Xiaomi-Robotics-1 VLA Model Trained on 100K Hours of Trajectories
July 15, 2026
Xiaomi-Robotics-1 is a vision-language-action model trained on 100,000 hours of real-world manipulation data via an auto-labeling pipeline. It enables zero-shot mobile manipulation and rapid adaptation to new tasks with minimal fine-tuning data.
HOW THIS AFFECTS YOU
●
builderYou can use this for more robust, language-conditioned robotic manipulation in unseen environments.
●
researcherThe scalable auto-labeling pipeline offers a blueprint for large-scale VLA pre-training.