Xiaomi Introduces embodied AI Model Xiaomi-Robotics-U0

Xiaomi on Wednesday unveiled Xiaomi-Robotics-U0, a new multimodal autoregressive foundation model with 38 billion parameters designed for embodied AI.

The company stated that the model integrates four core capabilities into a single framework: embodied scene generation, embodied transfer, robot interaction video generation, and general-purpose image generation and editing.

According to Xiaomi, the model can generate robot-ready environments from text prompts. It is also capable of adapting existing robot trajectories to new scenes while preserving motion characteristics and producing robot interaction videos based on task instructions.

The Xiaomi-Robotics-U0 model leverages internet-scale visual knowledge for embodied AI applications, aiming to advance the development of intelligent robots that can interact with and operate within physical environments.