China’s First Training Center for “Robot Caregivers” Addresses the Challenges of Embodied Intelligence

One of the standout performances in the 2026 Spring Festival Gala was a skit by Cai Ming titled “Grandma’s Favorite,” which sparked discussions about robotic caregiving. Recently, reporters visited China’s first practical training facility for “Artificial Intelligence + Elderly Care”—the Embodied Robots and Intelligent Rehabilitation Products Training and Validation Center in Qingdao. This center aims to tackle the challenges of implementing embodied intelligence in real-world applications.

Spanning nearly 10,000 square meters, the training center resembles a futuristic arena filled with robots: service robots, medication delivery robots, AGVs, specialized robotic dogs, companion robots, floor cleaning robots, moxibustion robots, exoskeleton robots, chess-playing robots, cooking robots, and wheelchair-pushing robots. Within this 9,600-square-meter space, 210 robots from 43 different companies are training in various scenarios across five major categories: disability and dementia prevention screening, rehabilitation for disabled individuals and children with autism, daily living assistance, emotional support, psychological intervention, companionship, emergency response, safety inspections, and autonomous vehicle coordination.

During a demonstration in the service robot coordination training room, a humanoid service robot named “Xingdong Q5,” which stands 1.65 meters tall and features a black-and-white design, was observed following a medication delivery routine. The robot navigated to a bedside, where another embodied robot took the medication and delivered it to an “elderly” figure. “The training for the medication delivery robot encompasses various skills, including grasping, moving, placing, delivering, and pouring. The items delivered are not limited to medication but also include bottles, fruits, cups, and food,” explained Zou Yilong, a trainer from Qingdao Data Group. According to manufacturer assessments, a robot requires a minimum of 10,000 hours of training. Currently, this robot has completed 100 hours of training, which involves data collection, cleaning, and labeling to create high-quality datasets, ultimately leading to the development of mature skills.

Medication management is a crucial aspect of daily services in nursing homes. By training and utilizing medication delivery robots, the workload of medical staff can be significantly reduced, allowing them to focus on more complex tasks. In another training room, trainers were teaching robots how to make beds, fold clothes, and push wheelchairs, while robotic dogs were being trained for item delivery and fire safety inspections.

“The training center is divided into different functional areas. The first floor focuses on the coordination training of embodied robots and robotic dogs; the second floor specializes in brain-computer interface technology; the fourth and fifth floors are designated for resident areas, including training and validation rooms for moxibustion and massage robots; while the sixth floor is primarily for collaborative training on physical and cognitive rehabilitation,” said Zhang Shuai, Deputy Director of the Qingdao Civil Affairs Bureau. The training center has also established a data platform that aggregates four types of data: equipment, training, patients, and rehabilitation. It has developed standardized training processes and collects various needs from welfare institutions to provide realistic training scenarios for robots. Additionally, data from the enterprises stationed at the center must be fully accessible.

“Currently, the center is training robots for medication delivery services and four-legged robotic patrols,” Zhang Shuai added.

In late 2025, the Qingdao Civil Affairs Bureau announced a recruitment notice for application scenarios of embodied intelligent robots and intelligent rehabilitation products in the elderly care sector, attracting nearly 100 companies nationwide, including leading enterprises like Beijing Zhongke Tailing, Baidu Intelligent Cloud, and New Stoneware Unmanned Vehicle Group, thereby covering both upstream and downstream enterprises.

What will the training center do after companies are onboard? “We focus on both training and validation. This involves not only verifying mature products but also assessing similar products from different manufacturers to continually improve and refine them to meet consumer and market demands,” said Yu Heng, Deputy Director of the Qingdao Civil Affairs Bureau’s Network and Data Office. Many training facilities lack real-world scenarios; therefore, the training center has created 11 training scenarios based on the rehabilitation service needs of local welfare homes, where professionals train the robots. Residents in the center, along with pharmacy staff, can directly participate in the training and validation of robots, providing immediate feedback on service quality.

“At the training center, a collaborative model has developed among the resident enterprises, fostering a rich variety of application scenarios and forming a more comprehensive ‘Artificial Intelligence + Elderly Care’ ecosystem,” Yu Heng noted. The training center has launched seven key project initiatives, including medication delivery robots for nursing homes, safety inspection robots, fire inspection robots, home care robots for empty-nest seniors, and diagnostic rehabilitation devices for Alzheimer’s patients, attracting collaboration from multiple enterprises. For example, China Mobile, Hisense Group, and Qingdao Data Group are jointly working on the home care robot project for empty-nest seniors, addressing critical challenges together.

Experts suggest that the next phase of brain-computer interfaces will transition from purely medical scenarios to consumer-level applications and capability enhancement, with embodied intelligence serving as an optimal medium. On the second floor, in the brain-computer interface field, Yuanshu Technology showcased a series of newly developed portable EEG recording and analysis systems that integrate brain-computer interfaces, neurofeedback, and other related functions, making research more accessible and cost-effective.

In addition to embodied robots, the training center features a variety of high-tech intelligent rehabilitation products, such as voice-controlled smart beds that can lie flat and elevate, along with mobile washbasins and dining tables for the convenience of disabled patients; home adaptation bathing equipment with smart temperature-controlled showerheads; anti-decubitus and sleep-assisting massage water beds; smart nursing turning beds; and anti-snoring pillows for restful sleep. These innovations aim to enhance rehabilitation treatments through advanced technology.

The training center aims to establish a complete closed-loop system from production to application. Many enterprises developing robots often struggle with practical applications, and the training center addresses the disconnection between R&D and market application. After more than a month of training and validation, some robots that do not meet market needs are quickly “eliminated.” For instance, one manufacturer’s medication box was too small to meet the needs of nursing homes for medication management. “The medication delivery robots currently in training are continuously being improved; for instance, we need to redesign the box door from a hinged to a pop-up style for easier robot access,” Zhang Shuai stated.

At the same time, the advantages of many other robotic products have been uncovered. For example, a consumer-grade AI-powered exoskeleton robot, due to its moderate price, long-lasting power, and stable assistance, not only benefits rehabilitation training for children in welfare institutions but has also received positive feedback from consumers. Additionally, staff discovered that wheeled robots, with their stable chassis and faster movement, are more suitable for servicing nursing homes compared to legged robots.

“Previously, enterprises operated in isolation, resulting in varying data quality. Now, through standardized and large-scale data production, we aim to establish a leading high-quality dataset in the rehabilitation field in China, providing high-quality data services to industry enterprises,” Zhang Shuai explained. Currently, the data collected by the training center primarily targets residents within welfare homes. In the future, the center will continuously adjust the resident enterprises and enrich training subjects and scenarios, expanding validation scenarios to all 297 nursing institutions across the city, thus injecting more “data fuel” into the embodied intelligence industry.

Moving forward, the training center plans to invite all nursing institutions and resident enterprises for discussions and exchanges, encouraging broad participation in training validation, and facilitating deep supply-demand matching. This will form a complete closed-loop system of “training testing + functional rehabilitation + data sharing + standardized validation + market promotion,” promoting the deep integration of government, industry, academia, research, and application in the embodied intelligence sector, thereby accelerating the benefits of embodied intelligence for countless families.