China's National Data Administration has announced plans to promote data standards for embodied intelligence in the future. Furthermore, the agency will instruct local data bureaus to carry out relevant work in an organized manner and will actively support companies that increase their investment in data.
Director Liu Liehun chaired a symposium on September 10th, which gathered research institutes, robot manufacturers, data firms, and a humanoid training center. The agency posted its account on September 13th. Although the statement does not specify a concrete date, name a draft, or open consultations, it places the data regulator in the discussion about embodied AI standards alongside the Ministry of Industry.
This announcement complements the existing set of documents. The Technical Committee of the Ministry of Industry and Information Technology released the Six-Part Standard System for Humanoid Robots and Embodied Intelligence (2026 edition) in February. This system covers general basic aspects, including safety and ethics, as well as data lifecycle rules in the chapter on intelligent computing. The industry standard YD/T 6770-2026 on reference testing methods for embodied intelligence is effective from June 1st. Standard YD/T 6771-2026, concerning the quality of datasets based on parameters such as completeness, consistency, diversity, reliability, and security, comes into force on November 1st.
The national guiding document GB/Z 218.1-2026 on the quality of real-world embodied intelligence data has been announced, with developers including robotics and cloud technology companies, as well as the Beijing Humanoid Robotics Innovation Center. Participants named by the National Data Administration represented the entire data chain, including the Institute of Automation of the Chinese Academy of Sciences, the Beijing Institute of General Artificial Intelligence, JD.com, Guanglun Intelligence, the Shijingshan Humanoid Robot Data Training Center, and the Beijing Humanoid Robotics Innovation Center. Two of these institutions also appear in the list of developers of GB/Z 218.1-2026, highlighting the continuity between current recommendations and new regulatory interest.
The agency defines embodied intelligence as a data-driven industry whose demand for high-quality, diverse, and large-scale multimodal records is constantly growing. Previously in 2026, it included embodied intelligence in the list of advanced fields in its plan for high-quality industry datasets. The novelty lies in the explicit promise to promote data standards and guide local bureaus—a signal aimed at unifying dozens of training grounds and corporate bodies, although annotation formats or evaluation metrics have not yet been specified. Companies should focus on the entry into force of YD/T 6771 in November as the nearest quality criterion while awaiting any National Data Administration project that aligns with or expands upon the level of MIIT and GB/Z.