A son in another city checks his phone and sees a stick-figure silhouette fall and stay down on the screen. No video of his father's bare room, just a shape and an alert. By the time he calls, the community worker is already moving.
That small, privacy-preserving scene in a Beijing suburb is the kind of deployment China now wants by the thousands. The country's answer to its ageing crisis is no longer a brochure of gadgets but a national programme to put robots and AI monitoring into real homes and institutions — and to measure whether they actually work.
Key takeaways
- In May 2025 the Ministry of Industry and Information Technology and the Ministry of Civil Affairs opened a Smart Elderly-Care Service Robot Pairing R&D and Scenario Application Pilot (智能养老服务机器人结对攻关与场景应用试点), running 2025–2027, that pairs robot makers with care providers.
- The pilot sets hard deployment minimums: home robots must be verified in at least 200 households with at least 200 units deployed; community or institution robots need at least 20 sites with at least 20 units, each validated for no less than six months.
- A Beijing economic-development zone put a "smart elderly health monitoring" scene into use in 2024 using a "stick-figure" (火柴人) AI camera and smart mattresses that detect falls and abnormal vital signs without filming the room.
- Shanghai had built 77 smart nursing homes (智慧养老院) by the end of 2024 against a target of 100 by the end of 2025, layering monitoring and chronic-disease management onto daily care.
- The demographic pressure is structural: official statistics cited in 2025 reporting put China's population aged 60 and over above 300 million (about 22%) at the end of 2024, with those 65 and over at roughly 220 million.
The pilot that forces real use
The May 2025 notice, document number 工信厅联通装函〔2025〕212号, is unusual in how concrete it is. Most "robot plus elderly care" talk stays at the vision stage. This one specifies numbers, durations and who must be in the room. Robots are split across home, community and institution settings, and across needs: care for the disabled and cognitively impaired, emotional companionship, health promotion, smart environments and daily-living assistance.
The deployment thresholds are the point. A home-service robot must be tried in no fewer than 200 households with at least 200 units, and the trial must run at least six months. Community and institution products need at least 20 sites and 20 units. The ministries also ask the pairs to write standards for safety, reliability and age-appropriateness. In other words, the state is using procurement-style pressure to convert prototypes into products that have survived contact with real older people.
What is already on the ground
The pilot did not start from zero. Beijing's economic-development zone opened its first smart elderly health-monitoring scene in 2024. The centrepiece is a front-end AI camera that renders the monitored person as a "stick figure" (火柴人) on a screen — pose, movement and behaviour are analysed, but the room is not filmed, which is the privacy bargain that makes households willing to install it. The same system pairs with a smart mattress that tracks heart rate, breathing, wetness and sleep, and auto-alerts staff. Falls, night-time leaving-bed and distress calls trigger a message to the care station and to family phones.
Shanghai took the institution route. Under a 2023–2025 smart-nursing-home plan, the city had built 77 smart nursing homes by the end of 2024, heading toward a target of 100 by the end of 2025. These are not robot showrooms; they are operating homes where monitoring devices, vital-sign sensors and internet-hospital links wrap the resident's whole stay. Separately, at the December 2025 Shanghai Silver-Hair Tech Industry Week, Fourier Intelligence (傅利叶智能) demonstrated its M4 lower-limb exoskeleton, the kind of device that helps a stroke or paralysis patient relearn walking — a concrete example of rehabilitation robotics moving from lab to public event.
The corporate and local push
The private sector is being pulled in the same direction. In June 2025, with the Shanghai elderly-care technology industrial park opening in Minhang, Tencent launched an upgraded "Silver-Hair Tech Partner Programme" (银发科技伙伴计划) alongside the city's science and civil-affairs bureaus. The programme, which gathered more than 40 eldercare-technology companies, offers a dedicated R&D fund described as being in the tens of millions of yuan (≈ US$1–14 million) and commits to live testing in no fewer than 10 nursing homes.
The logic underneath all of it is arithmetic. With more than 300 million people aged 60 and over, China cannot train and place enough human carers to close the gap, and the ministries' own framing names the goal explicitly: relieve family caregiving pressure and fill the labour shortfall in communities and institutions. Robots and AI monitoring are being treated as infrastructure for a problem that will only deepen.
According to Li An, chief scientist at BrainNet (脑机网), the hard part of an eldercare robot is not the hardware but earning a place in a stranger's home without becoming a threat.
Honest limitations
The strongest fact here — the 2025–2027 pilot and its deployment thresholds — comes straight from a published government notice on the State Council's own portal, corroborated by Xinhua and People's Daily, so it is firm. The Beijing and Shanghai deployments are real but reported at the city and district level: the Beijing "stick-figure" monitoring scene is covered by Beijing-based news reporting, and the Shanghai smart-nursing-home counts come from city-policy trackers rather than a single audited national register. Treat the 77-of-100 figure as a reported progress number, not a certified total.
The demographic figures (300 million aged 60 and over, 220 million aged 65 and over at the end of 2024) are widely cited in 2025 reporting of official statistics; they describe the pressure, not the pilot's results. The Tencent programme's fund size and "more than 40 companies" are the company's own launch claims.
This article focuses on the national pilot and representative city and corporate deployments. It does not cover the many smaller municipal monitoring pilots, nor does it assess whether these robots improve outcomes or merely add devices. Six-month home trials are exactly the evidence that does not yet exist in public form, and the pilot's own requirement for that evidence is the right thing to watch.
What readers can do now
- If you care for an older relative in China, ask your district civil-affairs office about smart-monitoring pilots before buying a camera yourself — privacy-preserving "stick-figure" and smart-mattress setups are already running in places like Beijing's development zone.
- If you build or buy eldercare tech, use the pilot's thresholds as your own bar: insist on a multi-month, multi-household trial with documented override and failure rates, not a showroom demo.
- If you follow policy, watch the 2025–2027 pilot's published入围名单 (shortlist) and later assessment results — that list will be the first real map of which robots earned the right to scale.