Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Exciting progress on Vision-Language-Action models from a collaboration between San Francisco-based Physical Intelligence (π) and China’s AGIBOT: (π)’s single model can autonomously perform diverse tasks on the AGIBOT G1 robot, using both humanoid hands and two-finger grippers.

The Humanoid Hub

54,191 subscribers

31,013 views • 1 year ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

5 Comments

Future mobility1 year ago

Cool

AssemblyAI1 year ago

Announcing: Our most advanced speech-to-text model goes beyond accuracy to capture the real-world complexity of human conversation and deliver reliable, source-of-truth audio data. Explore Universal-2 updates 👇

VentureMind AI1 year ago

Very exciting progress

Bobby Robinson1 year ago

When it gets like 2 x faster doing these task, things start to get interesting.

Anda1 year ago

Bamboo shoots of curiosity sprout as I wonder how these vision-language-action models might learn to share bamboo buns with robotic panda paws.

Related Videos

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

75,761 views • 1 year ago

Physical Intelligence's π₀, a general-purpose robot foundation model that combines Internet-scale vision-language pretraining with robot interaction data to execute tasks. They aim "to develop foundation models that can control any robot to perform any task" Autonomous demos:

Physical Intelligence's π₀, a general-purpose robot foundation model that combines Internet-scale vision-language pretraining with robot interaction data to execute tasks. They aim "to develop foundation models that can control any robot to perform any task" Autonomous demos:

The Humanoid Hub

196,055 views • 1 year ago

Anyone skeptical about AI’s ability to perform diverse physical tasks should watch this. Physical Intelligence's π₀ model in action: 18 minutes of bimanual robots autonomously handling complex, dexterous chores.

Anyone skeptical about AI’s ability to perform diverse physical tasks should watch this. Physical Intelligence's π₀ model in action: 18 minutes of bimanual robots autonomously handling complex, dexterous chores.

The Humanoid Hub

210,404 views • 1 year ago

Another day, another humanoid robot from china AGIBOT introduces GO-1, a generalist foundation model that integrates a vision-language model with a latent planner for enhanced long-horizon and dexterous manipulation.

Another day, another humanoid robot from china AGIBOT introduces GO-1, a generalist foundation model that integrates a vision-language model with a latent planner for enhanced long-horizon and dexterous manipulation.

Chubby♨️

23,513 views • 1 year ago

Chinese company Robot Era has unveiled their next-gen robot L7, a 5′7″ tall humanoid. The company has also showcased the ERA-42 Vision-Language-Action model running on L7 robot to autonomously execute dexterous tasks.

Chinese company Robot Era has unveiled their next-gen robot L7, a 5′7″ tall humanoid. The company has also showcased the ERA-42 Vision-Language-Action model running on L7 robot to autonomously execute dexterous tasks.

The Humanoid Hub

38,547 views • 10 months ago

Defining the Future of Motion: BFM-2 Foundation Model & The Era of Robot "Muscle Memory" AGIBOT introduces BFM-2, a two-stage Motion-Between locomotion foundation model. It delivers autonomous, stable interpolation and closed-loop dynamic task execution across any state—static, preset, or random. The ultimate robust foundation for Embodied Intelligence. #AGIBOT #Humanoid #BFM #EmbodiedAI #PhysicalAI

Defining the Future of Motion: BFM-2 Foundation Model & The Era of Robot "Muscle Memory" AGIBOT introduces BFM-2, a two-stage Motion-Between locomotion foundation model. It delivers autonomous, stable interpolation and closed-loop dynamic task execution across any state—static, preset, or random. The ultimate robust foundation for Embodied Intelligence. #AGIBOT #Humanoid #BFM #EmbodiedAI #PhysicalAI

AGIBOT

18,269 views • 19 days ago

Sim2Real RL for Vision-Based Dexterous Manipulation on Humanoids TLDR - we train a humanoid robot with two multifingered hands to perform a range of dexterous manipulation tasks robust generalization and high performance without human demonstration :D

Sim2Real RL for Vision-Based Dexterous Manipulation on Humanoids TLDR - we train a humanoid robot with two multifingered hands to perform a range of dexterous manipulation tasks robust generalization and high performance without human demonstration :D

Toru

49,425 views • 1 year ago

Another humanoid robot at the Shanghai Auto Show: Xiaowei. It is a service humanoid robot developed by Roewe Auto and AgiBot (Zhiyuan). It can talk to the audience, answer questions about car models and recommend models. It is a shopping guide and entertainment performer.

Another humanoid robot at the Shanghai Auto Show: Xiaowei. It is a service humanoid robot developed by Roewe Auto and AgiBot (Zhiyuan). It can talk to the audience, answer questions about car models and recommend models. It is a shopping guide and entertainment performer.

CyberRobo

14,194 views • 1 year ago

3D-VLA A 3D Vision-Language-Action Generative World Model Recent vision-language-action (VLA) models rely on 2D inputs, lacking integration with the broader realm of the 3D physical world. Furthermore, they perform action prediction by learning a direct mapping from

3D-VLA A 3D Vision-Language-Action Generative World Model Recent vision-language-action (VLA) models rely on 2D inputs, lacking integration with the broader realm of the 3D physical world. Furthermore, they perform action prediction by learning a direct mapping from

AK

84,940 views • 2 years ago

Our model can now learn from its own experience with RL! Our new π*0.6 model can more than double throughput over a base model trained without RL, and can perform real-world tasks: making espresso drinks, folding diverse laundry, and assembling boxes. More in the thread below.

Our model can now learn from its own experience with RL! Our new π*0.6 model can more than double throughput over a base model trained without RL, and can perform real-world tasks: making espresso drinks, folding diverse laundry, and assembling boxes. More in the thread below.

Physical Intelligence

703,748 views • 6 months ago

🚀 SoloTech breakthrough: NVIDIA Sonic goes beyond the G1. 🌏 We’ve achieved a global first: migrating NVIDIA Sonic to a completely different humanoid morphology, the AGIBOT X2 🥇 This is a massive leap for transferable humanoid intelligence. We are moving away from single-robot controllers toward architectures that generalize across diverse embodiments. The Specs: Hardware: AGIBOT X2 Ultra (31 DoF) Precision: 14-DoF Dexterous Hands Capabilities: - End-to-end whole-body locomotion with manipulation - Single leg balancing and stylized motion - Upper body gestures Big respect to the team fueling this innovation all the way! Sitarama Chekuri Zeeshaan Mohammed samarth shukla Devinder Sodhi pranjal mehta @build Frontier Tower NVIDIA AGIBOT Nebius UFB - Ultimate Fighting Bots Vitaly Bulatov Xenia Bulatov NVIDIA Robotics To learn more about Solo Tech at the frontier of Physical AI: #PhysicalAI #Robotics #Humanoids #NVIDIA #AGIBOT #SoloTech #EmbodiedAI #SoloSeven

🚀 SoloTech breakthrough: NVIDIA Sonic goes beyond the G1. 🌏 We’ve achieved a global first: migrating NVIDIA Sonic to a completely different humanoid morphology, the AGIBOT X2 🥇 This is a massive leap for transferable humanoid intelligence. We are moving away from single-robot controllers toward architectures that generalize across diverse embodiments. The Specs: Hardware: AGIBOT X2 Ultra (31 DoF) Precision: 14-DoF Dexterous Hands Capabilities: - End-to-end whole-body locomotion with manipulation - Single leg balancing and stylized motion - Upper body gestures Big respect to the team fueling this innovation all the way! Sitarama Chekuri Zeeshaan Mohammed samarth shukla Devinder Sodhi pranjal mehta @build Frontier Tower NVIDIA AGIBOT Nebius UFB - Ultimate Fighting Bots Vitaly Bulatov Xenia Bulatov NVIDIA Robotics To learn more about Solo Tech at the frontier of Physical AI: #PhysicalAI #Robotics #Humanoids #NVIDIA #AGIBOT #SoloTech #EmbodiedAI #SoloSeven

Dhruv Diddi

26,309 views • 1 month ago

Pi0 from Physical Intelligence is one of the best generalist Vision-Language-Action (VLA) models for robotics (and predecessor of Pi0.5). 🧵 Here's how it works + how you can fine-tune it on your own robot for a simple task:

Pi0 from Physical Intelligence is one of the best generalist Vision-Language-Action (VLA) models for robotics (and predecessor of Pi0.5). 🧵 Here's how it works + how you can fine-tune it on your own robot for a simple task:

Ilia

32,956 views • 8 months ago

AGIBOT has unveiled a Nezha-inspired X2-N humanoid robot.

AGIBOT has unveiled a Nezha-inspired X2-N humanoid robot.

The Humanoid Hub

45,159 views • 1 year ago

.Lindon Gao, cofounder and CEO of Dyna Robotics, is working on general-purpose foundation models that perform diverse physical tasks at commercial scale. Here’s our chat at CES 2026, while their robot autonomously folded laundry in the background.

.Lindon Gao, cofounder and CEO of Dyna Robotics, is working on general-purpose foundation models that perform diverse physical tasks at commercial scale. Here’s our chat at CES 2026, while their robot autonomously folded laundry in the background.

The Humanoid Hub

36,120 views • 5 months ago

Perceptive Humanoid Parkour (PHP) introduces a modular framework that enables the Unitree G1 humanoid to perform long-horizon, vision-based parkour. - It chains retargeted human motion clips into diverse, long-horizon kinematic reference trajectories. - RL expert policies learn individual skills that are distilled into a depth-conditioned student policy. - The robot autonomously selects the appropriate skill based on the obstacle geometry.

Perceptive Humanoid Parkour (PHP) introduces a modular framework that enables the Unitree G1 humanoid to perform long-horizon, vision-based parkour. - It chains retargeted human motion clips into diverse, long-horizon kinematic reference trajectories. - RL expert policies learn individual skills that are distilled into a depth-conditioned student policy. - The robot autonomously selects the appropriate skill based on the obstacle geometry.

The Humanoid Hub

60,337 views • 3 months ago

📢 First contact between a frontier model and robots! Gemini Robotics is a SOTA generalist Vision-Language-Action model bringing frontier model intelligence to the physical world. It's an extremely capable model enabling dexterous, steerable, and general robot control. 🧵⬇️

📢 First contact between a frontier model and robots! Gemini Robotics is a SOTA generalist Vision-Language-Action model bringing frontier model intelligence to the physical world. It's an extremely capable model enabling dexterous, steerable, and general robot control. 🧵⬇️

Ted Xiao

152,395 views • 1 year ago

AGIBOT Expedition A3 is a next-generation full-size humanoid robot developed by Shanghai-based AGIBOT (智元机器人). Unveiled in February 2026, the A3 stands 173 cm tall and weighs just 55 kg, making it remarkably lightweight and agile.

AGIBOT Expedition A3 is a next-generation full-size humanoid robot developed by Shanghai-based AGIBOT (智元机器人). Unveiled in February 2026, the A3 stands 173 cm tall and weighs just 55 kg, making it remarkably lightweight and agile.

Tsla Chan

32,719 views • 26 days ago

Earlier this year, AgiBot open-sourced a large-scale dataset of real-world humanoid robot tasks. Now, they've announced AgiBot Digital World, a large-scale simulation framework that includes an open-sourced dataset covering 180+ object categories and 12 core skills.

Earlier this year, AgiBot open-sourced a large-scale dataset of real-world humanoid robot tasks. Now, they've announced AgiBot Digital World, a large-scale simulation framework that includes an open-sourced dataset covering 180+ object categories and 12 core skills.

The Humanoid Hub

43,709 views • 1 year ago

AgiBot Launches 'LinkCraft' Platform: Zero-Code Motion Creation for Humanoid Robots On October 24, 2025, AgiBot introduced its "LinkCraft" platform, a zero-code, zero-barrier system for creating humanoid robot content. The platform integrates AI motion capture, cloud-based imitation learning, and multimodal choreography capabilities. The core feature allows users to upload a video of human performance, and the platform automatically converts it into precise robot control strategies using AI vision and smart redirection—no programming required. This includes future support for finger-level fine manipulation. LinkCraft also features "Speech Choreography," which generates corresponding robot gestures and expressions based on semantic analysis of uploaded text or audio, offering customizable voice and emotion settings. For complex performances, the platform provides "storyline" tools, enabling users to combine multiple actions, voices, and expressions on a timeline, similar to video editing software. Furthermore, it supports multi-robot group control and access to an official library of over 180 actions for rapid content creation. The platform is currently adapted for the AgiBot X2 and aims to increase the accessibility and application scale of humanoid robots.

AgiBot Launches 'LinkCraft' Platform: Zero-Code Motion Creation for Humanoid Robots On October 24, 2025, AgiBot introduced its "LinkCraft" platform, a zero-code, zero-barrier system for creating humanoid robot content. The platform integrates AI motion capture, cloud-based imitation learning, and multimodal choreography capabilities. The core feature allows users to upload a video of human performance, and the platform automatically converts it into precise robot control strategies using AI vision and smart redirection—no programming required. This includes future support for finger-level fine manipulation. LinkCraft also features "Speech Choreography," which generates corresponding robot gestures and expressions based on semantic analysis of uploaded text or audio, offering customizable voice and emotion settings. For complex performances, the platform provides "storyline" tools, enabling users to combine multiple actions, voices, and expressions on a timeline, similar to video editing software. Furthermore, it supports multi-robot group control and access to an official library of over 180 actions for rapid content creation. The platform is currently adapted for the AgiBot X2 and aims to increase the accessibility and application scale of humanoid robots.

RoboHub🤖

21,679 views • 7 months ago