❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers

📝 The paper  is available here:
https://nvlabs.github.io/GEAR-SONIC/

Our Patreon if you wish to support us: https://www.patreon.com/TwoMinutePapers

🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible:
Adam Bridges, Benji Rabhan, B Shang, Cameron Navor, Charles Ian Norman Venn, Christian Ahlin, Eric T, Fred R, Gordon Child, Juan Benet, Michael Tedder, Owen Skarpness, Richard Sundvall, Ryan Stankye, Shawn Becker, Steef, Taras Bobrovytsky, Tazaur Sagenclaw, Tybie Fitzhugh, Ueli Gallizzi
 
My research: https://cg.tuwien.ac.at/~zsolnai/
Thumbnail design: https://felicia.hu

#nvidia

Two Minute Papers's resource offers insights, tutorials, and resources for researchers and enthusiasts interested in computer science and artificial intelligence. Readers can learn about  research papers, breakthroughs, and trends in the field of AI. With concise summaries, analysis, and visualizations, Two Minute Papers provides  guidance and expertise for understanding complex research topics in a digestible format.

Two Minute Papers

NVIDIA's new AI system called Sonic is a lightweight multimodal robot controller trained on 100 million frames of human motion. It accepts diverse inputs — video, voice, music, or text — and translates them into robot motor commands using universal tokens. Despite being trained on 128 GPUs over 3 days, the resulting model has only ~42 million parameters and runs on a smartphone. Key innovations include a root trajectory spring model that dampens sudden commands to prevent robot self-injury, and the ability to understand whole-body movement without human-labeled action data. All models are being released for free. The project is led by Jim Fan at NVIDIA's humanoid robots lab.

NVIDIA’s New AI Changed Robotics Forever