Cooking, washing dishes, wiping tables, folding clothes, restocking shelves in supermarkets—these everyday household and workplace tasks are becoming invaluable "teaching materials" for training embodied intelligence robots. The growth of this industry relies heavily on vast amounts of real-world human behavior data, and the role of embodied intelligence data collector has quickly gained traction as a result. This occupation has carved out a fresh pathway for flexible employment for ordinary people. However, as the field is still in its infancy, challenges such as inconsistent equipment quality, uneven pay scales, and a lack of industry standards are surfacing, placing the sector at a crossroads of opportunity and challenge.

Flexible Hours and "Paid Housework" Offer New Job Prospects

No need to clock in or sit in an office, with full control over one's schedule, earning money by doing chores at home—this is how Wang Xiaohua, a 62-year-old resident of Baicheng, Jilin Province, perceives the role of an embodied intelligence data collector. After retiring, Wang learned about this novel part-time opportunity during a casual conversation. After registering locally, she was handed a lightweight data-collection headband. Since then, her daily household chores—stir-frying in the kitchen, wiping tables with a cloth, bending to fold laundry by the wardrobe, scrubbing pots and pans at the sink—have all become part of her data collection scenarios. The entire process is straightforward and easy to grasp. After creating an account on the platform, Wang simply claims daily household tasks online, records her movements, and the device automatically uploads the data for quality review in the background. By the next day, she knows her effective working hours, and at month's end, her pay is deposited directly into her account. She often jokes with her friends that she is "getting paid to do housework."

Ms. Li, a mother of two, has also become a "life coach" for AI. "I collect data for about six hours on weekdays and rest on weekends, earning over 4,000 yuan a month," she said. Li expressed satisfaction with the job, as it allows her to care for her family while earning an income. Knowing that the data she helps produce will serve as "data fuel" to advance embodied intelligence robots adds a deeper sense of purpose to her work.

Robust Demand as Data Gaps Drive Industry Growth

Wang and Li are not isolated cases but rather a microcosm of the part-time data collection workforce. 2026 is being viewed as the inaugural year for large-scale embodied intelligence data, with data collection centers being established in various regions and a wave of data companies emerging rapidly. The robust market demand is rippling through to the job market. A scan of social media platforms and recruitment websites reveals a multitude of data collection postings, the vast majority being flexible part-time roles spanning real-world settings such as homes, supermarkets, factories, and hotels. The household movement data Wang collects at home ultimately flows to Beijing-based data firm Jiyuan Zhihang. The company's Chairman and CEO, Gao Shaolong, explains that the focus of embodied intelligence has shifted from basic motor skills like running and jumping to understanding and perceiving the physical world. The industry has reached a consensus that developing general-purpose embodied intelligence robots requires 10 billion hours of multi-scenario real-world interaction data. Yet, according to industry estimates, only 500,000 hours of compliant, usable real-world physical interaction data currently exist domestically, creating a massive data gap that is fueling explosive industry growth.

"The scale and variety of full-time collectors are limited, so part-time workers will remain the backbone of the data collection industry for a long time," Gao revealed, noting that the number of part-time collectors his company can deploy has approached one million. The company offers an effective hourly rate of 70 yuan for embodied intelligence data collectors, with some regions adding special subsidies, making the income quite appealing in lower-tier markets.

Ongoing Challenges Highlight the Need for Unified Standards

The appealing narrative of flexible work and earning from home is attracting many to data collector roles, but an investigation reveals that this new profession is not as effortless as advertised. Differences in hardware and collection protocols among data companies mean some workers find it taxing to repeatedly perform the same motions while wearing heavy equipment. In June, a young man named Xiaoyuan took on a part-time data collection role through an online ad. His gear included not only a somewhat bulky head-mounted camera but also a mechanical gripper to simulate robotic grasping. While wearing the camera, Xiaoyuan had to avoid sudden head movements to prevent invalid captures. Additionally, he had to keep his arm suspended in position for extended periods, and after repeating the same motion dozens or even hundreds of times, his arms grew sore and numb, while his hands ached from constant gripper use. "This is far more demanding than I imagined," Xiaoyuan said.

Moreover, the industry currently lacks a standardized pay structure, leading to income disparities across different channels, and some workers admit the pay falls short of their expectations. The investigation found that direct hires by some staffing firms, requiring 5 to 8 hours of work per day, can earn 200 to 300 yuan daily. However, others who secure gigs through intermediaries see their wages reduced. Xiao Pan started part-time data collection through an intermediary in July, with an hourly wage of 15 yuan. The physically intensive recording sessions meant a full day's effort, and after insurance deductions, she took home 147 yuan for the day.

Gao Shaolong attributes these wide variations in labor intensity and compensation to the industry's early stage and the lack of unified standards. Different companies have distinct model training objectives, requiring different data formats and collection methodologies. Encouragingly, the industry is gradually refining its software and hardware, with multiple firms developing lighter and more ergonomic collection devices. Looking ahead, Gao and his peers are optimistic about the future of data collectors as a vocation. Still, he stresses that the industry needs further regulation: soft and hardware must be continuously upgraded to lessen physical strain on workers, and universal industry norms need to be established to clarify responsibilities, safeguard the rights of flexible workers, and ensure this new occupation can both support the embodied intelligence industry and genuinely unlock employment dividends.