Never before has the speed of technological development made people feel that time passes faster than it does today. Concepts once considered science fiction decades ago are gradually becoming familiar images in everyday life. Artificial intelligence has gone beyond computer screens, present in kitchens, offices and even on the streets. In that vast picture, human-shaped robots emerge as a clear symbol of the ambition to bring technology into the physical world, where every movement requires accuracy, flexibility and adaptability.
Unlike purely software systems, human-shaped robots must learn how to exist in real space, interact with objects and environments like humans. This makes the development process much more complex. To make a robot able to open microwave oven doors, fold clothes or arrange things neatly, it needs to be "teached" by thousands of repetitions of seemingly simple operations. Behind each smooth movement is a huge amount of data about movement, arching angles, impact force and visual feedback.
In China, where robot technology is considered a pillar for the next development phase, classes for humanoid robots are being built on an increasingly large scale. This is not only a story about technical innovation, but also associated with the goal of improving productivity, compensating for labor shortages and affirming its position in the technology competition with the US.
Heartbreaking class for robots in China
Inside a brightly lit office in Shanghai, a computer science student spends a whole week pretending to open a microwave oven door. Wearing virtual reality glasses and a support frame on his arm, he repeats the same movement hundreds of times a day. The human-shaped robot standing next to him does not directly work, but observes, records and learns every movement. On other days, he imitates folding clothes or stacking wooden blocks. This monotonous work is the foundation for robots to understand and reproduce basic human behaviors.
Such people call themselves "network labor", playing an essential role in China's robot development strategy. The country's government considers robot training data the key to improving productivity and solving labor shortages in the future. When physical artificial intelligence was identified as a national priority, the wave of investment in humanoid robots exploded, leading to a huge demand for motion data.
The global success of large language models, trained from a huge data warehouse, has strengthened the belief that a similar approach can be applied to robots. However, data for robots is much more complex, including image information, articulation movement, rotation direction and interaction with the environment. These data cannot be easily collected from the internet or created by complete simulation.
To solve that problem, the Chinese government has built robot training centers. These facilities often occupy large areas, operated by robot companies and equipped with dozens of humanoid robots. Here, trainers recreate everyday tasks in a repetitive way to create important movement data.
By the end of 2025 in China, dozens of state-owned robot data collection centers have been announced, with many facilities in operation. One of the largest centers in the suburbs of Beijing is built on an area of more than 10,000m2, providing many simulated training settings for automobile assembly lines, smart homes and elderly care facilities. According to a project representative, this process is like teaching children to walk, requiring many practices for robots to form stable movement capabilities.
At another center in Hubei, nearly a hundred human-shaped robots are controlled by humans to practice folding clothes, ironing and wiping tables hundreds of times a day. These standardized data can be shared throughout the industry, helping small businesses access resources they are not capable of building themselves. However, experts also warn of the risk of infrastructure surplus and bubble formation when too many centers are built at the same time.
Potential of robot helpers
The rise of humanoid robots in China is not happening spontaneously. It is strongly driven by industrial policies, similar to how this country once supported the electric vehicle industry in the past. From investment funds for physical artificial intelligence, grants for access to computing resources, to opening specialized courses in universities, humanoid robots are receiving comprehensive support.
In the period when robots could not be commercialized widely, the largest orders came from the public sector. Data training centers and state-owned enterprises became the main customers, buying a large number of robots to serve research, training and testing. This approach recalls the formation of the electric vehicle industry, when state agencies and public enterprises played the role of first buyers, creating conditions for products to be completed before reaching consumers. With humanoid robots, the long-term goal is to create "machine helpers" who can support humans in production, services and daily life, undertaking boring tasks but requiring high accuracy.
However, great potential also comes with risks. The number of human-shaped robot companies in China has exceeded 100, raising concerns about overcapacity and fierce competition. The country's economic planner has warned about the risk of a bubble, showing the necessary caution even when expectations are high.
In addition, researchers still debate the effectiveness of collecting motion data from humans. This method takes a lot of time and manpower, while other approaches such as digital simulation or learning from operating robots are also being considered. Whether "data plants" will truly bring long-term economic efficiency is still an open question.
Despite many challenges, China's human-shaped robot development strategy shows a clear vision. When robots are trained to perform repetitive tasks with high accuracy, people have the opportunity to free themselves from heavy work, focusing on creative and decisive activities. Machine helpers are not ready to appear in every family, but classrooms for robots are laying the foundation for that future.
In the context of technology accelerating non-stop, humanoid robots have become vivid proof of how humans are teaching machines to adapt to the real world.From the heart-stopping movements in training centers, a new generation of robots is gradually being formed, promising to change the way people work and live in the coming years.
It's a bit of a bit of a bit of a bit of a bit of a bit.
It's a bit of a bit of a bit of a bit of a bit of a bit.
