At present, China's humanoid robot and embodied intelligence industry is in a crucial window period of transition from technological verification to commercial application. As a cutting-edge field of the deep integration of artificial intelligence and high-end manufacturing, embodied intelligence embeds AI capabilities into physical carriers, enabling intelligent agents to complete perception, decision-making, and action in real environments, and humanoid robots are precisely its core carriers.
The General Office of the Ministry of Industry and Information Technology and the General Office of the State-owned Assets Supervision and Administration Commission of the State Council have issued a notice on jointly carrying out the 2026 Humanoid Robot and Embodied Intelligence real-scene Training Special Action. Adhering to application-driven development, we will focus on key scenarios in industrial, special, and service sectors, and promote the construction of real-scene training Spaces, the cultivation of innovative application consortia, the breakthrough of operational skills, and the verification of application deployment in an integrated manner. Through real-world scenario training, we will continuously optimize the algorithms of the intelligent model with the body, accumulate high-quality real machine data, and enhance the performance of key components of the body. Explore the establishment of a full life cycle management and guarantee mechanism for humanoid robots and embodied intelligent products.
At present, China's humanoid robot and embodied intelligence industry is in a crucial window period of transition from technological verification to commercial application. As a cutting-edge field of the deep integration of artificial intelligence and high-end manufacturing, embodied intelligence embeds AI capabilities into physical carriers, enabling intelligent agents to complete perception, decision-making, and action in real environments, and humanoid robots are precisely its core carriers. In recent years, driven by both policy guidance and market demand, China has continuously made breakthroughs in model algorithms, core components, and complete machine integration. However, it still faces pain points such as insufficient adaptation to real scenarios, lack of accumulation of real machine data, unstable performance of key components, and unsmooth industrial collaboration mechanisms. Most products are still in the "display mode" It is difficult to handle regular operations under high-intensity and complex working conditions. The launch of the real-scene training special action is precisely aimed at these bottlenecks, taking "on-site machine practice and practical training iteration" as the path to break through the key link from technology research and development to industrial application.
This special operation has established a clear implementation system and goal blueprint, with core tasks that are both systematic and practical. The action clearly defines six key tasks: creating standardized real-scene training Spaces, focusing on specific scenarios such as production and manufacturing, equipment maintenance, warehousing and logistics, medical care and health, and emergency rescue, and selecting real operation units as training carriers to avoid repetitive construction and resource waste; Form an innovation application consortium, with user units and whole machine enterprises as the core, and link algorithm, component enterprises and research institutes to build a collaborative ecosystem of industry, academia, research and application. Deeply engage in the research and development of practical operation skills, optimize the anti-interference and self-adaptation capabilities of embodied intelligent model algorithms in real working conditions, and overcome core skills such as dexterous operation and autonomous navigation. Promote the verification and regular deployment of real-scene applications to form a cluster effect of "verifying one, deploying a batch, and driving a large area". Strengthen the guarantee of elements such as data governance, standard norms, and talent cultivation, explore business models like "humanoid robots as a service", and lower the application threshold; Distill mature experiences and establish a full life cycle management and guarantee mechanism.
In terms of scene layout, the special operation precisely focuses on three major areas, taking into account both industrial value and people's livelihood demands. In the industrial sector, for intelligent manufacturing factories, practical training in assembly, quality inspection, operation and maintenance, warehousing and logistics, etc. is carried out to facilitate the digital transformation of the manufacturing industry. Special fields focus on high-risk environments, promote the application of scenarios such as chemical inspection, power maintenance, emergency rescue, and disaster prevention and mitigation, replace manual labor to complete high-risk tasks, and build a solid line of defense for safe production. The service areas cover various livelihood scenarios such as catering and retail, medical care and health preservation, and domestic services, promoting the integration of robots into daily life and enhancing the intelligent level of services. Through full-scenario real-scene training, not only can the practical performance of the product be rapidly refined, but also high-quality real machine data can be accumulated, providing core support for algorithm iteration. Simultaneously, it can drive the performance improvement of key components such as servo motors, reducers, and sensors, and consolidate the foundation of the industrial supply chain.
This special operation has set clear phased goals and drawn a clear timetable for the development of the industry. By the end of 2026, key products such as humanoid robots will have completed application verification and regular deployment in a number of representative scenarios, officially entering the "operation mode". Condense and form over a hundred high-value application scenarios, enriching the spectrum of embodied intelligent applications; It will drive the formation of a large-scale application capacity of tens of thousands of units, marking that China's humanoid robot industry has entered a new stage of large-scale commercialization from a hundred-unit pilot program. 2026 will also be the first year of commercial verification for the industry. Meanwhile, the action covers ten major provinces and municipalities including Beijing, Shanghai, Jiangsu and Zhejiang, as well as related central enterprises. Relying on the advantages of the National Artificial Intelligence Innovation and Application Pilot Zone, it is advanced with unified standards and paths to ensure that the practical training results rapidly spread across the country.
Empowered by policies, driven by scenarios, and with collaborative efforts, the humanoid robot and embodied intelligence industries are now entering a golden period of development. As an important means of new quality productivity, its industrialization and implementation can not only promote the upgrading of manufacturing, solve the problems of special operations, and improve the quality of public services, but also drive the coordinated breakthroughs in multiple fields such as artificial intelligence, high-end manufacturing, and new materials, and build an industrial cluster with global competitiveness.
Looking ahead, as the special campaign progresses in depth, humanoid robots and embodied intelligence will gradually permeate all fields of production and life. China will also seize the initiative in the global competition of humanoid robots and embodied intelligence industries by relying on its complete industrial system.





