Relationship Between Eating and Chatting During Mealtimes with a Robot

ABSTRACT

Eating with someone makes mealtimes more enjoyable and enriches our lives. However, lifestyle changes and the current COVID-19 pandemic have forced many people to frequently eat alone. Communication robots can be good mealtime partners. People would not worry about matching their mealtime schedules with robots, and they present no risk of disease transmission. Chatting is an important component of mealtime interaction. Thus, we developed a chatting system that can respond with natural timing and investigated the relationship between eating and talking when eating with a robot. We combined the good points of speech content recognition and volume recognition. Conversation systems only based on speech-content recognition experience long response-lag times because they use complex technologies. Using volume recognition to recognize human speech, which is faster than speech-content recognition, we aimed to reduce this lag by using filler responses, such as “I see,” before the speech-content recognition finished Using this system, we conducted an experiment to analyze the relationship between utterances and eating behaviors from the recorded videos and questionnaire answers of 25 participants. The results suggest that the recognition of picking-up and eating motions could support the recognition of utterances from humans but not necessary in deciding when robots should start talking. © 2021, Springer Nature Switzerland AG.