Nishio Toshiaki, Yoshikawa Yuichiro, Sakai Kazuki, Iio Takamasa, Chiba Mariko, Asami Taichi, Isoda Yoshinori, Ishiguro Hiroshi
Department of Systems Innovation, Graduate School of Engineering Science, Osaka University, Osaka, Japan.
Faculty of Engineering, Information and Systems, University of Tsukuba, Tsukuba, Japan.
Front Robot AI. 2021 Mar 22;8:633045. doi: 10.3389/frobt.2021.633045. eCollection 2021.
In recent years, communication robots aiming to offer mental support to the elderly have attracted increasing attention. Dialogue systems consisting of two robots could provide the elderly with opportunities to hold longer conversations in care homes. In this study, we conducted an experiment to compare two types of scenario-based dialogue systems with different types of bodies-physical and virtual robots-to investigate the effects of embodying such dialogue systems. Forty elderly people aged from 65 to 84 interacted with either an embodied desktop-sized humanoid robot or computer graphic agent displayed on a monitor. The elderly participants were divided into groups depending on the success of the interactions. The results revealed that (i) in the group where the robots responded more successfully with the expected conversation flow, the elderly are more engaged in the conversation with the physical robots than the virtual robots, and (ii) the elderly in the group in which robots responded successfully are more engaged in the conversation with the physical robots than those in the group in which the robots responded with ambiguous responses owing to unexpected utterances from the elderly. These results suggest that having a physical body is advantageous in promoting high engagement, and the potential advantage appears depending on whether the system can handle the conversation flow. These findings provide new insight into the development of dialogue systems assisting elderly in maintaining a better mental health.
近年来,旨在为老年人提供心理支持的通信机器人越来越受到关注。由两个机器人组成的对话系统可以为老年人提供在养老院进行更长时间对话的机会。在本研究中,我们进行了一项实验,比较两种具有不同类型身体的情景式对话系统——实体机器人和虚拟机器人,以研究将此类对话系统实体化的效果。40名年龄在65岁至84岁之间的老年人与实体桌面大小的人形机器人或显示器上显示的计算机图形代理进行了互动。根据互动的成功程度,老年参与者被分成不同的组。结果显示:(i)在机器人以预期的对话流程做出更成功回应的组中,老年人与实体机器人的对话比与虚拟机器人的对话更投入;(ii)机器人做出成功回应的组中的老年人比机器人因老年人意外发言而做出含糊回应的组中的老年人与实体机器人的对话更投入。这些结果表明,拥有实体身体在促进高参与度方面具有优势,而且这种潜在优势似乎取决于系统是否能够处理对话流程。这些发现为开发有助于老年人保持更好心理健康的对话系统提供了新的见解。