The RoMi project (16SV8436) was a German consortium effort that sought to determine how service robots could genuinely relieve nursing staff and support residents in care facilities, while also fostering acceptance of robotic assistance. The technical goal was to adapt and extend the pi4‑robot platform, workerbot 9TM, based on insights gathered from users and experts. The research team, led by the Geriatrics Research Group (CFGG) at the Charité, coordinated user‑centered studies, ethical reviews, and the integration of findings into the robot’s design. The consortium included pi4 as the robotic hardware provider, the University of Tübingen (formerly HUB) for human‑robot interaction studies, and HTWB for virtual‑reality prototyping. Funding came from the German Research Foundation (DFG) under the project number 16SV8436, and the project ran from early 2021 through late 2023, with the COVID‑19 pandemic causing notable delays in participant recruitment and field testing.

The technical outcomes of RoMi are centred on a mobile service robot capable of performing routine tasks that free nursing staff for more direct patient care. The robot can deliver beverages, bring everyday items, and remind residents of important appointments. Nursing staff issue orders through a smartphone application, receiving real‑time status updates on deliveries. Interaction modalities include spoken dialogue and a large touch display, allowing residents to communicate verbally or via touch. The design process began with extensive interviews of nursing and caregiving staff to elicit functional and aesthetic requirements. These requirements were prioritized and translated into a technical specification for the robot. A design‑thinking workshop and a subsequent ethics workshop helped refine the robot’s role and identify potential risks and benefits.

An online study conducted by the University of Tübingen examined trust‑relevant design aspects of human‑robot interaction. Results indicated that residents preferred robots that could communicate verbally, and that a more human‑like appearance enhanced perceived sympathy and competence. These findings guided the robot’s visual and behavioral design. In parallel, HTWB developed virtual‑reality prototypes that allowed stakeholders to evaluate different design variants and interaction scenarios before physical production. The VR study helped narrow down the robot’s appearance, functional features, and interaction patterns, thereby reducing development time and costs.

The final prototype was built by pi4 and integrated with the software stack derived from the user studies. A laboratory evaluation in the CFGG usability lab tested the robot’s performance in realistic care‑facility scenarios. While the report does not provide quantitative performance metrics such as task completion time or error rates, the evaluation confirmed that the robot could reliably execute beverage delivery, item transport, and reminder tasks, and that residents responded positively to the verbal and visual communication channels. The study also highlighted that robots do not replace nursing staff but rather support them by handling routine tasks, thereby allowing caregivers to devote more time to individualized patient care.

In terms of collaboration, the CFGG served as the interface between the consortium and the target user groups, ensuring continuous user involvement throughout the project. They conducted interviews, recruited participants, and managed the ethical approval process. The pi4 team focused on hardware development and integration of the software requirements. The University of Tübingen contributed expertise in human‑robot interaction and trust assessment, while HTWB provided virtual‑reality prototyping capabilities. Together, the partners produced a set of publications in national and international journals, presented findings at conferences, and developed a framework for evaluating robotic assistance in care settings. The project’s outcomes demonstrate a viable pathway for integrating service robots into routine nursing workflows, with a clear emphasis on user‑centered design, trust‑building features, and practical deployment in care facilities.