Interactive service robots: Developing evaluation criteria for interaction quality in guiding daily life activities

Abstract, Yoshiaki Mizuchi:

Interactive service robots must effectively convey information and procedures to users, mainly when teaching work procedures and requesting assistance in case of errors. To build and enhance such interaction capabilities for service robots, the evaluation of interaction quality is essential.

However, evaluating interaction quality in daily life scenarios poses several challenges, primarily due to its subjective nature. These include:

• the time and cost associated with conducting interaction experiments to collect data for analysis and evaluation
• the inefficiency of questionnaire-based subjective evaluations in situations requiring prompt results
• the ambiguity surrounding which evaluation metrics humans prioritize when guided by robots in their actions.

To address these challenges, we have collected multimodal interaction data through a robot competition task, in which a robot guides human behavior, utilizing a cloud-based VR platform. Furthermore, we examined evaluation metrics for interaction quality and a criteria determination process for fair and efficient evaluation. In this talk, I will present findings from case studies on the evaluation of interaction quality in the context of guiding daily life activities.

Bio:

Yoshiaki Mizuchi received the B.E., M.E., and Ph.D. degrees from Soka University, Tokyo, Japan, in 2010, 2012, and 2016, respectively. He was a Project Researcher at the Principles of Informatics Research Division, National Institute of Informatics, Tokyo, Japan, from 2016 to 2021. He is currently an Assistant Professor at the College of Engineering, Tamagawa University, Tokyo, Japan. His current research interests include human-robot interaction in daily life environments and the development of service robots utilizing immersive virtual reality techniques.