Xiaomi has introduced Xiaomi-Robotics-U0, a 38-billion-parameter multimodal autoregressive foundation model designed for embodied AI applications. The model represents the Chinese electronics giant’s bet on the convergence of AI and robotics.
The model unifies four core capabilities: embodied scene generation, embodied transfer, robot interaction video generation, and general-purpose image generation and editing. According to Xiaomi, it can generate robot-ready environments from text prompts, adapt existing robot trajectories to new scenes, and produce interaction videos from task instructions.
One of the model’s key innovations is its ability to leverage internet-scale visual knowledge for embodied AI. This allows it to draw on vast amounts of visual data to understand physical environments and generate appropriate robot behaviors.
Xiaomi has been steadily building its robotics capabilities, following earlier work with humanoid robots and quadrupedal robotic dogs. The U0 model is designed to serve as the intelligence layer for future Xiaomi robots.
The announcement positions Xiaomi alongside other major Chinese tech companies investing heavily in embodied AI. The 38B parameter model balances capability with deployment efficiency.
