China News Service, Hangzhou, October 17 (Qian Chenfei) From October 13th to 16th, the 2026 third meeting of the International Telecommunication Union Embodied Intelligence Standards Focus Group and the Embodied Intelligence Standards Advanced Training Course were held in Hangzhou, Zhejiang. Participating experts, scholars and business representatives exchanged views on key technologies for embodied intelligence, international standard construction and industry applications.
Embodied intelligence is moving from performing preset tasks to sensing, learning, acting and collaborating with people in open and dynamic environments. As robots accelerate their entry into industrial production and daily life scenarios, issues such as how to standardize multi-modal data, how to achieve interoperability between different systems, and how to evaluate robot capabilities have become increasingly prominent, becoming a common concern for industrial development and international standards research.
Wang Yuntao, chairman of the International Telecommunication Union's Embodied Intelligence Standards Focus Group and deputy chief engineer of the Artificial Intelligence Research Institute of the China Academy of Information and Communications Technology, pointed out that international standardization is not a pre-constraint for technological innovation, but a practical need for embodied intelligence to move towards global large-scale application. At present, the focus group has brought together more than 500 experts from 57 countries and regions, established nine major groups, and promoted research around infrastructure and terminology, multi-modal data, embodied intelligence models, system integration and interfaces, open source ecology and other directions. Dozens of standard work projects have been carried out in an orderly manner.
In his view, transnational technological collaboration is the key to standard construction. Industry, academia and research entities in various countries should continue to deepen the linkage, improve benchmark testing and evaluation systems, and promote the safe and reliable application of embodied intelligence in all walks of life with common international rules.
China’s participation in the formulation of international standards for embodied intelligence has achieved concrete results. ITU-T F.748.66 "Embodied Intelligence System Framework and Capability Requirements", led by the China Academy of Information and Communications Technology, was approved and came into effect in December 2025, clarifying the definition of embodied intelligence, system framework, and key capability requirements such as perception, decision-making, execution, interaction, and learning. This standard provides a common technical basis for global research and development, and also marks that China's research results have been transformed into globally recognized standard results through international consultation.
On-site advanced training class on embodied intelligence standards. Photo provided by the organizer
At the advanced training class on embodied intelligence standards held on the 16th, the China Academy of Information and Communications Technology, the Department of Electronic Engineering of Tsinghua University, and the Embodied Intelligence Testing Laboratory jointly released the "Embodied Intelligence Development Report (2026)". The report shows that the embodied intelligence industry has moved beyond the laboratory prototype stage and entered a critical window period for commercialization. Beijing, Guangdong, Jiangsu, Zhejiang and other places have gradually formed characteristic industrial clusters, and the upstream, midstream and downstream industrial chain supporting facilities have been accelerated.
It is worth noting that the report points out that to realize the robot’s “ChatGPT” moment, the model requires tens of millions of hours of effective data. Driven by high data demand and large-scale data gaps, dozens of domestic companies have entered the embodied data track. Among them, simulation synthesis and real data evolve in parallel, and the boundaries of capabilities are gradually becoming clearer.
Zhou Enkai, co-founder and COO of Beijing Motphys Technology Co., Ltd., feels this deeply. When giving a keynote report, he said that with the development of the industry, the industry's focus is gradually shifting from robots to data and training. Obtaining high-quality data, improving training efficiency, and narrowing the gap between the simulation environment and real operations have become important issues to promote the implementation of robots.
"We built the MotrixLab training platform based on the self-developed MotrixSim physics engine to explore the 'real-simulation-real' closed loop, convert real scenarios into training resources, and then conduct large-scale training and verification through simulation." In Zhou Enkai's view, promoting the unification of data and simulation asset standards will help improve the standardization of training and evaluation, reduce the connection cost in technology implementation, and help embodied intelligence enter real scenarios such as factories.
"Standard discussions require the support of technical practice." Zhou Enkai said, "We have accumulated a lot of experience in the field of physical simulation. What we bring is not only a full-stack self-developed product, but also a set of simulation infrastructure solutions that start from the physical bottom and adapt to multiple computing power. When standards discussion requires industrial experience, we are willing to bring this experience in and provide a reference from China for the construction of global embodied intelligence standards."
On the same day, experts, scholars and business representatives from many countries in the field of embodied intelligence also gave special reports and held discussions on the themes of "key technologies of embodied intelligence", "industry application and deployment" and "making embodied intelligence deployable, comparable and trustworthy". (over)
