An exploratory study to design and evaluate a dialogue system for a robot assisting the elderly in a care environment

With a growing gap between the number of care professionals to the elderly, socially assistive robots (SARs) show a potential in helping the elderly with activities of daily living (ADL). Communication using natural language is one of the most natural forms of human-robot interaction. We conduct an exploratory study to design and evaluate a dialogue system for a care robot assisting the elderly with daily activities in a care environment. We implement a dialogue system for a care robot using the Rasa framework and tested the dialogue system with users. Our evaluations obtained an F1-score of over 60% for entity extraction and response selection to unseen data, with 50% successful task completions and 29.1% partial successes. We conducted a qualitative analysis by observing the interactions to find the limitations of the system and provide recommendations for future developments of the system.