China's Unitree Robotics Unveils Unified AI Humanoid Robot

Unitree Robotics of China introduced the UnifoLM‑OminiA‑0.3, a unified multimodal AI model for humanoid robots that combines reasoning, dialogue, vision, object recognition, navigation and whole‑body control in a single architecture. In video demonstrations the robot receives verbal commands, interprets its surroundings and carries out tasks such as moving a cushion onto a sofa, counting medication boxes, sorting laundry, loading a dishwasher and adjusting a hospital bed. The system can interrupt an ongoing action instantly when instructed, showcasing real‑time planning and coordination without the delays of earlier fragmented AI setups. Unitree positions the technology as an example of “embodied AI,” aimed at domestic assistance, caregiving and other environments where autonomous, adaptable robots could work alongside people.

The unified approach is presented as a step forward from previous robots that relied on separate subsystems for voice, vision and motion, promising smoother, faster responses and greater safety in home and healthcare settings.