A Component-Based Approach for Developing Actions in Non-Humanoid Elderly Care Robots

Abstract

Developing expressive motion for non-humanoid companion robots is a growing area of research in human–robot interaction. This study presents a hybrid design approach that integrates designer-led ideation with user-based empirical validation to develop physical actions for Lemmy, a two-wheeled companion robot for elderly users. The research followed a three-stage process: designer workshops produced an initial action catalogue; pairwise comparison testing with older adults identified preferred variants of four key motion components, speed, posture, path, and turn type, and validation studies compared complete actions composed from preferred, non-preferred, and random combinations. High-fidelity 3D animation was employed to enable rapid iteration, experimental control, and parallel progress while the physical prototype remained under development. Results from 25 elderly participants showed consistent preferences for dynamic posture, smooth turning, and moderate speed, confirming that component-level optimization enhances perceived comfort and naturalness. Crucially, the findings demonstrate that once user-preferred motion components are established, they can be recombined to generate new, user-aligned actions without exhaustive re-testing. The proposed approach thus offers a scalable, designer-accessible method for creating expressive motion in non-humanoid companion robots for eldercare contexts.