BEIJING, April 19 – In a groundbreaking event, the world’s first half-marathon for humanoid robots commenced in Beijing today. Twenty teams from universities, research institutions, and enterprises participated in the 21.0975-kilometer race, marking the inaugural “human-machine co-run” event. Competitors ranged from the 1.8-meter-tall “Tiāngōng” with its large stride to the compact 82-centimeter “Xiǎo Hǎi” and the endurance-focused “Xíngzhě Èr Hào” capable of six-hour operation.
Liang Liang, Deputy Director of the Beijing Economic and Technological Development Area Management Committee, emphasized that this marathon represents a systemic test of humanoid robots’ comprehensive environmental adaptability, moving beyond controlled laboratory settings. The event highlights significant technological leaps in humanoid robot mobility and endurance capabilities.
1. Humanoid Robots in the Spotlight
Public fascination with humanoid robots has surged following high-profile demonstrations. During the Spring Festival Gala, over ten Unitree humanoid robots performed traditional yangko dances alongside human dancers. Recently at the Zhongguancun Forum, nearly 100 humanoid robots from more than ten companies served as guides, hosts, and performers. The “Kuāfù” humanoid robot, developed through collaboration between Leju (Shenzhen) Robot Technology and the Beijing Institute for General Artificial Intelligence, demonstrated tai chi and wayfinding capabilities. Meanwhile, Tsinghua University’s championship-winning T1 humanoid robot performed complex maneuvers including synchronized push-ups and soccer tricks.
Investment momentum continues to build, with over ten municipal governments including Beijing, Shanghai, and Shenzhen establishing industrial funds exceeding 100 million yuan. According to IT Orange data, more than 50 companies in the embodied intelligence sector secured over 6 billion yuan in financing during Q1 2025 alone. The industry expansion reflects China’s strategic focus outlined in this year’s Government Work Report, which prioritizes future industries including embodied intelligence.
Industrial applications are accelerating. Ubtech’s Walker S1 humanoid robots have conducted material handling trials at Zeekr’s automotive factory, while Leju’s “Kuāfù” humanoid robot manages box transportation and parts sorting. Dongfeng Liuzhou Motor recently ordered 20 Ubtech humanoid robots for factory deployment – the first bulk industrial procurement of humanoid robots. Though current efficiency reaches approximately 70% of skilled workers, these humanoid robots primarily handle repetitive tasks in final assembly stages where conventional automation falls short.
Beyond manufacturing, Unitree humanoid robots have entered global markets, with products available on JD.com and AliExpress platforms. Nearly 30 Chinese universities employ them for research in locomotion, environmental perception, and decision-making algorithms. Specialized applications are emerging too, such as Guangdong Power Grid’s extreme-environment humanoid robots that locate faults in remote mountainous areas.
2. Why the Acceleration?
The recent progress contrasts sharply with historical challenges. Since Waseda University’s WABOT-1 took 45 seconds per step in 1973, development remained sluggish despite milestones like Honda’s ASIMO and Boston Dynamics’ Atlas. The breakthrough came with AI large language models in 2022, according to Wang Tianmiao, Honorary Director of Beihang University’s Robotics Institute. “Traditional robotics relied on programmed instructions, while large models simulate human cognition with generalization capabilities,” he explained.
Professor Xiong Rong of Zhejiang University notes that conventional approaches required mathematical modeling of biomechanics, limiting adaptability. “New-generation AI allows robots to acquire tacit knowledge through mass data training,” she stated. Industry consensus confirms that AI infusion has transformed humanoid robots from pre-programmed devices into systems capable of environmental interpretation and autonomous decision-making.
A pivotal moment occurred in 2022 when Tesla unveiled Optimus, demonstrating autonomous box-handling in factories. “Optimus understands instructions, makes decisions, and decomposes tasks without fixed trajectories – revealing humanoid robots’ true potential,” remarked Zhang Jin, President of Siasun Robotics. Professor Zhao Mingguo of Tsinghua University added that Tesla’s supply chain approach reduced costs dramatically, with key components like reducers and drivers becoming more accessible.
Current development focuses on three interconnected systems:
- Brain: Handles perception, judgment, and task decomposition
- Cerebellum: Manages motion control and real-time adjustments
- Body: Executes physical operations
The Beijing Humanoid Robot Innovation Center’s “Huìsī Kāiwù” platform represents a leap in multi-scene adaptability, though significant challenges remain. Zhang Weinan of Harbin Institute of Technology identifies “insufficient cross-scenario data” as the primary bottleneck for brain development, while specialized computing chips require further advancement.
Despite viral videos showing humanoid robots performing delicate tasks, experts caution about household adoption timelines. Wang Tianmiao predicts various embodied intelligence applications emerging in 2-3 years, with humanoid robots becoming prominent around 2035. Xiong Rong estimates 3-5 years for specific scenarios and 7-10 years for general home use. Leju founder Leng Xiaokun summarized the challenge: “Future home humanoid robots must be as affordable as appliances and as intelligent as PhDs.”
3. The Advantage of Chinese Manufacturing
Shanghai’s Zhīyuán Xīnchuàng Technology embodies China’s manufacturing edge, having produced 731 humanoid robots at its “mass production factory.” The facility features rigorous testing protocols, including 2,000 deep-squat trials per unit under 30kg loads. Morgan Stanley’s “Humanoid Robot 100” report notes that 56% of top-valued robotics companies are Chinese, crediting “mature supply chains, local application opportunities, and strong state support.”
Three strategic advantages fuel China’s humanoid robot development:
- Industrial Synergy: Automotive giants like Xiaopeng and Xiaomi leverage shared supply chains in batteries, chips, and sensors. Xiaopeng founder He Xiaopeng notes their humanoid robots currently match “autonomous driving L2 capability,” targeting commercially viable L3 systems.
- Engineering Talent: China’s 17.65 million engineers (2020) and 300+ university robotics programs create unparalleled human capital. Institutions like Harbin Institute of Technology’s State Key Laboratory of Robotics (established 1986) provide foundational research.
- Policy Frameworks: National guidelines position humanoid robots as “disruptive products” following computers and EVs. Local governments in Beijing, Shanghai, and Shenzhen have implemented targeted support policies.
Industrial robot producers like Topstar actively monitor the sector. “Our 200,000+ manufacturing clients provide immediate application scenarios,” stated founder Wu Fengli. Siasun’s Zhang Jin emphasized China’s complete industrial chain: “Each component has specialized developers, enabling healthy ecosystem development.”
4. Visions and Concerns for the Future
Official guidelines prioritize three application domains: special operations (explosive handling, rescue), manufacturing (automotive, electronics), and public services. While automotive factories currently deploy humanoid robots for simple tasks, efficiency limitations persist. Agility Robotics acknowledges their Digit humanoid robot operates just 1.5 hours before requiring equal charging time, underperforming human workers.
Training facilities bridge this gap. Shanghai’s Zhīyuán maintains a 4,000-square-meter “vocational school” where data collectors in VR headsets teach humanoid robots household skills through hundreds of repetitions. Similar facilities exist at national innovation centers. “Real-world interaction data requires massive collection efforts,” explained Peng Zhihui, Zhīyuán co-founder, noting over half their team develops “brain and cerebellum” systems.
Contrary to job-replacement fears, factory managers report persistent labor shortages. One factory’s recruitment dropped from 900 to 300 workers annually, with positions remaining unfilled. “People prefer food delivery over factory work,” noted an anonymous manager. Professor Xiong Rong views humanoid robots as solutions for “dangerous, repetitive work amid future labor shortages.”
Ethical considerations are intensifying. Wang Tianran, Academician of the Chinese Academy of Engineering, emphasizes that “constraints on robot behavior directly affect public acceptance.” Liu Zhe of Peking University notes that Isaac Asimov’s Three Laws of Robotics cannot address modern ethical complexities: “Today’s algorithmic decision-making operates as a ‘black box’ requiring transparency frameworks.”
Zhang Weinan cautions that emotional interaction capabilities could create new dilemmas, necessitating clear “tool-attribute” definitions. As humanoid robots evolve from technical marvels into societal participants, their development trajectory will profoundly reshape human self-perception and civilizational foundations.