South Korean Startup RLWRLD Records Workers’ Movements to Train AI Robot Systems

A South Korean artificial-intelligence startup is strapping body cameras to hotel staff, warehouse workers, and convenience store employees to build an AI software layer for robots. RLWRLD, based in Seoul, is collecting detailed motion data from skilled workers across industries to develop what it calls a robotics foundation model — an AI brain designed to run across a range of robots in factories, industrial sites, and eventually homes.

Among those being recorded is David Park, a nine-year veteran of the five-star Lotte Hotel Seoul, who folds banquet napkins and wipes glassware while cameras mounted on his head, chest, and hands capture every movement. RLWRLD is running similar data-collection sessions with logistics workers at CJ and food display staff at Japanese convenience chain Lawson. The company unveiled its robotics foundation model in May 2026 and expects AI robots to be deployed at industrial scale around 2028.

After converting worker footage into machine-readable data, RLWRLD engineers repeat the same tasks wearing VR headsets and motion-tracking gloves to capture fine details such as joint angles and force applied. The company is among a smaller group developing AI for five-fingered robotic hands designed to mimic human touch — a capability it sees as critical for use in homes and diverse industrial settings.

South Korea’s government announced a $33 million project to record the skills of master technicians into a database for AI-powered manufacturing, citing an aging and shrinking workforce. Major companies are moving in parallel: Hyundai Motor plans to introduce humanoid robots at its Georgia factory in 2028, while Samsung Electronics aims to convert all manufacturing sites into AI-driven facilities by 2030.

Labor groups have raised concerns that robot deployments could eliminate jobs and erode the skilled workforce South Korea is now counting on to train those same systems. Lotte Hotel, for its part, estimates humanoids could eventually handle 30% to 40% of back-of-house event preparation work, though it expects robots to be ready for cleaning tasks no earlier than 2029.

Source: mint – technology

This article was generated by AI and cites original sources.