By Zoey
Inside a luxury banquet hall at Seoul’s five-star Lotte Hotel, hotel employee David Park carefully folds a white banquet napkin into a perfectly layered square while cameras strapped to his head, chest, and hands record every movement in precise detail.
For Park, the task is routine work refined through nearly a decade in hospitality. For South Korean artificial intelligence startup RLWRLD, however, every finger movement, wrist rotation, and subtle adjustment represents valuable data that could eventually teach robots to perform the same task with human-like precision.
The company is part of South Korea’s rapidly expanding push into what technology firms are calling “physical AI” — machines powered by artificial intelligence that can perceive, make decisions, and physically interact with the real world rather than simply process digital information.
RLWRLD is building a massive database of human skills collected from workers across industries, ranging from hospitality and logistics to retail and manufacturing. The goal is to create an advanced AI software system capable of powering robots that may eventually work in factories, warehouses, hotels, restaurants, and even private homes.
At the hotel in Seoul, Park is among several hospitality employees whose daily routines are being transformed into machine-readable training material. Engineers record workers folding linens, polishing wine glasses, arranging cutlery, and preparing banquet halls as part of a broader effort to teach robots the complex dexterity required for delicate physical tasks.
“I’ve been doing this about once a month,” Park said as he adjusted the motion-tracking devices attached to his body.
Nearby, hotel staff prepared for real guests while engineers monitored streams of motion data being fed into RLWRLD’s growing AI systems. Park joked that the cameras strapped to his hands felt uncomfortably tight, but the recordings continued as he methodically wiped glasses and arranged silverware with practiced efficiency.
The startup is gathering similar data from employees at major South Korean logistics company CJ Group, where warehouse workers are filmed lifting, gripping, and sorting packages. It is also working with Japanese convenience store chain Lawson to study how workers stock shelves and organize food displays.
The effort reflects a growing belief within South Korea’s technology sector that the next major AI revolution will not center only on chatbots or software assistants, but on robots capable of physically operating in human environments.
Unlike traditional industrial robots designed to repeat narrow factory tasks, physical AI systems are intended to adapt to unpredictable situations, manipulate objects, and function with increasing autonomy.
The competition has become one of the most fiercely contested sectors in global technology.
American companies including Tesla and numerous Chinese robotics firms are investing billions of dollars into humanoid robots and advanced automation systems. South Korean companies, however, believe the country’s manufacturing expertise gives them a unique advantage.
Industry leaders argue that while U.S. firms dominate large language models such as ChatGPT and Gemini, South Korea can compete more effectively in robotics because of its deep industrial base and highly skilled workforce.
The South Korean government has embraced that strategy aggressively.
Last month, officials unveiled a $33 million national initiative aimed at capturing the “instinctive know-how” of experienced workers and converting it into AI training systems for manufacturing robots. Policymakers hope automation can offset the country’s aging population, shrinking labor force, and slowing productivity growth.
Major corporations are already setting ambitious deployment targets.
Hyundai Motor Company plans to introduce humanoid robots developed by its robotics subsidiary Boston Dynamics into factories beginning with its Georgia production facility in 2028. Meanwhile, Samsung Electronics says it aims to transform all manufacturing facilities into AI-driven factories by 2030 using humanoids and specialized robots across production lines.
Billy Choi, a professor at Korea University’s Human-Inspired AI Research Center, said South Korea’s manufacturing-heavy economy naturally positions the country to focus on industrial humanoids.
“South Korea has a highly developed manufacturing sector and the focus is squarely on humanoids tailored specifically for those industries,” Choi said.
Still, the rapid push toward automation has sparked growing anxiety among labor unions and workers worried about job losses and the erosion of skilled professions that have long defined South Korea’s economic success.
Earlier this year, Hyundai’s labor union warned that widespread deployment of humanoid robots could trigger what it described as an “employment shock.” The concerns became so prominent that President Lee Jae Myung publicly addressed the issue, describing artificial intelligence as an unstoppable technological force and urging workers to adapt to changes arriving faster than expected.
Labor groups argue that while AI may replicate technical skills, human craftsmanship and creativity remain irreplaceable.
Kim Seok, policy director at the Korean Confederation of Trade Unions, warned that excessive dependence on robots could undermine the development of future skilled workers by reducing opportunities for humans to learn trades firsthand.
“Mastery of skills is ultimately a human achievement,” Kim said. “Even if AI can replicate existing abilities, the continuous development of craft will remain fundamentally human.”
At RLWRLD’s research facility, engineers are attempting to bridge the gap between human movement and machine capability through increasingly sophisticated data collection systems.
Once footage from workers is converted into usable digital information, engineers repeat the same actions themselves while wearing virtual-reality headsets, motion-tracking gloves, and body sensors. The resulting data allows the company to measure tiny details such as finger pressure, grip strength, joint movement, and hand positioning.
The company’s laboratory occupies part of a cluttered upper-floor suite inside Lotte Hotel, where tangled wires, laser tracking systems, computers, and prototype robots fill former luxury rooms.
During demonstrations, engineers guide wheeled robots equipped with human-like black metal hands through tasks such as lifting cups, opening boxes, and organizing small objects. Some tests remain imperfect, with robots occasionally dropping or knocking over items. Yet the progress has accelerated rapidly.
Recent internal footage showed a humanoid robot carefully opening a cardboard box, placing a computer mouse inside, sealing the package, and placing it onto a conveyor belt.
One of RLWRLD’s main ambitions is developing advanced five-fingered robotic hands that more closely imitate human touch and flexibility.
Most industrial robots today rely on simpler two- or three-fingered grippers designed for repetitive manufacturing work. RLWRLD believes more sophisticated human-like hands will eventually become essential for robots expected to operate in homes, hotels, hospitals, and service industries where delicate handling and close human interaction are required.
Experts say the hospitality sector offers especially valuable training opportunities because hotel work often combines precision, adaptability, and social awareness.
Although current humanoid systems remain far slower than human workers, hotel executives believe progress is moving quickly. RLWRLD estimates that robots could eventually handle a significant portion of behind-the-scenes hotel operations, including cleaning, preparation, logistics, and room servicing.
Lotte Hotel executives hope robots will begin assisting with operational tasks by 2029 and are exploring future robot rental services for hotels and other service businesses.
Even supporters of automation acknowledge limits remain.
Park, reflecting on his own work preparing banquet halls and interacting with guests, said robots may eventually handle repetitive support tasks but are unlikely to fully replace the human side of hospitality.
“If you look at the entire process of preparing for an event in back-of-house areas, we think humanoids might be able to take over about 30% to 40% of that workload,” Park said. “It will be difficult for them to replace the remaining work, which involves actual human-to-human interaction.”
Original article
Comments