New training centers across China run like unusual classrooms. Young workers wear VR goggles and exoskeletons. They repeat the same motions for hours<\/strong>: folding clothes, opening and closing microwave doors, wiping tables, and stacking boxes.<\/p>\n\n\n\n Sensors capture each gesture with high precision. The systems track joint movement and hand rotation. Engineers convert these motions into massive datasets. Teams use this data to teach humanoid robots in China how to handle common tasks.<\/strong> They target homes, shops, factories, and care facilities.<\/p>\n\n\n\n The scale of this robot education is vast. One of the largest facilities sits in Beijing\u2019s Shijingshan district. It spans over 107,000 square feet. Inside, researchers built scenes that mimic restaurants, warehouses, kitchens, and bedrooms.<\/p>\n\n\n\n Another center operates in Hubei province. Operators remotely control nearly a hundred humanoid robots there. Each day, the machines fold 300 pairs of shorts. They iron clothes 500 times, wipe tables 500 times, and pack 500 boxes of medicine<\/strong>. Dozens of such centers now operate across the country. Most work directly with companies building humanoid hardware and AI systems.<\/p>\n\n\n\n Chinese authorities built these centers to solve a critical shortage of training data. Unlike language models, robots need data that connects vision, movement, and task context<\/strong>. Text from the internet does not provide that link. Teams also cannot simply download this information from the web.<\/p>\n\n\n\n These \u201crobot schools\u201d aim to produce large, consistent datasets. AI systems that control humanoid robots in China need precisely labeled movement records. Standardized data from the centers can also spread across the industry. That raises the baseline for everyone, including smaller firms.<\/p>\n\n\n\n Experts say manual data creation moves slowly and costs a lot. China therefore develops digital simulations in parallel. Companies also collect \u201con-the-job\u201d data in real workplaces. State-run centers give the industry an early boost. They do it through hundreds of thousands of well-labeled movement trajectories.<\/p>\n\n\n\n China has made AI and humanoid robotics a national priority. In a few years, more than 150 startups have emerged. Examples include AgiBot, Unitree Robotics, and Galaxy General Robot.<\/p>\n\n\n\nSimulation, Practice, and Full-Scale Training<\/h3>\n\n\n\n
The Data Shortage for Humanoid Robots in China<\/h2>\n\n\n\n
From Industrial Giants to Humanoid Pioneers<\/h2>\n\n\n\n