Xiaomi's robotics technology makes progress, successfully achieving stable factory operations and trial operations in youth apartments

·by Henderson·AI
Xiaomi's robotics technology makes progress, successfully achieving stable factory operations and trial operations in youth apartments

On November 3, Lei Jun posted on social media that at the 2026 Wuzhen World Internet Conference, Xiaomi Group was awarded the conference's "Outstanding Contribution Award," while the Xiaomi robotics team received the "Leading Technology Award" for their "key technologies for a general-purpose robot operating base model aimed at scalable deployment." He expressed gratitude for the recognition and took the opportunity to report on two recent advancements in Xiaomi's robotics business.

Lei Jun pointed out that in industrial scenarios, Xiaomi's humanoid robot successfully challenged 8 hours of stable operation at the automotive factory's central console side cover sorting workstation, meeting the workstation's all-day production requirements. During the operation, the robot was able to autonomously correct anomalies and achieve system-level collaboration with the production line system. The 8-hour operation time corresponds to a full shift, and this duration is more convincing than single-point action demonstrations, indicating that the robot must withstand fluctuations in rhythm, incoming materials, and anomaly handling to qualify for production line-level operational capability.

Latest developments in Xiaomi's robotics business in industrial and living scenarios

In living scenarios, Xiaomi's embodied base model has begun "trial operations" in Xiaomi Youth Apartments, undergoing multiple rounds of iterative optimization based on Xiaomi-Robotics-1, establishing an Agent closed-loop system that encompasses data collection, model inference, and task execution. Currently, the system is capable of completing continuous tasks for over 30 minutes, covering 3 rooms and involving more than 50 tasks. Unlike demonstrations involving a single room and single task, cross-room continuous operations require the model to maintain an uninterrupted task chain amid constantly changing environments, which is a challenge that the general-purpose robot operating base model needs to address.

These two scenarios correspond to the most challenging ends of deploying general-purpose robots: one side is the automotive assembly workshop with standardized processes but strict rhythms, while the other is the highly unstructured living space. Xiaomi stated that it will continue to focus on key areas such as data, models, ontology, and scene applications in the future, and these two "internships" and "trial operations" are accumulating operational data in real-world scenarios for the model.

H
About the author
Henderson