A Concept of Emotion-Driven Human-Robot-Inter-Action for Nursing Care Scenarios

Abstract

‘Social robotics’ is a fairly recent field of robotics addressing questions of how mobile robotic platforms must be extended with sensors and data analysis possibilities, enabling them to recognize and interpret the emotional state of their human counterparts and the current situational context to react in an adequate manner. One key application scenario for social robots is in the field of nursing homes and elderly residents. This implicates that social robots must be able to interact with diverse types of roles, ranging from the elderly home inhabitants via relatives and guests to professional health care providers. In order to communicate adequately with humans in a social manner, the interaction strategies must include the information about the emotional states of humans. In this contribution, we propose a concept of an emotion-driven human-robot-interaction scheme, where possible perceived emotional expressions (such as ‘happy’, ‘surprised’ or ‘sad’) are used to trigger corresponding reaction strategies from a repository, which can then be used to react and stabilize the situation. Using the scenario of nursing care as an example, all necessary technical components of social robots are described, and possible dialogues between nursing home residents and robot are simulated with a large language model, whose prompts are enriched by additional information about the possible detected emotional states of the residents.