The RAPit subproject of the AdaMeKor consortium delivered a comprehensive set of technical and scientific outcomes for an adaptive multi‑component robotic system aimed at assisting patient transfer in care settings. Central to the effort was the development and evaluation of three distinct control modalities for the KUKA Jaco robotic arm: a conventional gamepad, a SpaceMouse, and a voice‑controlled interface. A systematic quantitative and qualitative comparison of these modalities was carried out with a cohort of care recipients and caregivers, yielding actionable insights into usability, learning curves, and error rates that will inform the design of personalized control schemes for future deployments. In parallel, the team produced an annotated dataset of robotic arm motion sequences, which has been made publicly available to the research community and can serve as a benchmark for motion‑planning algorithms and machine‑learning models.
A key innovation was the introduction of a neural‑network‑based motion‑prediction module that reduces the degrees of freedom of the arm by proposing the most probable next movement. Early tests indicate that the network can narrow the search space by up to 60 % while maintaining a prediction accuracy above 90 % for the most common transfer motions. This capability is expected to streamline the control interface and reduce cognitive load for users.
The project also advanced the design of a robotic care bed that incorporates an automated segment mattress. Each mattress segment can be independently positioned in height, allowing the bed to form a stable seating niche, lift a patient for standing, or adjust the head‑rest for comfort. CAD models and full‑scale animations of two selected concept ideas were produced, demonstrating the feasibility of integrating the segment mattress with the robotic arm and the existing bed infrastructure. The active support system, which provides a handhold for the patient, was shown to facilitate a smoother transition from bed to wheelchair, thereby reducing the risk of falls.
The integration of the robotic arm onto a nightstand and its subsequent coupling with the partner’s care bed at the University of Oldenburg enabled a realistic demonstration platform. Evaluation sessions conducted on this demonstrator informed the refinement of control modalities and highlighted the need for adaptive force feedback during patient handling.
The RAPit project ran from 15 March 2020 to 14 September 2023 under the German Federal Ministry of Education and Research (BMBF) grant 16SV8534. It was coordinated by Dr. Serge Autexier of the Cyber‑Physical Systems and Robotics Innovation Center in Bremen, with contributions from the German Research Center for Artificial Intelligence (DFKI) and the University of Oldenburg. The consortium’s modular structure allowed the nine work packages (AP1–AP9) to progress in parallel, even under pandemic‑induced restrictions. While the inclusion of end users in the design process experienced some delays, the project ultimately produced 18 concept sketches, a comprehensive evaluation matrix based on 23 criteria, and a set of design guidelines for future robotic transfer solutions. The results are documented in a final report and have been presented at several conferences, ensuring that the knowledge generated will be accessible to both academia and industry stakeholders working on assistive robotics for healthcare.