The inaugural World Humanoid Robot Championships took place from August 15-17 at Beijing's National Speed Skating Oval "Ice Ribbon" and the National Stadium "Bird's Nest," featuring "steel and iron" athletes from around the globe competing in various events. Throughout this year, Chinese humanoid robot companies have accelerated technological iteration through competitive sports: from half-marathons to boxing rings, and soccer fields, high-intensity competitions have become crucial stages for testing product performance. The flourishing domestic events not only demonstrate China's progress in humanoid robotics but also drive the development of the entire industry chain.

**From Ping Pong to Mars Exploration**

At this humanoid robot "Olympics," 280 teams from 16 countries brought over 500 robots of various forms to compete in athletics, soccer, dance, boxing, and other sports. Simultaneously, specialized competitions involving cargo handling, pharmaceutical sorting, and other applications suitable for factories, warehouses, hospitals, and hotels tested the "labor skills" of humanoid robots.

Despite being a workday, the venue was nearly packed. Spectators frequently exclaimed "Look how flexible its hands are!" one moment, then lamented "Oh no, it fell!" the next. While audiences showed enthusiasm, the robotic "athletes" faced numerous challenges: on the track, robot "runners" occasionally veered off course while sprinting; in the boxing ring, some "fighters" slipped and fell flat as soon as the bell rang; in the dance area, "performers" moved smoothly to applause but occasionally dropped small components.

However, these incidents sparked not ridicule but heated discussions among spectators: "This robot's center of gravity is obviously too high" and "The algorithm of the first-place runner on the track must be very advanced."

The 1500-meter humanoid robot athletics race on the morning of August 15 drew significant attention. Hangzhou Unitree Technology's "Spring Festival Gala model" H1 robot won the championship with a time of 6 minutes 34 seconds, while Beijing Humanoid Robot Innovation Center's "Tiangong" robot placed second at 6 minutes 55 seconds. The "Tiangong" robot no longer required remote control and could autonomously complete the entire race.

"Tiangong" won the championship at Beijing Yizhuang's humanoid robot half-marathon in April. A company representative explained that when "Tiangong" ran the half-marathon, the team was already planning to achieve fully autonomous running in the next competition, with their focus on enhancing robot intelligence.

Compared to the half-marathon, humanoid robots showed significant improvements in step stability and speed.

"We didn't build this ping pong-playing robot to win world championships!" said Zhang Shanghang, researcher at Peking University's School of Computer Science and director of the Embodied Large Model Center at Beijing Institute for General Artificial Intelligence. His team's ping pong robot became one of the most attention-grabbing "technical stars" outside the competition arena.

Zhang explained that this robot wasn't designed for competitive victories but serves as a "technical platform" for verifying extreme capabilities. "Ping pong requires extremely high dynamic capture and millisecond-level response execution, authentically simulating technical challenges in future key scenarios like high-speed sorting on industrial assembly lines and highly dynamic environments in deep space exploration."

"Competition platforms accelerate technical verification and iteration, with the ultimate goal of making products 'face the stars and seas,' applying verified mature capabilities to extreme complex tasks like polar exploration, mining operations, and even lunar and Mars exploration," Zhang analyzed. Behind robots entering competitive arenas lies a powerful synergy chain of R&D and commercial demands.

As a non-profit research institution, the Beijing Institute for General Artificial Intelligence uses open-source strategies to build a "large-small brain collaboration framework" for humanoid robots—where large brain models handle high-level cognitive decisions and spatiotemporal planning, while small brain models precisely control action execution, allowing partner companies to "use directly" and quickly deploy to specific scenarios like pharmacy services and hotel reception.

"But this is just the beginning," Zhang stated. "The team's ultimate goal is to apply the 'quick reflexes' skills honed on ping pong tables to polar ice caps, kilometer-deep mines, and even lunar and Mars exploration missions—that's exploring robots' true limits."

**Distinctive China-US Industry-Academia-Research Transformation Mechanisms**

This summer, at the RoboCup Robot World Cup in Salvador, Brazil, Tsinghua University's Tsinghua Vulcan team won the adult division championship, while China Agricultural University's Shanhai team secured second place. Both used T1 robot platforms from Chinese humanoid robot company Accelerated Evolution. Notably, the third-place US team and the German team that won the small-size division also chose the same robot platform.

"This isn't coincidental," Accelerated Evolution Vice President Zhao Weichen explained. In fierce global competition, China's humanoid robot industry has achieved significant technological improvements over the past two years. Behind this phenomenon lies the divergence in development paths between Chinese and foreign industry chains.

Liu Shaoshan, director of the Embodied Intelligence Center at Shenzhen Institute of Artificial Intelligence and Robotics (AIRS), traced the global robotics industry's evolution: In early industry development, German and Japanese companies led in core components, while US companies dominated software algorithms. The turning point came around 2015.

"At that time, many Western enterprises weren't optimistic about robotics potential, viewing it merely as a branch under automation, and gradually reduced investment," Liu analyzed. "In contrast, China's intelligent new energy vehicles began explosive development during this period, driving maturation of the entire component supply chain. In this market competition wave connecting multiple industries, domestic component manufacturers achieved technological breakthroughs in motor controllers, high-precision sensors, and other core areas. The industry gradually gained large-scale talent reserves and rich application scenarios, plus national strategic attention, laying solid foundations for China's humanoid robot industry takeoff."

This supply chain advantage directly manifests in Chinese companies' hardware product competitiveness. Zhao Weichen demonstrated technical details: "Accelerated Evolution's self-developed new robot joints have gear backlash control ranges optimized 10 times over some competitors, significantly improving operational precision. This generational advantage can build 1-2 year technical barriers, while our costs are only 200 yuan higher than competitors."

However, he noted that the industry generally believes hardware advantages aren't lasting barriers in today's intensely competitive market; the more important competition lies in building developer ecosystems.

Many industry professionals noted that in China, humanoid robot industry ecosystem thinking has become similar to consumer electronics—benchmarking the "Apple model" by integrating hardware platforms, operating systems, and development tools to achieve technological revolution and popularization.

Across the Pacific, Jeff Burnstein, president of the Association for Advancing Automation, observed different landscapes. At the 2025 World Robot Conference on August 9, he noted that while humanoid robot concepts are equally popular in the US, only two cases have actually entered mass production: Agility Robotics' Digit robot and Tesla's Optimus robot. The former began "working" in logistics company warehouses in June, handling cargo boxes, while the latter's mass production plans encountered problems.

Burnstein believes large-scale humanoid robot applications will eventually become reality; the key question is timing. In his view, customers need companies to provide efficient, accurate, and low-cost solutions meeting demands in different scenarios. This aligns with Chinese companies' logic of "promoting scenario implementation through practical testing."

In the first half of this year, Chinese humanoid robot companies invested products in half-marathons, boxing rings, and soccer fields, refining technology through high-intensity robot competitions.

"Playing soccer is robots' 'actual combat,' with real confrontation and scenarios," Zhao Weichen explained. "In intense competition, robots' visual perception, autonomous movement, multi-machine collaboration, impact resistance, and stability when autonomously getting up after falls all undergo extreme testing. These capabilities align with underlying needs in household services and industrial scenarios."

Liu Shaoshan analyzed that under increasingly specific application demands, the next technological breakthrough bottleneck for humanoid robots lies in hardware, especially "refined" issues like force sensing and material durability required for "dexterous hand" operations.

"How to integrate all links of the humanoid robot industry into a mature system is the real test of a country's competitiveness in this field," Liu analyzed. China and the US currently have distinctive industry-academia-research transformation mechanisms. The US relies on market-driven industrial ecosystems with higher efficiency, exemplified by UC Berkeley's RISE Lab, which uses a "professor-led platform—student entrepreneurship—venture capital support" model, incubating seven unicorns including Databricks over the past five years, with total valuations exceeding $30 billion. This system's core lies in venture capital's deep involvement in early research, forming a transformation chain from "doctoral dissertations—startups—tech giants."

Although domestic institutions like Tsinghua and Beijing Institute of Technology are rapidly approaching research levels of top US universities like MIT, overall ecosystem gaps remain.

**Promising Applications in Healthcare and Elderly Care**

At the recent 2025 World Robot Conference, humanoid robots demonstrated broad application prospects in industrial, medical, elderly care, and companion scenarios.

As an exhibiting company, Hexagon Manufacturing Intelligence's Greater China Commercial Operations President Zhan Yanan explained that their first humanoid robot AEON is undergoing application testing in automotive and aerospace industries, primarily performing operational tasks, inspection patrols, reality capture, and work assistance.

Zhan noted that inspection of large complex workpieces like automotive bodies and aircraft components has long relied on skilled technicians. Robots, with flexible mobility and high-precision scanning capabilities, have significantly improved factory automation levels.

Multiple robot companies are collaborating with intelligent manufacturing enterprises, with the latter providing training scenarios for robots, promoting robot iteration through "trial and error."

UBTech Robotics Vice President and Research Institute Director Jiao Jichao introduced their collaborations with Dongfeng Liuqi, Geely Auto, BYD, Foxconn, SF Express, and other industry enterprises. The company's Walker S series industrial humanoid robots have entered multiple factories for practical training.

At this year's World Robot Conference, their new-generation humanoid robot Walker S2 demonstrated hot-swap autonomous battery replacement technology. Jiao explained this is the world's first humanoid robot with autonomous battery replacement capability, completing the process in just 3 minutes, enabling 24/7 uninterrupted operation and potentially improving industrial efficiency.

"The current humanoid robot industry is in a transitional phase from laboratory prototypes to commercialized applications in specific scenarios, with core breakthroughs concentrated in deepening embodied intelligence and scenario verification in vertical fields like industrial manufacturing," said Zhong Xinlong, director of the AI Research Office at CCID's Future Industry Research Center.

Leading products represented by "Tiangong," UBTech's Walker S, and Fourier Intelligence's GR-1 are no longer mere technical demonstrations but are deeply integrating into real production environments. In automotive manufacturing, humanoid robots are beginning to attempt complex tasks like final assembly, logistics, and quality inspection.

"This breakthrough's essence lies in systematic improvements in perception, decision-making, and control capabilities, enabling them to initially adapt to incompletely structured industrial environments and complete delicate tasks previously only operable by humans."

Additionally, humanoid robots show promise in healthcare and elderly care. Recently, domestic robot company Fourier released a humanoid robot focused on interactive companionship functions. Company representatives explained that compared to industrial scenarios, elderly care humanoid robots must possess higher perceptual sensitivity, more compliant control algorithms, and stronger safety protection capabilities.

Earlier this year, Chinese Academy of Sciences Institute of Automation researcher Zhao Xiaoguang explained that her team's robot prototypes are conducting demonstration applications in nursing homes, monitoring elderly people's health 24/7 and replacing care staff for "room checks."

Digital Huaxia Robot Company's leader Wu Wei noted that humanoid robot applications are entering deep waters: "Previously, robots were mainly used for relatively simple applications like guidance tours and entertainment performances. Starting this year, related companies are all considering how humanoid robots can improve functionality and achieve commercial loops."

However, experts believe that despite positive progress in application exploration, humanoid robots still face three core challenges before large-scale popularization: high costs, core technology maturity, and application ecosystem completeness.

Zhong Xinlong identified cost as the primary obstacle to commercial implementation. Currently, high-performance humanoid robot hardware costs, particularly core components like high-precision reducers, torque sensors, and high-performance servo motors, remain extremely expensive, making selling prices far exceed most enterprises' and consumers' affordability.

Regarding technology, robot stability, battery life, and environmental adaptability still need improvement. Finally, software and application ecosystems remain in early stages, lacking standardized operating systems and rich applications, limiting robots' functional expansion and scenario adaptation capabilities.

**Will Humanoid Robots Enter Homes on a Large Scale?**

"Looking ahead, humanoid robots are expected to develop into standardized intelligent terminals, but their popularization path will significantly differ from traditional electronic consumer products, showing gradual characteristics of commercial-first, then household applications," Zhong Xinlong explained.

Humanoid robots won't rapidly enter mass consumer markets like smartphones or personal computers. Initial development will focus on industrial, commercial, and specialized service B2B markets, where economic value created can effectively offset high costs.

Zhong further explained that industry chain maturation, cost reductions from large-scale production, and core technology iteration breakthroughs will bring humanoid robot prices to ordinary households' acceptable ranges. Only when they can provide irreplaceable household service value (like housework and elderly care) will C2C markets truly open.

Achieving convenient purchasing and after-sales requires building entirely new ecosystem systems, including modular hardware designs for easy maintenance and replacement, standardized software interfaces to attract developers, and extensive professional sales and service networks. This process will be lengthy, requiring coordinated advancement from industry, research, and policy levels.

**Robot Solutions Also "Going Global"**

Chinese robots are not only exploring practical applications domestically but also gaining overseas user popularity due to high-tech standards and cost-effectiveness. Singapore's Straits Times recently reported that Chinese industrial robot exports have surged as many new factories emerge across Southeast Asia.

The report mentioned that Chinese company Siasun Robot produces mobile robots for moving goods in factories, warehouses, and ports, plus robotic arms for tasks like automotive welding. Some foreign companies that previously used Siasun robots in China later placed orders for their facilities in Malaysia, Singapore, Germany, and elsewhere.

Siasun Robot Technical Director Zhang Lei recently explained that their products sell well in North American markets: "As Chinese manufacturing goes overseas, Chinese robot solutions are also brought along. We have significant advantages in ecosystem, supply chain, and cost-effectiveness."

Efort Intelligent Robot Vice General Manager Zhang Wei noted that their company mainly expands into European, Southeast Asian, and Japan-Korea markets, primarily selling industrial robots. He explained that Europe is a traditional global automation industry highland. According to IFR (International Federation of Robotics) 2024 statistics, among the world's top 20 robot-selling countries or regions, European countries occupy 8 positions, making it a "lighthouse" market for Chinese industrial robots.

Southeast Asia, as China's manufacturing spillover market, maintains close connections with Chinese industrial chains culturally, geographically, and industrially.

"We've participated multiple times in German, Italian, and other European automation exhibitions, finding some local automation integrator partners in Europe. Chinese industrial robots, leveraging China's manufacturing-rich scenarios, have incubated and iterated numerous automation solutions with Chinese advantages. Additionally, Chinese industrial robots' cost benefits from large-scale production also bring competitive advantages to European customers," Zhang Wei said.

"Chinese robot overseas market expansion shows diversified layouts, with industrial and service robots currently being export leaders, while humanoid robot exports remain in early stages of brand display and technical verification," Zhong Xinlong analyzed.

Although humanoid robot exports haven't formed scale, Chinese companies have successfully promoted products to Middle Eastern and North American high-end markets through international exhibitions and collaborative R&D, mainly for research, education, and corporate image display, laying foundations for subsequent market development.

Zhong noted that Chinese robot industry cooperation with overseas markets has entered deep integration stages, forming complex patterns where technology introduction coexists with independent innovation, and domestic and international supply chains interdepend. In upstream industries, although China accelerates domestic substitution in core components like reducers, controllers, and servo systems, some high-end products still require imports, maintaining close cooperation between Chinese robot companies and suppliers from Germany, Japan, and other countries.

In technological R&D, Chinese companies and research institutions actively participate in international academic exchanges, establishing joint R&D projects with global top universities and laboratories. Additionally, China actively participates in and even leads international robotics industry standard formulation.

However, like most industries going global, robot exports face challenges. Zhang Wei identified core first-stage challenges in building complete chains covering products, pre-sales, sales, and after-sales: "For example, industrial robot products must meet local European CE certification, plus require local pre-sales technical support capabilities."

"For Chinese robot companies to achieve full-chain globalization, they must shift from product thinking to systematic, localized global operational thinking, building comprehensive capabilities adapted to overseas markets," Zhong Xinlong stated. "This requires companies to possess globalization strategies in top-level design, treating overseas markets as important components of endogenous growth for long-term planning and resource investment."

Zhong suggested that to support globalization strategy implementation, companies should first build solid localized service networks. As complex productivity tools, robots' stable operation highly depends on timely technical support and maintenance. Companies can establish overseas subsidiaries and build strategic cooperation with capable local integrators or service providers.

Second, establish compliance and quality management systems meeting international standards. Product exports must cross strict technical and legal thresholds. Companies need to thoroughly research and obtain various certifications in target markets.

Third, implement deeply integrated supply chain and talent localization strategies. Companies should establish regional production or assembly centers in key markets, shortening delivery cycles, reducing logistics costs, and avoiding tariff risks. Supporting this, companies must advance talent localization, actively attracting and cultivating relevant talent with international vision and local market experience. This is fundamental assurance for achieving transitions from "product exports" to "brand exports" and "capability exports."