Xiaomi Robot Debuts 38B World Model
CyberRobooo · x · 2026-07-14
Xiaomi Robotics introduced Xiaomi-Robotics-U0, a 38B-parameter autoregressive world foundation model.
It unifies multiple generative capabilities, including:
- Text-to-image
- Image editing
- Multi-view embodied scene generation
- Scene transfer
- Embodied video generation
Its core purpose is to generate data for robot training:
- A single real robot trajectory can be expanded into variants with different backgrounds, lighting, and object placements
- It preserves original actions and task logic while covering scenarios rarely seen in reality
The technical report highlights the following results:
- After post-training with data generated by U0, the robot's average task success rate in unseen backgrounds and complex lighting improved from 36.9% to 63.2%
The post concludes with a clear message:
- World models aren't just for predicting the world
- They can also generate new training worlds for robots
Related event: Xiaomi Robotics Unveils U0: a 38B Embodied World Foundation Model(3 posts)→
More from Embodied
- LIBERO-MAX: Robot Policies Drop 11–26 Points When the World Changes Mid-Task — drmapavone · 2026-10-27
- Photos from ROSCon UK: BBC's robotic whale, da Vinci surgical robot, PR2 and Baxter — kamathsblog · 2026-10-27
- Agility's Digit robot to expand into two more GXO US warehouses from early 2027 — jonstephens85 · 2026-10-27
- Real2Sem2Real turns one phone video into 2,000+ robot demos, matching 150 teleop demos — Scobleizer · 2026-10-27
- Wayve CEO argues end-to-end deep learning should power driving, safety and world models — alexgkendall · 2026-10-27
- Mundo turns a few room photos into an explorable, simulatable 3D world — at 1.5 FPS — mark_k · 2026-10-27