Robot learning in action at the @UW. 👏
More than 60 researchers joined us for NVIDIA Robotics Day this week to explore NVIDIA Cosmos, Isaac GR00T, Isaac Lab, along with the Newton physics engine.
They also heard from NVIDIA researchers and got hands-on experience training virtual robots to manipulate objects in simulation. 🦾
Thanks for joining us. 💚
🤖 Scaling physical AI with human experience.
NVIDIA brings 1M hours of human demonstrations to the robotics community, with 100K hours open sourced, featuring egocentric video, head and hand tracking, and language annotations.
Explore the dataset on @HuggingFace and start building. 👇
This or that, robotics edition 🤖 What will senior research manager @UmarIqb pick?
The future of physical AI goes beyond what’s onboard. 🤖
Physical AI systems need reliable connectivity and distributed AI infrastructure to complement on-device computing. 6G will be the connective tissue for billions of physical AI devices, including robots.
Explore what 6G will bring to robotics at #NVIDIAGTC Washington D.C. 🔗 nvda.ws/3WQ4iQI
Build Training-Ready Datasets in Two Prompts Featuring Encord and Voxel51 | Cosmos Labs teapaw.org/i/broadcasts/2nQqvlbLO…
From Human Demonstrations to Robot Training Data With RLWRLD | Cosmos Labs teapaw.org/i/broadcasts/2kEqrOvzp…
Safety is essential to bringing robots and autonomous vehicles into the physical world. 🦺
Join NVIDIA and @ansidotorg at #NVIDIAGTC D.C. to explore how reasoning, simulation, real-world testing, and industry standards help advance physical AI safety from development through deployment.
🎙️Featuring NVIDIA’s Marco Pavone and Riccardo Mariani, and ANSI’s Laurie E. Locascio.
📅 Wednesday December 2 @ 4:30 PM EST
Add to schedule 👉 nvda.ws/4eLfT9S
Have questions about safety for robotics and autonomous vehicles?
Join us in person to ask the experts. 💬 nvidia.com/gtc/dc/session-ca…
What’s driving the next wave of physical AI? See for yourself at #NVIDIAGTC in Washington DC.
Learn how open models, simulation, and accelerated computing are advancing robotics and autonomous vehicles.
Explore sessions and build your agenda 👉 nvda.ws/3WSWIoq
Change the lighting and tabletop without recording the robot task again. 🤖
NVIDIA Cosmos Data Factory augments cube-manipulation footage with yellow lighting, dim lighting, and a red tabletop.
The workflow checks the generated clips, retries failed variations, then adds masks, tracks, and captions for training.
Explore video augmentation for robotics → nvda.ws/4zSHheC
Bring your robot data questions to Cosmos Labs. 🤖
NVIDIA, Encord and @Voxel51 go live Nov. 19 at 11 a.m. PT with Cosmos Data Factory demos, robotic inspection use cases, and Q&A.
Set a reminder 👉 youtube.com/watch?v=c5or9gzM…
Accelerate Visual AI Agents on NVIDIA Jetson with VSS 3.3 and Cosmos teapaw.org/i/broadcasts/2AMLNKEeX…
Human motion is a starting point for robot learning. 🦿
NVIDIA SOMA Retargeter maps human motion to different humanoid robots, accounting for their proportions, joint structures and range of motion.
Developers can configure their robot, refine movement interactively and process motion collections into reference trajectories for downstream policy training.
Explore the workflow ➡️ nvda.ws/4jm1LGy
NVIDIA Robotics retweeted
Wondering what’s in store at #NVIDIAGTC D.C.? 👀
Look no further. We’ve got your week covered.
Add physical AI and robotics sessions to your agenda 👉 nvda.ws/4zTERwi
From skepticism to possibility. Hear what changed @Ken_Goldberg’s outlook.👇
A year ago @Ken_Goldberg was a skeptic. At CoRL 2026 he told me why that changed:
• Agentic robotics as a paradigm shift
• Why factories won't adopt black box policies
• Bottom-up vs general-purpose robots
• Humanoids
• Open-source work with @NVIDIARobotics
Full video: youtu.be/YnaGXVEzapo
20 million downloads for NVIDIA Cosmos on @HuggingFace! 🎉
To the developers and researchers building robots, autonomous vehicles and vision AI agents with open world models: thank you for experimenting, sharing feedback and building with us.
Tell us what you're building with Cosmos next! 💚
Make a robot arm feel weightless. 🦾
With NVIDIA Holoscan 5, build a gravity-compensation loop that lets a @SeeedStudio reBot Arm support its own weight while you guide it by hand. Then measure latency and jitter with NVIDIA Nsight Systems.
Follow the step-by-step tutorial 👉 nvda.ws/4AFniAa
What if one delivery run could become a whole set of edge-case tests?
@CocoRobotics’ robots make deliveries through platforms like @UberEats, @DoorDash, Deliveroo, and @woltapp.
Using NVIDIA Cosmos, Coco turns real fleet video into new variations of a scene, so its robots can practice handling rare incidents before encountering them on the street.
Try it out 🔗 nvda.ws/4C37rN6







