In Zhejiang Jiaxing, a new generation of young workers is training robots to fold clothes, sweep floors, and tidy rooms. These "data collectors" serve as the human teachers behind the machines, guiding them through everyday tasks with precision and care.
At the Zhejiang Tsinghua Flexible Electronics Technology Research Institute's embodied intelligence training ground, each robot working diligently has a young person nearby. These individuals, known as "data collectors," step into simulated living spaces daily, donning VR goggles and holding mechanical claws to begin their teaching and data collection work.
Yang Huiqin, a post-2000s graduate from Jiaxing Nanhu University, started as an intern at the facility and later joined full-time. The work appears simple—standing in fixed areas and demonstrating actions for robots—but she approaches each movement with caution. To synchronize human actions with robots, the collectors must first learn to control their own habits. "Human movements are broad, and we all have unconscious small gestures. But when demonstrating for robots, the range of motion must be controlled, and those inappropriate gestures cannot be made," she says.
Tasks that come easily to humans prove challenging for robots. Initially, Yang Huiqin felt anxious near the machines, fearing that an exaggerated movement might cause damage. "I worried about harming its parts," she recalls. Working alongside robots gave her a sense of mutual connection, but she often must stand behind them, maintaining distance to avoid hindering their limb movements while monitoring tasks like flower arranging or folding clothes. She enjoys giving the robots names like "Xiao Bai" and "Jing Ling," making them feel more alive.
Despite the repetitive nature of the work, Yang Huiqin finds satisfaction in seeing the robots' accuracy and efficiency improve slowly. Within just six months, she progressed from a team member to a team leader and now serves as a project planner for data collection, coordinating her team's daily tasks to meet deadlines.
The demand for data collectors is rising alongside the growth of the embodied intelligence industry and the increasing number of robots. The institute currently employs nearly 300 collectors, mostly post-2000s graduates from local vocational schools or other regions, with some holding bachelor's degrees. Partnerships with schools offer internships, allowing students to enter the field early.
On July 30, at the Zhejiang Tsinghua Flexible Electronics Technology Research Institute's embodied intelligence training ground, Yang Huiqin trained a robot in folding clothes.
On July 30, at the same training ground, Zhu Sikai, an intern from Zhengzhou Information Technology Vocational College, trained a robot to grasp objects on a table. As a first-time data collector, he found the experience novel and hopes for greater breakthroughs in human-machine collaboration in the future.
On July 30, at the training ground, Zhu Sikai wore professional VR goggles and held a mechanical claw to control the robot's upper limb movements in sync.
On July 30, at the same venue, Zhang Ruoxuan, an intern from Zhengzhou Information Technology Vocational School, operated a screen behind a robot. That day, he trained the robot to pick up socks from a sofa and place them into a basket.
On July 30, at the training ground, a post-2000s data collector trained a robot in picking up various objects from a table.
On July 30, at the same training ground, a post-2000s data collector trained a robot to grasp an object in front of it.
On July 30, at the training ground, the post-2000s data collector Zuo Junjie trained a robot to pick up waste paper from a coffee table and throw it into a nearby yellow desktop trash bin.
On July 30, at the training ground, a robot was being trained to arrange food items on a supermarket shelf.
Comments