- info@digitalxnode.com
- GF 27, TDI Center, Near Jasola Apollo Metro Station 110025
In South Korea, a rapidly expanding AI startup is teaching robots to act like people by recording the actions and expertise of seasoned workers.
At the upscale Lotte Hotel Seoul, employee David Park carefully folded banquet napkins while wearing cameras on his head, chest, and hands. Every move he made was logged and saved as data that can later help train robots to carry out the same job.
The project is run by South Korean startup RLWRLD (pronounced “real world”), which is building a huge database of human workplace skills to develop advanced AI systems for robots.
These employees help gather data showing how people grip items, arrange products, move goods, and handle delicate tasks.
RLWRLD’s core aim is to create an AI software platform that can run on many kinds of robots in factories, workplaces, and eventually even homes.
The company believes one of the toughest challenges in robotics is replicating the flexibility and precision of human hands. For that reason, RLWRLD is heavily focused on developing robots with humanlike hand motion.
David Park, who has worked for years in hotel banquet services, said he often takes part in recording sessions where his techniques are captured with wearable cameras and sensors.
After folding napkins neatly, Park also showed how he cleans wine glasses and tableware while engineers logged every motion in detail.
RLWRLD is part of a growing set of South Korean firms investing in what specialists call “physical AI.”
South Korea views robotics as a major chance to boost its standing in the global AI sector.
Yet competition is fierce. Large companies from the United States and China, including Tesla, are putting billions into humanoid robots and AI-powered machines.
Just as chatbots like ChatGPT and Gemini learn from massive volumes of online text, robots also need huge amounts of real human behavior data to master physical jobs.
South Korean experts think the country has an edge thanks to its strong manufacturing base and highly trained workforce.
To back this effort, the South Korean government recently started a $33 million project aimed at capturing the skills and methods of experienced workers to help train AI-driven manufacturing systems.
The goal is to lift productivity while helping the nation address its aging and shrinking workforce.
RLWRLD recently launched its own robotics foundation model, which serves as a central AI system for robots.
The company expects industrial AI robots could start large-scale rollout around 2028.
Major corporations are preparing for the same future:
Experts say South Korea’s robotics strategy is mainly centered on humanoid robots built for industrial work settings.
Despite enthusiasm for robotics, labor unions and employees worry that AI machines could displace human jobs.
Hyundai’s labor union recently cautioned that robots might trigger a serious employment crisis. In reply, South Korean President Lee Jae Myung called AI development an unstoppable trend and urged workers to adjust to fast tech changes.
Labor leaders argue that while robots can mimic existing skills, genuine craftsmanship and innovation still originate with people.
They also warn that heavy robot adoption could limit chances for future skilled workers to gain practical experience.
Inside one RLWRLD lab at Lotte Hotel Seoul, engineers test robots fitted with black metal hands designed to closely mirror human motion.
During demos, robots slowly lifted cups, packed items into boxes, and placed objects onto conveyor belts.
Some robots still make errors, like knocking things over, but the technology keeps advancing.
Many robots today use basic gripping tools instead of full humanlike hands. However, RLWRLD is among a smaller group of companies building five-fingered robotic hands that can do more delicate work.
Experts believe these advanced hands may be especially important when robots eventually enter homes and interact more directly with people.
Hospitality workers, in particular, provide useful training data because their roles demand precision, care, and fine movements.
Although current humanoid robots are still far slower than people, companies think they will soon be helpful assistants.
For instance, cleaning a hotel room now takes robots several hours, while human staff complete the same job in about 40 minutes.
Still, Lotte Hotel expects robots to start helping with cleaning and back-of-house tasks by 2029.
The company also intends to launch robot rental services for hotels and service industries, with possible expansion into private homes later.
David Park believes robots may eventually handle around *30% to 40%* of behind-the-scenes hotel work. However, he says machines will have trouble replacing tasks that need genuine human interaction, communication, and emotional understanding.
