Socially Pertinent Robots in Gerontological Healthcare
Gerontological day-care facilities are an underexplored and demanding environment for social robotics: users are vulnerable, interactions are unstructured, and the stakes of mishandled conversations are high. We report two waves of real-world experiments with a full-sized humanoid robot and over 60 patients and companions in Paris — finding broad user receptivity, with acceptance closely tied to the robot’s robustness to environmental clutter and flexibility across varied interaction styles.
1 Inria
2 Czech Technical University in Prague
3 Bar-Ilan University
4 University of Trento
5 Heriot-Watt University
6 ERM Automatismes
7 PAL Robotics
8 Assistance Publique - Hopitaux de Paris
01 The problem / 問
Social robotics in healthcare presents a uniquely demanding deployment context: the environment is unstructured and cluttered, users include cognitively and physically vulnerable populations, and the stakes of misunderstood or mishandled interactions are high. Gerontological day-care facilities — where elderly patients attend for medical supervision and social engagement — are particularly underexplored as test environments for interactive robots. The H2020 SPRING project set out to evaluate whether a full-sized humanoid robot equipped with multimodal conversational capabilities could be practically useful and accepted in such a setting.
02 The approach / 法
The paper reports two waves of real-world experiments conducted in a gerontological day-care facility in Paris, involving over 60 end-users including patients and companions. The robot’s software architecture integrates modules for speech recognition, multi-party dialogue management, person tracking, and non-verbal behaviour generation — all designed to handle the ambiguity and variability of spontaneous social interactions. Acceptability and usability were measured using the Acceptability Evaluation Scale (AES) and System Usability Scale (SUS), providing quantitative user-study data grounded in established HRI methodology. Results show that users were broadly receptive to the technology, with acceptance tied closely to the robustness of perception and the flexibility of interaction handling — the robot was most positively received when it could gracefully manage environmental clutter and varied conversational styles.