Method to record and analyze the operation of seal robot in elderly care

Abstract

Robot therapy, a mental health care through interactions with robots, has attracted attention as a new method of dementia care. In particular, the therapeutic seal robot named “PARO” is being widely used. When using PARO in elderly facilities, caregivers called “handlers” encourage the elderly people to interact with PARO. However, the usage of PARO was left to the handlers itself. Therefore, there was no intended effect in certain cases. To solve this prob-lem, this study aims to develop a method to record the behaviors of handlers and the reactions of elderly people during the robot therapy and a method to support planning by analyzing the recorded data. A Bayesian network was applied to analyze the relationship between the handler’s behavior and the elderly people’s reactions. To verify usefulness, the experiment was conducted at four elderly facilities between November 2019 and January 2020. The participants were 12 handlers and 21 elderly people. We observed the robot therapy using PARO for 20 min, and subsequently, conducted interviews. Consequently, a model that vi-sualized the relationship between the handler’s behav-iors and the elderly people’s reactions was obtained from 40 observed cases. The interviews confirmed that the model was useful for planning a robot therapy.