As embodied intelligence technology becomes deeply integrated with application scenarios, humanoid robots are moving from laboratories to the front lines of production. Recently, the Office of the Ministry of Industry and Information Technology and the Office of the State-owned Assets Supervision and Administration Commission jointly issued a notice announcing the launch of the "2026 Humanoid Robot and Embodied Intelligence Field Training Action Plan." It clearly states that by the end of 2026, key products such as humanoid robots will complete application verification in multiple representative scenarios, achieve routine deployment, and officially enter "operation mode."
This initiative places particular emphasis on the term "field." The notice requires focusing on core areas such as industrial manufacturing, warehousing logistics, emergency rescue, and medical care, to create a series of field training spaces that are trainable, testable, and verifiable. These spaces are not simple testing platforms but real scenarios with standardized production capabilities. By following the principles of "reusing existing resources and minimal intervention," robots can directly conduct training at workstations and service points, ensuring seamless alignment between technological implementation and real production needs.
To bridge the last mile between technology and industry, the action encourages user units, system integrators, algorithm teams, and research institutions to form "innovative application consortiums." This mechanism aims to establish a long-term collaborative model: user units are responsible for opening up scenarios and providing real business data, while system integrators focus on task planning and human-robot collaboration based on scenario requirements. Supply chain companies and research institutions provide support in the background, offering iterative hardware and key common technology solutions. Through this joint effort, it not only enhances the robot's ability to resist disturbances and adapt to abnormal working conditions but also strengthens full-body motion control and spatial semantic understanding accuracy through the accumulation of high-quality datasets.
Regarding the current industry's concern about scalability, the action emphasizes a promotion path of "verify one, deploy a batch, and drive a region." For complete system solutions that pass verification, they will be scaled and deployed routinely in similar scenarios. Meanwhile, to lower the investment threshold for enterprises, the notice explicitly encourages exploring the "humanoid robot as a service" model, accelerating marketization through operational leasing or performance-based payment methods.
