Open Source Securities: The humanoid robot half marathon shows the advantages and technical shortcomings of China's industrial chain and is more clear and clear

The Zhitong Finance App learned that Open Source Securities released a research report saying that the world's first humanoid robot half marathon was successfully held in Yizhuang, Beijing, marking that China has gained a leading edge in large-scale implementation of high-intensity humanoid robot sports scenarios. The competition comprehensively tested key technologies such as the coordination of the robot's joint module and the stability of the control algorithm. At the same time, shortcomings such as motors, transmission mechanisms, and “cerebellar” operation and control were revealed. The bank recommended Tesla chains, common key technologies, those benefiting from the lead screw, and those related to the cerebellum.

The main views of Open Source Securities are as follows:

Humanoid robots “ran” into reality, and we saw the future

On April 19, the world's first humanoid robot half marathon was held in Yizhuang, Beijing. The event was led by the Beijing Municipal Government, including 21 humanoid robots including Tiangong Ultra and Songyan Power N2. CCTV5 was broadcast live on the “large screen+TV+mobile”, which attracted public attention with its sense of technology and innovation. In the end, Tiangong Ultra developed by the Beijing Humanoid Robot Innovation Center won the championship with a score of 2 hours 40 minutes and 42 seconds. Its shareholders Premium Choice, Xiaomi Group, and Beijing Electromechanical each held 28.57% of the shares. Matsunobu Power won second place. Founder Jiang Zheyuan was the co-founder of Stardom Era.

The full length of this marathon was 21.0975 km. The track covered various complex road conditions such as crack roads, steep slopes, and stone roads. It comprehensively tested the robot's joint module coordination, control algorithm stability, energy management, structural design, and signal anti-interference capabilities. It can be said that it was an extremely difficult “comprehensive test”. As the world's first “man-robot co-run” event, the purpose of this marathon is not to run fast, but to use it as a comprehensive inspection platform to conduct extreme exercises under stress tests in an unideal environment and long-distance sports, promote industrial chain linkage, and form a synergy of standardization and large-scale implementation. With strong industrial chain collaboration and scene verification capabilities, China has taken the lead in large-scale implementation of high-intensity motion scenarios for humanoid robots, and other countries have yet to have similar verification conditions.

Compared to the World Artificial Intelligence Conference in July 2024 (some robots still require lifting assistance), this competition directly allowed robots to run the full course, showing a leap forward in key technology. At the same time, the competition raised public awareness and acceptance of humanoid robots through mechanisms such as high-exposure live streaming and “popularity award” interaction, and paved the way for future social applications of human-robot integration. As a figurative display of the national strategy for humanoid robots, the “one race, one session” industrial ecology strategy is also warming up for the upcoming World Humanoid Robot Games to be held in August.

Clear shortcomings, cerebellar control, motors, high-efficiency actuators, etc., point the direction for industrial development

Through this competition, the technical shortcomings that the development of the robotics industry needs to focus on are more clear and clear. For example, problems such as performance decay or fall due to joint overheating, insufficient dynamic balance capacity, and the need for multiple power exchanges indicate the direction for subsequent industrial development: (1) Motors: the most important power source and heat source, which directly affects battery life and heat load. They need to balance high power density and heat dissipation, and place high requirements on structural design, magnet arrangement and coil winding; (2) transmission mechanisms (screws, speed reducers): directly affect motion efficiency, energy consumption control and response speed, and place higher requirements for precision, rigidity, friction loss and motor matching control. (3) “Cerebellum”: The core of dynamic balance, including drive control, encoder, controller, etc. The challenge is how to reduce instruction delays, improve the compatibility of various hardware, and improve decision-making ability through continuous training; (4) Lightweight materials: Self-weight is the key to battery life and thermal management, and weight reduction is a direct means to improve energy efficiency ratio and extend battery life.

AI and software determine hardware, and Tesla's leading edge is obvious; domestic production focuses on common key technologies and supply chains

AI and software determine hardware. Tesla has built a complete AI system from models, data, computing power to chips, and has established a highly supporting supply chain system support. The robot production is expected to be 5,000 units in 2025. The next two years will grow at a rate of 10 times each. Industrialization is leading the way, and the focus is on the Tesla chain from an investment perspective. Domestic robots such as Huawei and Yushu are speeding up to catch up and make up for shortcomings, driving opportunities in the domestic supply chain. The domestic chain focuses on common key technologies, including key components such as actuators, screws, motors, speed reducers, and sensors.

Investment recommendations and beneficiary targets

(1) Tesla Chain: Recommended Targets: Zhenyu Technology, Fengli Intelligence, Wuzhou Xinchun; Beneficial Targets: Top Group, Sanhua Intelligent Control, Mingzhi Electric; (2) Common Key Technologies: ① Motors: Recommended Targets: Raysai Intelligence; Beneficial Targets: Buke Co., Ltd., Huichuan Technology; ② Lead Target: Saimo Intelligence, Radick; ③ Reducer Recommended Targets: Zhongda; ④ Recommended Targets: Raysai Intelligence; ⑤ Other Beneficial Targets: Gugao Technology; Other Beneficiary Targets: Gugao Technology Intelligent manufacturing, Zhejiang Rongtai.

Risk warning: risk of macroeconomic fluctuations; mass production of robots falls short of expectations; supply chain development falls short of expectations.