto__jean

clankr 的头像

clankr3 个月前

Really impressed by the teamwork of these 3 robots. I'm curious, do world action models support such collaboration systems, or are they fundamentally infeasible for this kind of interaction?

相关视频

🤔 Can we train one policy to control a wide range of robots, from drones to quadrupeds, navigators to bimanual manipulators, and more? 🦾Introducing CrossFormer: a single policy that can perform manipulation, navigation, aviation, and locomotion:

🤔 Can we train one policy to control a wide range of robots, from drones to quadrupeds, navigators to bimanual manipulators, and more? 🦾Introducing CrossFormer: a single policy that can perform manipulation, navigation, aviation, and locomotion:

Ria Doshi

85,220 次观看 • 2 年前

Excited to share RoCo: Dialectic Multi-Robot Collaboration with Large Language Models. We propose a novel approach to multi-robot collaboration that leverages LLMs for both high-level communication and low-level path planning. w/ Shreeya Jain, Shuran Song

Excited to share RoCo: Dialectic Multi-Robot Collaboration with Large Language Models. We propose a novel approach to multi-robot collaboration that leverages LLMs for both high-level communication and low-level path planning. w/ Shreeya Jain, Shuran Song

Mandi Zhao

88,722 次观看 • 3 年前

Tired of teleoperating your robots? We built a way to scale robot datasets without teleop, dynamic simulation, or even robot hardware. Just one smartphone scan + one human hand demo video → thousands of diverse robot trajectories. Trainable by diffusion policy and VLA models as-is. Introducing: Real2Render2Real 👉

Tired of teleoperating your robots? We built a way to scale robot datasets without teleop, dynamic simulation, or even robot hardware. Just one smartphone scan + one human hand demo video → thousands of diverse robot trajectories. Trainable by diffusion policy and VLA models as-is. Introducing: Real2Render2Real 👉

Max Fu

69,381 次观看 • 1 年前

Multi launches today! Multi: Multiplayer Collaboration for macOS - Shared control for any app or display - Drawing & deep links for faster communication - Native app; crafted to feel like part of macOS Get early access:

Multi launches today! Multi: Multiplayer Collaboration for macOS - Shared control for any app or display - Drawing & deep links for faster communication - Native app; crafted to feel like part of macOS Get early access:

Multi

811,945 次观看 • 3 年前

With the recent progress in large-scale multi-task robot training, how can we advance the real-world deployment of multi-task robot fleets? Introducing Sirius-Fleet✨, a multi-task interactive robot fleet learning framework with 𝗩𝗶𝘀𝘂𝗮𝗹 𝗪𝗼𝗿𝗹𝗱 𝗠𝗼𝗱𝗲𝗹𝘀! 🌍 #CoRL2024

With the recent progress in large-scale multi-task robot training, how can we advance the real-world deployment of multi-task robot fleets? Introducing Sirius-Fleet✨, a multi-task interactive robot fleet learning framework with 𝗩𝗶𝘀𝘂𝗮𝗹 𝗪𝗼𝗿𝗹𝗱 𝗠𝗼𝗱𝗲𝗹𝘀! 🌍 #CoRL2024

Huihan Liu

28,080 次观看 • 1 年前

Tired of collecting demonstrations all day to train your robot? Introducing MimicGen, an autonomous data generation system for robotics. Using just 200 human demos we generated a large multi-task dataset of 50K demos! #CoRL2023 #NVIDIAResearch 👇 🧵 1/

Tired of collecting demonstrations all day to train your robot? Introducing MimicGen, an autonomous data generation system for robotics. Using just 200 human demos we generated a large multi-task dataset of 50K demos! #CoRL2023 #NVIDIAResearch 👇 🧵 1/

Ajay Mandlekar

93,696 次观看 • 2 年前

What if robots could think longer on harder problems without saying a single word?🤔 We introduce RD-VLA (Recurrent-Depth VLA): a latent, iterative reasoning architecture for robot control. ❌No Chain-of-Thought tokens. ❌No extra memory overhead. ✅Just reasoning—directly in latent space. 🧠🤖 Project page: 👇🧵

What if robots could think longer on harder problems without saying a single word?🤔 We introduce RD-VLA (Recurrent-Depth VLA): a latent, iterative reasoning architecture for robot control. ❌No Chain-of-Thought tokens. ❌No extra memory overhead. ✅Just reasoning—directly in latent space. 🧠🤖 Project page: 👇🧵

Jiafei Duan

13,949 次观看 • 7 个月前

Can robots self-improve by collecting data autonomously🤖? Introducing SOAR: a system for large-scale autonomous data collection 🚀 and autonomous improvement📈of a multi-task language-conditioned policy in diverse scenes without human interventions .

Can robots self-improve by collecting data autonomously🤖? Introducing SOAR: a system for large-scale autonomous data collection 🚀 and autonomous improvement📈of a multi-task language-conditioned policy in diverse scenes without human interventions .

Paul Zhou

47,729 次观看 • 2 年前

We are excited to share new experiments with AgiBot @AgiBot_zhiyuan on multi-task, multi-embodiment VLAs! With one model that can perform many tasks with both two-finger grippers and multi-fingered hands, we take another step toward one model for all robots and tasks.

We are excited to share new experiments with AgiBot @AgiBot_zhiyuan on multi-task, multi-embodiment VLAs! With one model that can perform many tasks with both two-finger grippers and multi-fingered hands, we take another step toward one model for all robots and tasks.

Physical Intelligence

76,321 次观看 • 1 年前

🚀 First step to unlocking Generalist Robots! Introducing 🤖LAPA🤖, a new SOTA open-sourced 7B VLA pretrained without using action labels. 💪SOTA VLA trained with Open X (outperforming OpenVLA on cross and multi embodiment) 😯LAPA enables learning from human videos, unlocking potential for robotic foundation model ❗Over 30x pretraining efficiency for VLA training 🤗Code and checkpoints are all open-sourced!

🚀 First step to unlocking Generalist Robots! Introducing 🤖LAPA🤖, a new SOTA open-sourced 7B VLA pretrained without using action labels. 💪SOTA VLA trained with Open X (outperforming OpenVLA on cross and multi embodiment) 😯LAPA enables learning from human videos, unlocking potential for robotic foundation model ❗Over 30x pretraining efficiency for VLA training 🤗Code and checkpoints are all open-sourced!

Seonghyeon Ye

33,352 次观看 • 1 年前

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Max Fu

40,466 次观看 • 2 年前

Introducing OFT—an Optimized Fine-Tuning recipe for VLAs! Fine-tuning OpenVLA w/ OFT, we see: -25-50x faster inference ⚡️ -SOTA 97.1% avg SR in LIBERO 💪 -high-freq control w/ 7B model on real bimanual robot -outperforms π₀, RDT-1B, DiT Policy, MDT, Diffusion Policy, ACT 🧵👇

Introducing OFT—an Optimized Fine-Tuning recipe for VLAs! Fine-tuning OpenVLA w/ OFT, we see: -25-50x faster inference ⚡️ -SOTA 97.1% avg SR in LIBERO 💪 -high-freq control w/ 7B model on real bimanual robot -outperforms π₀, RDT-1B, DiT Policy, MDT, Diffusion Policy, ACT 🧵👇

Moo Jin Kim

84,359 次观看 • 1 年前

Rollouts in the real world are slow and expensive. What if we could rollout trajectories entirely inside a world model (WM)? Introducing 🚀Ctrl-World🚀, a generative manipulation WM that can interact with advanced VLA policy in imagination. 🧵1/6

Rollouts in the real world are slow and expensive. What if we could rollout trajectories entirely inside a world model (WM)? Introducing 🚀Ctrl-World🚀, a generative manipulation WM that can interact with advanced VLA policy in imagination. 🧵1/6

Yanjiang Guo

96,028 次观看 • 11 个月前

If you have a policy that uses diffusion/flow (e.g. diffusion VLA), you can run RL where the actor chooses the noise, which is then denoised by the policy to produce an action. This method, which we call diffusion steering (DSRL), leads to a remarkably efficient RL method! 🧵👇

If you have a policy that uses diffusion/flow (e.g. diffusion VLA), you can run RL where the actor chooses the noise, which is then denoised by the policy to produce an action. This method, which we call diffusion steering (DSRL), leads to a remarkably efficient RL method! 🧵👇

Sergey Levine

153,174 次观看 • 1 年前

X Square Robot has introduced TwinDEX: a co-designed pair of three-finger, 9-DoF dexterous hands, one wearable for data collection, one matched robot end effector for deployment. The wearable and the robot hand share the same kinematics, geometry, materials, look, and sensors, so human demos map straight to the robot without retargeting. Human data collection improves throughput 5.3x compared to teleoperation, and a policy can be trained with a few hundred episodes without any robot data.

X Square Robot has introduced TwinDEX: a co-designed pair of three-finger, 9-DoF dexterous hands, one wearable for data collection, one matched robot end effector for deployment. The wearable and the robot hand share the same kinematics, geometry, materials, look, and sensors, so human demos map straight to the robot without retargeting. Human data collection improves throughput 5.3x compared to teleoperation, and a policy can be trained with a few hundred episodes without any robot data.

The Humanoid Hub

277,022 次观看 • 27 天前

Humanoids need a single, generalist control policy for all of their physical tasks, not a new one for every new chore or demo. A policy for walking can't dance. A policy for dancing can't support mowing the lawn. We need to scale up humanoid control for diverse behaviors, just like we did with language and video models. Watch this humanoid, supported by a single control policy, to start its day, mow the lawn, jog, dance, and even hold the elevator for a colleague. This is the "System 1" for humanoids—a robust, natural humanoid controller that will support general-purpose autonomy. Project Page: #Robotics #AI #NVIDIA #SONIC #FoundationModel 🤖🔥

Humanoids need a single, generalist control policy for all of their physical tasks, not a new one for every new chore or demo. A policy for walking can't dance. A policy for dancing can't support mowing the lawn. We need to scale up humanoid control for diverse behaviors, just like we did with language and video models. Watch this humanoid, supported by a single control policy, to start its day, mow the lawn, jog, dance, and even hold the elevator for a colleague. This is the "System 1" for humanoids—a robust, natural humanoid controller that will support general-purpose autonomy. Project Page: #Robotics #AI #NVIDIA #SONIC #FoundationModel 🤖🔥

Zhengyi “Zen” Luo

15,982 次观看 • 10 个月前

Introducing Omnigrasp: Grasping Diverse Objects with Simulated Humanoids. With Omnigrasp, we show that we can control a humanoid equipped with dexterous hands to grasp diverse objects (>1200) and follow diverse trajectories, with one policy! 🌐: 📜:

Introducing Omnigrasp: Grasping Diverse Objects with Simulated Humanoids. With Omnigrasp, we show that we can control a humanoid equipped with dexterous hands to grasp diverse objects (>1200) and follow diverse trajectories, with one policy! 🌐: 📜:

Zhengyi “Zen” Luo

92,214 次观看 • 2 年前