Unitree unveils UnifoLM-OminiA-0.3 multimodal humanoid robot for home and care

On July 20, Chinese AI firm Yushu Technology (宇樹科技) launched the UnifoLM‑OminiA‑0.3 model, a multimodal large‑language model that integrates vision, speech, reasoning and motion control. The model powers a humanoid robot demonstrated by Unitree, capable of understanding commands through voice and visual cues, operating autonomously, and resisting interference.

The robot performed household tasks such as moving pillows to a sofa, loading a dishwasher, adjusting a hospital bed, delivering medication, and sorting clothing. It combines modular hardware, high‑performance drivers and a multi‑modal perception system, enabling precise grasping and stable locomotion in domestic, clinical and care‑facility environments.

Developers cite the model’s ability to coordinate multiple tasks without separate subsystems as a step toward practical intelligent assistants in everyday settings.