In a sprawling facility in Beijing’s Shougang Park, a former art teacher named Fudi Luo spends her days engaged in a novel form of vocational training. CNBC reports that her students are not young people seeking to enter the workforce. Instead, they are humanoid robots, and her job is to teach them how to sort items on a factory line. Luo is one of approximately 100 instructors at the Beijing-based Humanoid Robot Data Training Center, a government-backed facility that represents the front line of China’s ambitious push to transform robots from laboratory curiosities into capable industrial and service workers.
The Mechanics of Robot Education
The training process is both painstaking and technically sophisticated. Using cameras, controllers, and motion capture technology, instructors like Luo guide their AI-enabled pupils through repetitive physical tasks, generating the training data that allows the robots’ underlying models to learn. “At first, the robot has no awareness, so I have to control it manually,” Luo explained to CNBC. “But once my movement generates data, the robot learns and then can perform the task by itself.”
A typical day for Luo involves eight hours of repetitive motions, covering a curriculum that includes housekeeping, massage, organizing store shelves, and metal repair. The work is physically demanding and mentally repetitive, a human performing the labor that will eventually make human labor less necessary. “The robot doesn’t know what being tired is, but I do!” Luo says.
On the same campus, startup Beijing Inspire-Robots Technology focuses on a particularly challenging aspect of robotic dexterity: the hand. Winston Zou, secretary of the company’s board of directors, told CNBC that a robotic hand typically requires 10,000 training iterations to master a single new skill, such as picking up an egg or lifting a piece of string. “Our current robotic hand can pick up an egg or even smaller objects and lift a string,” Zou said, describing a capability that sounds modest but represents a significant engineering achievement in the context of robotic manipulation.
A National Strategy for Embodied AI
The Beijing center is not an isolated experiment but a node in a rapidly expanding national network. The facility is backed by the city government and is part of a broader ecosystem of similar centers across China, all oriented toward the same goal: generating the vast quantities of high-quality training data needed to bring humanoid robots to commercial viability at scale.
This infrastructure investment reflects the Chinese Communist Party’s strategic designation of humanoid robots as a critical technology for the coming decade. In the same way that Beijing previously targeted electric vehicles and artificial intelligence as priority sectors, policymakers have identified embodied AI as an area where China must achieve global dominance by 2030. The results of this focus are already visible in the investment data: as EastFrontier reported, China’s robotics industry attracted a record $3.3 billion in the first quarter of 2026 alone, fueling the rapid growth of companies like Unitree Robotics, which recently filed for a $610 million IPO.
International Recognition and Competition
China’s aggressive push in humanoid robotics has not gone unnoticed by its international competitors. During Tesla’s fourth-quarter earnings call in January, CEO Elon Musk acknowledged the formidable challenge posed by Chinese manufacturers. “By far, the biggest competition for humanoid robots will be from China,” Musk stated. “China is incredibly good at scaling manufacturing.” A May 11 research report by the U.S. Chamber of Commerce and Rhodium Group characterized China’s approach as an “industrial policy of everything,” describing a shift from targeted sectoral intervention to a comprehensive strategy for technological dominance across multiple frontier industries simultaneously.
The training centers are a key enabler of this strategy. By systematically generating the embodied AI training data that robots need to learn physical tasks, China is building a data asset that will be difficult for competitors to replicate quickly. The combination of government funding, a large pool of relatively low-cost human trainers, and a clear national mandate creates a structural advantage in the race to commercialize humanoid robots.
From Training Center to Factory Floor
While the robots currently undergoing training still rely heavily on human assistance, their proponents are confident that full autonomy is within reach. Kenneth Ren, an overseas solution expert with RealMan Intelligent Technology who helps run the Beijing center, was careful to frame the technology’s purpose in terms of augmentation rather than replacement. “Our goal is to take on tasks that are dangerous to humans or repetitive work that people are unwilling or afraid to do,” Ren told CNBC. “We have no intention of replacing humans in any field.”
Whether or not that assurance holds over the long term, the immediate reality is that AI-powered robots are already moving beyond the training center and into real-world deployments. Across China, machines are currently undergoing test runs as restaurant chefs, bartenders, waiters, traffic cops, and bodega owners. As the industry reaches what analysts are calling a commercialization inflection point, the data generated in facilities like the one in Shougang Park will be the critical fuel powering the next generation of the automated workforce.