China, which showcased its Humanoid Robot technology with a performance at the Spring Festival gala, is moving to put robots into real-world use. It is training robots to fit actual industrial sites and building data to prepare for commercialization.
According to China's state-run Global Times on the 16th, regions including Anhui, Zhejiang and Shandong are opening robot training centers one after another to commercialize new robot technologies.
The "robot school" is where Humanoid Robots are trained to be deployed in real work. One training center in Shandong is teaching dozens of Humanoid Robots to carry trays, fold clothes and take water off shelves.
Robots require more complex data learning than other fields. In particular, data needed for Humanoid Robots to work—such as joint motion and speed, rotation, visual information, touch, pressure and force—are not available online, unlike text and images used for artificial intelligence (AI) training. They must be generated directly through interaction with robots.
As of the end of last year, China had established more than 40 state-run robot data collection centers. Of these, 24 are currently in operation. Typically, they place dozens of robots in office spaces measuring thousands of square meters (㎡), have workers repeat simple tasks with the robots one-on-one, and operate by accumulating data on key actions.
Leju, a Humanoid Robot company that established a robot training center in Shijiazhuang, Hebei, with state support, created environments such as an automobile assembly line, smart home and eldercare facility in a 10,000 ㎡ space and runs 16 Humanoid Robot training programs. For example, robots wearing Virtual Reality (VR) and motion-capture equipment return empty boxes, sort materials and package products.
The data generated at this center reaches 6 million entries a year, the most in China. The functions Humanoid Robots have acquired here exceed 20, and their task success rate is said to reach 95%.
At another center in Hubei, about 100 Humanoid Robots generate data by repeating actions hundreds of times, such as folding and ironing clothes and wiping tables.
According to China's state-run China Daily, Li Chao, co-founder and chief technology officer (CTO) of the robot company Deep Robotics, recently said, "Unless you train robots in real situations, you cannot truly advance the technology," adding, "With government policies that promote robot adoption, we can try actual deployments and discover new use cases. This is the part most envied overseas."
Robot training centers also help increase real revenue for robot companies. Three data collection centers in Jiangxi, Guangxi and Sichuan delivered Humanoid Robot sales worth 566 million yuan (about 121.9 billion won) to the Chinese robot company UBTECH Robotics.
Humanoid Robots are expected to be introduced first in manufacturing, such as automobile factories, and in logistics, which has many simple tasks. U.S. robot company Agility Robotics, after supplying its Humanoid Robot Digit to logistics companies Amazon and GXO and to auto parts company Schaeffler Group, decided last month to deploy seven units to Toyota's production plant in Canada.
Agility Robotics chief technology officer (CTO) Daniel Dees said, "No matter whether it is Germany, Korea, Japan or the United States, there is a shortage of workers for simple, repetitive tasks," adding, "Reshoring of U.S. manufacturing is only possible through the combination of humans and robots."
Korea also has many fields where Humanoid Robots can be introduced, given the shortage of workers in manufacturing and simple tasks. According to the Ministry of Employment and Labor (MOEL), as of 2025, the industries with the most unfilled positions, even when an enterprise actively recruits, are manufacturing (25,000 people) and transportation and warehousing (13,000 people), in that order. The unfilled rate was also highest in transportation and warehousing at 24.2%, followed by manufacturing at 16%.