This year's government work report proposes to cultivate future industries such as biomanufacturing, quantum technology, embodied intelligence, and 6G. As an important direction of embodied intelligence, humanoid robots have shown great potential in industrial manufacturing, life services, and other fields, and are expected to become disruptive products after smartphones and new energy vehicles. Currently, China has demonstrated a first mover advantage in the field of humanoid robots that is comparable to developed countries. However, it still faces problems such as insufficient technological innovation capabilities, relatively backward manufacturing processes, and high overall costs. Therefore, comprehensive measures need to be taken from the aspects of technological innovation, scene traction, industrial organization, and ecological creation to quickly create new competitive advantages. In recent years, China has successively issued policy documents such as the 14th Five Year Plan for Robot Industry Development and the Guiding Opinions on the Innovative Development of Humanoid Robots, and set up major projects for key support, gradually building an industrial development support system covering the construction of technological innovation system, expanding scene applications, and creating industrial ecology. The powerful perception, learning, and interaction capabilities of artificial intelligence integrated into physical entities enable humanoid robots to have a wider range of application scenarios. Relying on the long-term accumulation in the field of industrial robots and the rapid development of artificial intelligence technology, China has demonstrated a first mover advantage in the field of humanoid robots. Mainly manifested in the following aspects. One is the continuous breakthrough of key core technologies. In the key technology group of "limbs", the localization of core components such as six dimensional torque sensors and frameless torque motors is accelerating, and domestic enterprises are gradually breaking the monopoly of foreign manufacturers. In terms of key technology groups for the "brain" and "cerebellum", China has successfully developed models for the "brain" and "cerebellum". In July 2024, the National and Local Collaborative Humanoid Robot Innovation Center in Shanghai successfully launched its independently developed "Zhuque" brain model and "Xuanwu" cerebellar model at the World Artificial Intelligence Conference. The second is that the industrial chain is taking shape. Domestic manufacturers have basically achieved full coverage of their technological systems and successfully launched complete machine products, promoting rapid industrial development. According to data from the China Academy of Electronic Information Industry Development, the scale of the domestic humanoid robot industry will reach 3.91 billion yuan in 2023, a year-on-year increase of 85.7%; From 2024 to 2025, China's humanoid robots will achieve small-scale mass production; By 2026, the scale of the humanoid robot industry will exceed 20 billion yuan. The trend of industrial agglomeration development is increasingly evident, with a group of core enterprises gathering in regions such as the Pearl River Delta, Beijing Tianjin Hebei, and Yangtze River Delta, becoming an important force in building an ecosystem for the humanoid robot industry. Thirdly, the advantages of scenario application are obvious. China has many unique advantages in industrial production, medical and health care, commercial retail, and life services. The industrial manufacturing scene has a high degree of structuring and a relatively harsh working environment, making it an ideal scenario for the large-scale promotion of humanoid robots. At present, manufacturing companies such as GAC and SAIC are competing to lay out humanoid robots. The Uber S series humanoid robots have entered automotive factory training and are performing specific tasks such as sorting, handling, and quality inspection in frontline production areas. The domestic population situation has shown a new trend of low birth rate and aging, and the single economy and silver economy are rising. In life service scenarios such as family partners, emotional comfort, and elderly care, humanoid robots will have great potential. In fact, many hospitals have begun to use humanoid robots to assist in nursing work. The development of humanoid robot industry in China is facing challenges. At present, countries are accelerating the competition for technology dominance around software and hardware technologies such as sensor systems, intelligent control systems and transmission systems. There are still some problems in the development of humanoid robots in China, such as the large gap in the level of core components and the lack of high-quality data sets for large vertical models. Mainly manifested in the following aspects. There is a gap between the core component technology and process level and that of foreign countries. There are still bottlenecks in the manufacturing of frameless torque motors, hollow cup motors, and roller screws. For example, magnetic circuit and process design are the core technologies of frameless torque motors. Currently, the distributed fractional slot and carbon fiber binding technologies used by American companies are leading the world. Domestic frameless torque motors have significant differences in key parameters such as torque density, temperature rise, and maximum speed. The second issue is the lack of high-quality datasets for training large models. The big model algorithm is the core technology that affects the performance of the "brain" and "cerebellum" of humanoid robots, and can determine the intelligence level and motion ability of humanoid robots. It is the key for countries to compete for the technological high ground of the humanoid robot industry. However, there is a lack of high-quality datasets for training large model algorithms in China, and there is a significant gap between China and foreign countries in terms of data collection efficiency and data simulation technology. For example, our research has found that there is currently a lack of large models in China that can efficiently generate simulation data for humanoid robots. Only a specific scenario can be established in the physical world, and the humanoid robot can be commanded to continuously move to achieve data collection. Thirdly, high costs constrain industrial applications. The overall cost of humanoid robot products is relatively high. Xiaomi CEO Lei Jun once revealed that the cost of domestically produced humanoid robots is about 700000 yuan per unit, and the high overall cost hinders the industrial application of humanoid robots. On the one hand, the cost of core components is relatively high. The core components of humanoid robots lack unified technical standards, and customized production costs are high. In addition, high-precision core components such as frameless torque motors rely on imports, further increasing the overall product cost. On the other hand, there is insufficient scene traction and weak willingness of enterprises to mass produce. For example, in industrial manufacturing scenarios involving fragmented scenarios such as human-machine collaboration and intelligent handling, humanoid robot manufacturers find it difficult to determine which fragmented scenarios to develop production plans for in order to achieve the maximum return on investment. This raises doubts about the expected demand for complete machine product orders, making it difficult to implement production plans. Fourthly, the innovation ecosystem needs further improvement. On the one hand, the form of industrial organization needs to be improved. At the level of industrial collaboration, there is a lack of "chain owner" enterprises that can drive the collaborative development of innovative ecological communities. At the level of industrial agglomeration, there is a lack of mature innovation carriers such as humanoid robot industry clusters, incubators, and industrial parks, which drive the development of upstream and downstream agglomeration in the industrial chain. On the other hand, the supply structure of innovative factors is not optimal, particularly manifested in the weak adaptability of the supply structure of innovative factors to the demand structure. In addition, the efficiency of innovative services needs to be improved, especially due to a lack of funding support during the concept validation phase. According to research, research institutes and startups currently lack funding support for producing samples in the early stages of concept validation. There is significant uncertainty about whether research results can be successfully transformed into marketable products, and it is even more difficult to accurately match user groups on the demand side. To address the above issues, China needs to focus on the characteristics and prominent advantages of industrial development stages, strengthen strategic top-level design, improve the technological system, create a good industrial ecology, and promote the deep integration of humanoid robot technology innovation and industrial innovation. Firstly, driven by industrial demand, accelerate breakthroughs in key core technologies. We need to deploy innovation chains around the industrial chain and systematically layout the core technologies of the entire stack. Strengthen the "bottom-up" demand condensation and innovative organizational mechanism, condense the specific needs of enterprises for lightweight bones, high-precision sensors, etc. into a technical list with clear technical parameters and reasonable research and development cycles, and adopt models such as "unveiling and leading" and "horse racing" to organize research and development. Support the construction of a humanoid robot data center that includes an action library, object knowledge base, and data collection platform, accelerating the creation of an open-source humanoid robot dataset. Leading enterprises in the fields of digital twins and data simulation, as well as suppliers of humanoid robot "brain" and "cerebellum" technologies, form an innovative consortium to accelerate the development of virtual training environments and data simulation models. Secondly, strengthen the deep integration of intelligent and cutting-edge manufacturing technologies, and create "star" level whole machine products. Develop a basic version of the whole machine, focusing on basic functions such as high intelligence and high mobility of humanoid robots, and creating a universal whole machine platform that can support algorithm optimization and structural modification. Develop high-precision, low-cost interactive, and highly reliable functional robots for extreme environments, promote the industrialization of humanoid robots, and accelerate the reduction of product costs. Thirdly, expand the scenario traction effect and drive technological iteration and upgrading. Focusing on high-energy market operators with ecological leadership in the fields of industrial manufacturing and life services, selecting a group of key application scenarios with strong industrial driving force, further constructing a product demand list, and supporting humanoid robot manufacturers to accelerate the implementation of mass production plans based on the list. Organize the compilation of typical case sets for demonstration application scenarios, and promote the demonstration application of humanoid robot industry scenarios. Fourth, improve the form of industrial organization and cultivate a better ecological environment for industrial development. Cultivate and strengthen a group of humanoid robot ecological enterprises with wide coverage of core technologies and strong industrial driving capabilities, promote collaborative innovation and joint construction of production lines in the upstream and downstream of the industrial chain. Support various regions to build humanoid robot industrial parks and incubators based on regional resource endowments and industrial advantages, and accelerate the construction of advantageous clusters with strong innovation capabilities and good application scenarios. (Author Chen Zhi, Researcher at the China Academy of Science and Technology Development Strategy) (New Press)