Embodied AI Data Collectors: ¥200/day to Teach Robots Physical Tasks

具身智能数据采集员：日薪200元起，给机器人当老师

Companies are hiring part-timers at ¥200–250/day to teach robots physical tasks. The work splits into two types: teleoperating a dual-arm robot to sort blocks or stack cups, or repeating motions like folding clothes while sensors record trajectories. No degree or experience required—just height/weight measurements for glove fitting and a question about VR sickness. Global high-quality physical interaction data stood at roughly 500,000 hours by early 2026, less than 1/20,000 of LLM training data, so human collectors are needed to build datasets from scratch.

Why it matters: Embodied AI data collection is a real new job category, and the article provides a concrete data-gap figure plus two collection methods — enough substance. But it's an industry landscape piece, not a product launch or tech breakthrough, so it lands at 72.