Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

🚀 Introducing SoftMimic: Compliant humanoids for an interactive future — bringing humanoids into the real world 🤝 🔗 Current humanoids collapse when they touch the world — they can’t handle contact or deviation from their reference motion 😭 😎SoftMimic learns to deviate gracefully from motion references. It natively embraces...

pulkitology's profile picture

Pulkit Agrawal

18,025 subscribers

40,337 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

19 Kommentare

Profilbild von Pieter Abbeel
Pieter Abbeelvor 1 Jahr

impressive

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

😍Unlike today's stiff humanoids, soft mimic is robust to unexpected collisions 🔗

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

🤩Using the same motion reference, soft mimic picks up boxes of different sizes -- free generalization! 🔗

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

🙀🤯Soft mimic is extremely robust while still being safe and doing what you want the robot to do. 🔗

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

🥳Soft mimic is compliant enabling gentle interactions 🔗

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

🤯🤩Soft mimic can be quickly programmed for new tasks, zero-shot 🔗

Profilbild von James
Jamesvor 1 Jahr

It has invisible lats syndrome

Profilbild von Mengdi Xu
Mengdi Xuvor 1 Jahr

Very cool results. Congrats!

Profilbild von Sunil Gora
Sunil Goravor 1 Jahr

Impressive work! Isn't it learning the 'admittance control'?

Profilbild von Pulkit Agrawal
Pulkit Agrawalvor 1 Jahr

yes

Profilbild von czr
czrvor 1 Jahr

cool！

Profilbild von Sammy
Sammyvor 1 Jahr

Amazing

Profilbild von ζ Pedram ζ
ζ Pedram ζvor 1 Jahr

Would love to see code on this, but this is the most practical way to real scalability. Also minimum requirements for Kung Fu

Profilbild von Kaival Shah 🤔
Kaival Shah 🤔vor 1 Jahr

Super cool!

Profilbild von ζ Pedram ζ
ζ Pedram ζvor 1 Jahr

So nice

Profilbild von Michal Nauman
Michal Naumanvor 1 Jahr

Awesome work 🦾

Profilbild von Carlos DP 🤖🇺🇸
Carlos DP 🤖🇺🇸vor 1 Jahr

Very cool work! Similar to the goals of FALCON, but within a motion tracking context, I like it

Profilbild von Sarika saah🇺🇸
Sarika saah🇺🇸vor 7 Monaten

In the real world, true success lies in not falling over from bumps and touches, and SoftMimic has nailed this point. How easy is it to deploy this framework on robots like Unitree or Digit? 😌

Profilbild von Priyanka Agrawal
Priyanka Agrawalvor 11 Monaten

The future depends on robots.

Ähnliche Videos

Excited to share SoftMimic -- a new approach for learning compliant humanoid policies that interact gently with the world.
1:29

Excited to share SoftMimic -- a new approach for learning compliant humanoid policies that interact gently with the world.

Gabe Margolis

65,247 Aufrufe • vor 1 Jahr

🕸️ Introducing SPIDER — Scalable Physics-Informed Dexterous Retargeting! A dynamically feasible, cross-embodiment retargeting framework for BOTH humanoids 🤖 and dexterous hands ✋. From human motion → sim → real robots, at scale. 🔗 Website: 🧵 1/n
1:25

🕸️ Introducing SPIDER — Scalable Physics-Informed Dexterous Retargeting! A dynamically feasible, cross-embodiment retargeting framework for BOTH humanoids 🤖 and dexterous hands ✋. From human motion → sim → real robots, at scale. 🔗 Website: 🧵 1/n

Chaoyi Pan

66,077 Aufrufe • vor 11 Monaten

Holy…S😳 This isn’t fabric. It’s electronic skin,woven with dense fiber and textile sensors that can detect pressure, touch, deformation, and subtle contact changes in real time. Now imagine humanoid robots covered in it,especially on dexterous hands. humanoids could truly touch and understand the physical world and even humans. (from Shanghai JQ INDUSTRIES )
1:46

Holy…S😳 This isn’t fabric. It’s electronic skin,woven with dense fiber and textile sensors that can detect pressure, touch, deformation, and subtle contact changes in real time. Now imagine humanoid robots covered in it,especially on dexterous hands. humanoids could truly touch and understand the physical world and even humans. (from Shanghai JQ INDUSTRIES )

CyberRobo

174,143 Aufrufe • vor 7 Monaten

Today’s robots aren’t humanoids from sci-fi movies. They carry supplies, evacuate the wounded and fight. 2026: more than 28,000 robotic missions completed on the battlefield.
0:38

Today’s robots aren’t humanoids from sci-fi movies. They carry supplies, evacuate the wounded and fight. 2026: more than 28,000 robotic missions completed on the battlefield.

Defense of Ukraine

74,353 Aufrufe • vor 5 Monaten

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥
1:32

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥

Zhengyi “Zen” Luo

62,782 Aufrufe • vor 11 Monaten

Odyssey launches Odyssey-3, a foundation world model that sets a new state of the art on Physics-IQ Verified (66.1) for video-to-video physical simulation. The same pretrained model generates real-time interactive environments from prompts and adapts to control robots, humanoids, cars, and drones with limited task-specific data.
0:52

Odyssey launches Odyssey-3, a foundation world model that sets a new state of the art on Physics-IQ Verified (66.1) for video-to-video physical simulation. The same pretrained model generates real-time interactive environments from prompts and adapts to control robots, humanoids, cars, and drones with limited task-specific data.

Documenting AGI

38,907 Aufrufe • vor 22 Tagen

Humanoid robots shouldn't just follow pre-defined movements—they should perform! 🤖✨ Introducing UniAct: A unified model for multimodal motion generation and action streaming. Most humanoids are limited to pre-designed moves. UniAct changes the game by allowing robots to generate live action sequences from: 📝 Text instructions 🎶 Music rhythms 📉 Spatial trajectories 🔄 Cross-modal signals Whether it’s dancing to a beat or following a complex path, UniAct brings humanoid robots to life in real-time. 🚀 🔗 Project: 📄 Paper:
1:37

Humanoid robots shouldn't just follow pre-defined movements—they should perform! 🤖✨ Introducing UniAct: A unified model for multimodal motion generation and action streaming. Most humanoids are limited to pre-designed moves. UniAct changes the game by allowing robots to generate live action sequences from: 📝 Text instructions 🎶 Music rhythms 📉 Spatial trajectories 🔄 Cross-modal signals Whether it’s dancing to a beat or following a complex path, UniAct brings humanoid robots to life in real-time. 🚀 🔗 Project: 📄 Paper:

Siyuan Huang

10,688 Aufrufe • vor 9 Monaten

Unfortunately Frank, Rose, Bob and Gary the humanoids all just lost their jobs to a pair of faster, lower cost, more reliable robot arms. Superhuman robots will beat humanoids for every industrial application with scale.
0:55

Unfortunately Frank, Rose, Bob and Gary the humanoids all just lost their jobs to a pair of faster, lower cost, more reliable robot arms. Superhuman robots will beat humanoids for every industrial application with scale.

Simon Kalouche

480,428 Aufrufe • vor 5 Monaten

What does OpenMind do? Founder of OpenMind, Jan Liphardt, talks about our OM1 operating system for robots, our partners, and how robots interact in the real world. Elon Musk predicts 1 billion humanoids by 2040, so we'll need stablecoins to transition to a new machine economy.
2:26

What does OpenMind do? Founder of OpenMind, Jan Liphardt, talks about our OM1 operating system for robots, our partners, and how robots interact in the real world. Elon Musk predicts 1 billion humanoids by 2040, so we'll need stablecoins to transition to a new machine economy.

OpenMind

26,876 Aufrufe • vor 1 Jahr

Command humanoids *directly* with natural language? Introducing LangWBC, a generative, end-to-end policy that turns natural language into real-world whole-body humanoid control! 💬→🦿Smooth, robust, surprisingly intuitive! See more 👉 #RSS2025
0:45

Command humanoids *directly* with natural language? Introducing LangWBC, a generative, end-to-end policy that turns natural language into real-world whole-body humanoid control! 💬→🦿Smooth, robust, surprisingly intuitive! See more 👉 #RSS2025

Zhongyu Li

14,241 Aufrufe • vor 1 Jahr

Ever wonder how we got our humanoids to walk naturally using a motion tracker? Here is the secret: We have a world-class, game-engine-ready, real-time motion generator running under the hood! Graphics 🤝 Robotics Check out the full thread here 👇
0:27

Ever wonder how we got our humanoids to walk naturally using a motion tracker? Here is the secret: We have a world-class, game-engine-ready, real-time motion generator running under the hood! Graphics 🤝 Robotics Check out the full thread here 👇

Zhengyi “Zen” Luo

15,303 Aufrufe • vor 6 Monaten

Lightberry () lets humanoids be useful out of the box. They integrate with major manufacturers like Unitree and Booster to make their robots speak, read the room, and navigate the world - just like in Star Wars. Congrats on the launch, Stephan Koenigstorfer & !
1:11

Lightberry () lets humanoids be useful out of the box. They integrate with major manufacturers like Unitree and Booster to make their robots speak, read the room, and navigate the world - just like in Star Wars. Congrats on the launch, Stephan Koenigstorfer & !

Y Combinator

56,612 Aufrufe • vor 1 Jahr

X-Humanoid unveiled TG-VLA, which it calls the world's first whole-body VLA framework for full-size humanoid robots. The goal: move humanoids past "mobile dual-arm systems" to human-like, whole-body coordinated operation. Built on three technologies: - HEX: an open-source full-size humanoid VLA that learns across embodiments and coordinates the whole body. Claimed SOTA on real-world whole-body manipulation - HAF-VLA: hierarchical action flow that decomposes whole-body motion into subtasks. Best or joint-best on 6 benchmarks - DSRL-DCT: online RL for high-DoF humanoids in a compressed latent space. 100% success on mobile manipulation HEX is open-source.
2:48

X-Humanoid unveiled TG-VLA, which it calls the world's first whole-body VLA framework for full-size humanoid robots. The goal: move humanoids past "mobile dual-arm systems" to human-like, whole-body coordinated operation. Built on three technologies: - HEX: an open-source full-size humanoid VLA that learns across embodiments and coordinates the whole body. Claimed SOTA on real-world whole-body manipulation - HAF-VLA: hierarchical action flow that decomposes whole-body motion into subtasks. Best or joint-best on 6 benchmarks - DSRL-DCT: online RL for high-DoF humanoids in a compressed latent space. 100% success on mobile manipulation HEX is open-source.

The Humanoid Hub

19,563 Aufrufe • vor 3 Monaten

From dogs to mini companion humanoids, China robot makers face challenge of putting robots to work
2:01

From dogs to mini companion humanoids, China robot makers face challenge of putting robots to work

Reuters

130,499 Aufrufe • vor 2 Monaten

Scene from K-ICT WEEK in BUSAN 2026 at BEXCO. Interactive quadruped and humanoid robots are among the highlights of the robotics zone. Visitors, including children, can watch and interact firsthand with emerging AI technologies. The technology in that video represents two branches of physical AI that were advancing quickly in South Korea in 2026: four-legged robots (quadrupeds) and interactive humanoids.
0:21

Scene from K-ICT WEEK in BUSAN 2026 at BEXCO. Interactive quadruped and humanoid robots are among the highlights of the robotics zone. Visitors, including children, can watch and interact firsthand with emerging AI technologies. The technology in that video represents two branches of physical AI that were advancing quickly in South Korea in 2026: four-legged robots (quadrupeds) and interactive humanoids.

Theana

41,433 Aufrufe • vor 1 Monat

Amazon is training humanoids to move boxes. Makes sense! OmniRetarget is a data generation engine that enables complex loco-manipulation for humanoids. It uses offline retargeting from human MoCap datasets and augments data from single demos to produce 8 hours of trajectories that train RL policies in simulation, transferred zero-shot to real robots. Amazon FAR is the primary affiliation for the paper:
1:37

Amazon is training humanoids to move boxes. Makes sense! OmniRetarget is a data generation engine that enables complex loco-manipulation for humanoids. It uses offline retargeting from human MoCap datasets and augments data from single demos to produce 8 hours of trajectories that train RL policies in simulation, transferred zero-shot to real robots. Amazon FAR is the primary affiliation for the paper:

The Humanoid Hub

106,447 Aufrufe • vor 1 Jahr