The whole dayThu 19 Nov 2026
AI Compass
WeeklyArchiveAbout

NVIDIA open-sources MotionBricks real-time motion toolkit

Apache 2.0 Python runtime on NVIDIA Warp, Newton and TensorRT with commercial OpenUSD SOMA and Unitree G1 checkpoints, released on 19 November 2026.

Glowing green NVIDIA eye mark projecting cyan pose-keyframe arcs from a vaulting human silhouette over modular blocks to a running humanoid robot
Image: AI Compass illustration (NVIDIA marks)

NVIDIA released MotionBricks (NVlabs/motionbricks and nvidia/MotionBricks on Hugging Face) on 19 November 2026, open-sourcing the real-time latent generative motion framework from its SIGGRAPH 2026 paper (arXiv:2604.24833). Trained on more than 350,000 BONES-SEED motion-capture clips, a single neural backbone reaches 15,000 frames per second at 2 ms latency on TensorRT, filling 30 Hz transitions (6 to 16 tokens of four poses, or 0.6 to 1.9 seconds per chained step) between four recent context poses and up to four target poses.

Three-module VQ-VAE architecture, Smart Primitives and split licensing

Three learned networks share each generation: a root predictor for path and duration, a masked token transformer pose predictor and a multi-head VQ-VAE decoder. Reusable Smart Primitives snap locomotion and object interactions such as vaulting, crouch-strafing and sitting together zero-shot without animation graphs or retraining, complementing NVIDIA’s GR00T 2 robotics stack.

One MotionBricks runtime drives OpenUSD SOMA game characters (top) and physical Unitree G1 humanoid trajectories (bottom).Video: NVIDIA

Built on NVIDIA Warp, Newton physics and TensorRT, the repository provides Apache 2.0 source code, CC-BY-4.0 documentation, Inbetweener and Scene Preview demos across six gait styles, two primitive and scene editors, and commercially licensed ONNX and SafeTensors checkpoints (under the NVIDIA Open Model Agreement) driving both OpenUSD SOMA human characters and Unitree G1 humanoid robots.

Technical details

Backbone throughput and corpus
15,000 FPS and 2 ms latency on TensorRT; trained on 350,000+ BONES-SEED mocap clips
Runtime stack and hardware
Python on NVIDIA Warp, Newton and TensorRT (tested on Windows/Linux x86-64 RTX 4090)
Pretrained character and robot rigs
OpenUSD SOMA human skeleton and Unitree G1 humanoid (bundled ONNX + HF SafeTensors)
Component licensing split
Apache 2.0 code, CC-BY-4.0 docs, NVIDIA Open Model Agreement weights (commercial OK)

Release details & context

GR00T Whole-Body Control integration and physical hardware controller requirements

MotionBricks now powers the real-time motion-generation layer inside NVIDIA’s GR00T Whole-Body Control stack alongside offline sibling model GEAR-SONIC and the Newton-based SOMA Retargeter. While NVlabs/motionbricks distils NVIDIA’s uncut 2-minute-40-second Unreal Engine 5 showcase into readable Python apps, the earlier GR00T-WholeBodyControl/motionbricks preview retains the self-contained synthetic training loop.

NVIDIA’s Hugging Face model card notes that SOMA and G1 weight folders are non-interchangeable, sample BONES BVH/CSV clips carry an evaluation-only licence, and deploying generated kinematic trajectories onto physical Unitree G1 hardware requires a separate tracking controller and joint-limit safety checks.

Common questions

Can studios use MotionBricks pretrained SOMA and Unitree G1 models commercially?

Yes, the runtime code uses Apache 2.0 and the SOMA and Unitree G1 weights use the NVIDIA Open Model Agreement permitting commercial use and redistribution, whereas bundled sample motion clips are restricted to evaluation.

NVIDIA release timeline (13 launches tracked)

  1. Earlier9 Nov 2026
    NVIDIA introduces GR00T 2 world-action robotics model and releases 100k hours of human video

    NVIDIA unveils the GR00T 2 world-action model trained on 402,730 hours of human video and open-sources the 100,000-hour Human-1M-100K dataset on Hugging Face.

  2. Earlier10 Nov 2026
    NVIDIA open-sources Alpamayo 2 Edge reasoning model for onboard autonomous driving

    NVIDIA open-sources Alpamayo 2 Edge, running adaptive driving reasoning in 160 to 190 ms on DRIVE AGX Thor at 15.15 km per incident in AlpaSim.

  3. Earlier12 Nov 2026
    NVIDIA names CFO Colette Kress interim CEO as Jensen Huang resigns

    NVIDIA names CFO Colette Kress interim CEO and Harvey C. Jones its first-ever board chair after Jensen Huang resigns effective 12 November 2026.

  4. Earlier17 Nov 2026
    NVIDIA founder Jensen Huang says he would return as CEO if board asks

    NVIDIA founder Jensen Huang pledges full cooperation with the board’s independent investigation and says he would return as CEO if the board asks.

Explore the full NVIDIA dossier or browse all open source releases.

Also in the Thursday 19 November edition

Read all 16 launches on one page
  1. 01
  2. 02
  3. 03
  4. 04
  5. 06
  6. 07
  7. 08
  8. 09
  9. 10
  10. 11
  11. 12
  12. 13
  13. 14
  14. 15
  15. 16

Spotted an error? Email aiforsuccess.in+compass@gmail.com or tag @ai_for_success on X

Email Edition · Early Access

Get the 5-minute AI briefing

Daily and Sunday Weekly inbox editions open soon. Join the early-access list.

Zero spam · Privacy

! Please enter a valid email address (for example, you@domain.com).

Morning Edition Alerts

Get a 1-click alert at 11:00 IST

One quiet browser alert when each daily or Sunday Weekly edition goes live between 11:00 and 12:00 IST. No email required. Privacy