217 Robots 'Learn' Elderly Care Skills, Developing a Sense of Proportion
China has over 320 million people aged 60 and above, with more than 50 million elderly who are disabled or semi-disabled. The country faces a shortage of up to 10 million professional nursing caregivers. Embodied intelligent robots are seen as a key solution, but they need extensive training before formally taking on these duties.
Qingdao has opened the country's first training and verification center for elderly-care robots. Covering 9,600 square meters, the center currently hosts 217 robots that are learning the "sense of proportion" required for elder care through repeated exercises.
Within the training space, robots of various forms perform basic tasks like fetching medicine, folding clothes, and pushing wheelchairs in their designated roles.
Take the simple act of fetching a water cup, for instance. It is broken down into dozens of subtle steps: locating the bottle, recognizing its shape and size, moving the chassis to the table, extending the robotic arm, adjusting the grip angle, turning around to deliver, with the delivery height aligned to the elder's palm.
This single action alone requires over a thousand hours and more than ten thousand repetitions.
During the training, operators use VR devices to guide robots and record optimal trajectories. After each task, they select high-quality data and discard invalid samples. Bao Tianzhe, an associate professor at the University of Health and Rehabilitation Sciences, points out that the difficulty lies in the individuality of elderly people: "You can't predefine procedures as in industrial settings." This greatly increases the complexity of development.
The training facility adheres to the principle of "no showroom." Yu Heng, a director at Qingdao Civil Affairs Bureau, notes that many enterprises adopt an engineer mindset, focusing on what they can make rather than what the market needs. Products that perform well in labs often struggle in real care scenarios.
One example is the home-service robot "Wanda," which can sort clothes, make beds, and play chess with seniors, yet it responds slowly to dialect-spouting elderly users. Xu Jiuling, an operations manager, says dialect is a real pain point: robots fail to understand elderly speech, and seniors often misinterpret robot replies.
Companion robots face even more complex tasks. When an elder says, "My leg hurts," a robot should not merely reply, "Have more rest." It must provide specific advice using a medical knowledge base.
Qingdao adopts a process of "government-issued challenges, real-scene assessment, and data evaluation" to select products that can only enter people's homes after passing. Zhang Shuai, deputy director of the Qingdao Civil Affairs Bureau, says so.
Currently, care robots handle mostly auxiliary tasks like delivery, folding clothes, and wheelchair pushing, along with simple entertainment. They cannot yet perform physical-contact actions such as helping elders stand up, turning them over, or massaging. Zou Yilong from Qingdao Data Group says elderly bones are fragile, and existing robots' force-feedback technology does not meet safety standards, so human interaction is not yet possible.
The industry ecosystem is taking shape. The Qingdao center offers free access and venue to enterprises, attracting 45 firms and 210 robot models, including local leaders Haier and Hisense, as well as startups like Brainco.
Shandong is building 26 provincial embodied-intelligence training grounds, gathering 41 robot companies, 28 data-model firms, and 53 scenario-applications units. The Civil Affairs Vocational University has launched a micro-major in "Emotional Companion Robot Technology for Elders" to train inter-disciplinary talent.
As 217 robots practice fetching water, chatting, and patrolling, a clearer future emerges: the first hurdle for care robots entering the market is not technological brilliance but mastering a "sense of proportion" in human interaction.