Humanoid robots at CES 2026 were a recurring theme, with several companies showcasing their latest advances in robotics and AI, including Korean companies Hyundai and LG. But there are plenty of other humanoid robot vendors that are developing such models with plans to launch commercial versions for uses ranging from manufacturing to home assistance. To get there, robot developers must first train the robots to accurately perform repetitive jobs, like folding laundry, wiping tables, and carrying goods. China is one place where such programs are well underway, with robot schools already training humanoid robots to perform actions.
A report from the Korean website Chosun Biz detailed some of these initiatives, following the Spring Festival gala, an event where Chinese vendors showed off some of the advanced capabilities of their humanoids. Various provinces in the country have facilities dedicated to training humanoid robots by having them complete specific actions, like folding clothes, repeatedly. The report mentions a robot school in Shandong where humanoids are tasked with folding laundry, carrying trays, and retrieving water bottles from shelves. At a similar center in Hubei province, actions including clothes folding, ironing, and wiping tables are used to train as many as 100 robots at a time.
Why folding clothes is so important for humanoid robotics
Consumers may want to buy humanoid robots in the future to help with household chores and handle repetitive tasks in a controlled environment. The robots may load and unload the washer and dryer, fold clothes, and tidy up the home while humans are away or engaged in other tasks. While this is a speculative scenario, the fact is that humanoid robots will need access to data from the real world to understand their surroundings and learn to handle 3D objects with precise movements. Robots will also have to be able to adjust grip and force, depending on the objects they interact with, so they can safely operate around humans. These robots should be able to pick up glasses and other sensitive items without breaking or damaging them. That’s where the tedious task of folding laundry comes into play.
U.S. robotics company Figure explained the importance of folding laundry in August 2025, when detailing a towel folding experiment for the Figure humanoid robot (image above). “Folding laundry sounds mundane for a person, but this is one of the most challenging dexterous manipulation tasks for a humanoid robot,” the company wrote. “Towels are deformable, constantly changing shape, bending unpredictably, and prone to wrinkling or tangling. There’s no fixed geometry to memorize, and no single ‘correct’ grasp point. Even a slight slip of a finger can cause the material to bunch or fall.” A humanoid learning to fold clothes has to adapt in real-time to the changing conditions. Each piece of clothing has a unique shape and configuration that the robot needs to understand. The robot would also have to coordinate its hands and fingers to grip parts of the cloth material and perform a folding routine.
China’s massive learning centers for robots
Figure also explained that the laundry folding exercises yielded notable results for the humanoid robot. The humanoid learned how to pick up towels from a pile, perform folding tasks, correct mistakes, and employ “fine manipulation skills, like tracing an edge with a thumb, pinching corners, or unraveling tangled towels” to complete the job. These experiments can ensure that the AI models that the robot uses to see the world and perform tasks assigned by humans can also work in other environments. The same robot that can fold clothes can also learn to perform more complex tasks, like working on an assembly line.
LG introduced the LG CLOiD home robot at CES 2026, showing how the robot can fold clothes, among other tasks. Separately, Hyundai announced a new Atlas humanoid robot at CES 2026, created in partnership with Boston Dynamics, that will eventually work alongside human workers in car manufacturing factories, starting with simple jobs like sorting parts, explaining that robots need to learn how to perform tasks.
Chinese humanoid robot vendors are likely using similar principles to train their models, investing in dedicated training centers. The Chosun Biz report mentions a company called Leju that has created a robot school in China’s Hebei province in a 10,000-square-meter (107,640 square feet) facility which will eventually house simulated workspaces for robots, like car assembly lines and smart homes. Humanoids will use virtual reality equipment that includes motion sensors to perform specific actions that may include moving boxes or packaging materials. This center reportedly allowed humanoid robots to learn more than 20 functions, with reported task-success rates above 95%. The center generated six million pieces of data each year, the report notes.
