A semantic–kinematic mapping framework for user-centered transfer-assist nursing robot design

Abstract

Abstract To address the limited integration of affective expectations into transfer-assist nursing robot design, this study proposes a semantic–kinematic affective design framework and applies it to the development of a transfer-assist nursing robot. The framework first constructs a dynamic semantic lexicon through semantic descriptor collection and KJ clustering, thereby identifying users’ core affective expectations for robot-assisted holding and transfer motions. Second, OpenCV-based keyframe extraction, semantic differential evaluation, and imagery transition lines are integrated to establish the matching relationship between core semantic descriptors and representative motion patterns. Third, OpenPose is used to extract upper-limb keypoints, including the shoulders, elbows, and wrists, and two-dimensional motion contours are generated through temporal superposition to provide references for robot form design. Finally, the analytic hierarchy process is applied to determine the weights of design elements, including color, form, and function. Taking a transfer-assist nursing robot as an application case, the proposed process involved nine core dynamic semantic descriptors, 140 keyframes, and multi-level design-element weights. Finite element analysis and a preliminary user evaluation of the resulting robot concept indicated structural feasibility and generally positive perceived qualities. These evaluations provide preliminary evidence regarding the feasibility and perceived characteristics of the resulting design rather than comparative validation of the framework. The study provides a traceable hybrid quantitative–qualitative design pathway for incorporating user-perceived motion semantics into the early-stage design of assistive robotic systems.