QINGDAO, April 8 (Xinhua) -- In a workshop larger than a football field in China's eastern coast city of Qingdao, a sweater-folding robot repeats the same motion thousands of times a day, a kitchen-bot practices carrying plates down the hall, and robot dogs trot through corridors learning to open doors without bumping into walls.
This is the 9,600-square-meter Qingdao Training Center for Elderly Care Robots, where machines are being taught the surprisingly delicate art of caring for aging humans. While Silicon Valley fixates on AI replacing white-collar workers, China's embodied intelligence startups have quietly shifted to a sector where demand is urgent and human-robot friction is minimal, focusing on the country's 320 million seniors.
"We're building the platform where robots master the full skill set of elderly care," said Zhang Shuai, a local civil affairs official.
Inside, human trainers guide various robots through repetitions of grasping objects, maintaining balance and twisting switches while mastering the delicate motor skills humans often take for granted. Each attempt is logged in the back end, building datasets of precision movements for when these machines eventually enter real homes.
The center has become a hub for tech ambition, with 45 companies, including Haier, Hisense and Agibot, deploying 210 products for modular training across ten categories such as medication delivery, emotional companionship and dementia intervention. The facility creates what officials describe as a closed loop where robots are trained and tested, data is shared, standards are verified, and products are then pushed to market.
But the most telling moments happen when reality intrudes.
During early trials at a local nursing home, an elderly resident muttered "I want to lean back for a bit" in thick regional dialect. The smart nursing bed sat frozen. It understood standard commands like "raise the headrest," but it had never heard that phrase before.
Technicians recorded and fed the data back. The manufacturer quickly partnered with a large language model company to rebuild the comprehension layer. This feedback loop represents something rare in robotics, a genuine co-development with end users.
Unlike factory robots operating in controlled settings, elder-care machines must navigate the unpredictability of human aging, from slurred speech and sudden falls to emotional vulnerability.
In Beijing's southeastern suburbs, that experiment has opened to the public. A pioneering "smart elderly care robot station" debuted in March inside a 1,100-square-meter community center that feels part tech showroom, part senior center.
On a recent day, a 72-year-old woman surnamed Zhang rolled up her sleeve for a massage robot whose mechanical fingers kneaded with surprising nuance. "Lighter," she commanded, and it adjusted. Nearby, elderly residents challenged chess-playing robots and watched tea-brewing automatons perform elaborate ceremonies. Downstairs, a pancake-making robot flipped perfect golden rounds in just three minutes.
Upstairs, an aging-friendly smart home demonstration apartment spans roughly 60 square meters, equipped with intelligent wheelchairs and companion robots for seniors to test. A 200-square-meter outdoor leisure area features a special robot dog obstacle course where elderly visitors can compete for fun.
Vision Seek AI, a Beijing startup, deployed emotion-sensing companion bots and AI-powered therapy devices here. "It's a natural user experience laboratory," said Xia Jing from the tech firm. "Once validated here, we can replicate this to thousands of community stations nationally."
"This station integrates basic elderly care services with robot assistance, and feeds seniors' suggestions back to manufacturers to drive product iteration," said Zhang Li, a community official.
In January, China's Ministry of Civil Affairs called for widespread adoption of humanoid robots, brain-computer interfaces, and AI to develop key technologies for disability prevention, anti-aging, daily care and rehabilitation, pushing for integrated, safe, scaled deployment of meal assistance, mobility aid, feeding, dressing, hygiene help and emergency response products.
Yet for all the optimism, industry analysts say a certain distance remains between slick demonstrations and real home life.
Wang Youhua, a gerontology professor at Southwest University, said these robots are still toddlers, struggling with dialect, environmental noise and emergency situations.
The economics also bite harder, with capable elder-care robots currently costing more than human caregivers. "For the next two to three years, nursing homes will remain the main battlefield," Wang said. "Only after technology matures and costs drop will robots gradually enter family homes."
But the bet is clear that if robots can learn to care for China's grandparents -- with all their complaints, dialects and real-world demands -- they will eventually find their market. ■
