In collaboration with researchers and students in clinical psychology, medicine, and industry, we investigate how conversational agents and robots can make it easier for people to express personal experiences and emotions. This work aims to identify interaction strategies that support reflection and the development of long-term human–AI relationships.

A joint patent application has been filed based on this research.

Keywords: Self-Disclosure, Empathic Interaction, Reflection, Personalization

We developed a large language model-based mediation system designed to support communication between people with different values or perspectives. Instead of deciding which person is correct, the AI introduces empathetic expressions that help each participant recognize the other’s position. Our findings suggest that such interventions can reduce interpersonal tension and support more creative and sustained dialogue.

Keywords: AI-Mediated Communication, Empathy, Conflict Resolution, Large Language Models

Robots should be more than information-delivery devices. They can also help people reconsider assumptions, encounter alternative viewpoints, and engage in deeper reflection. In this project, we examine how interacting with multiple robots that express different perspectives affects human judgment and openness to alternative ideas.

Keywords: Human–Robot Interaction, Multi-Robot Systems, Decision Support, Reflection

How does human judgment change when several robots agree—or when one robot disagrees with the others?

This research investigates how consensus and dissent among robots influence trust, conformity, confidence, and deliberative decision making. The findings provide design implications for multi-agent systems that expose users to diverse perspectives without creating excessive social pressure.

Keywords: Multi-Robot Interaction, Trust, Conformity, Deliberation

We developed machine-learning models that estimate a person’s quality of life from video, speech, and conversational content collected during interactions with communication agents and robots. This approach enables lower-burden and more continuous assessment than conventional questionnaire-based methods and provides a foundation for personalized well-being support.

Keywords: Machine Learning, Multimodal Sensing, Quality of Life, Human–Robot Interaction

Using counseling data collected from professional therapists, we investigated conversational factors associated with empathetic interaction. We also evaluated telecounseling systems using messaging platforms and virtual reality, as well as counseling systems implemented in communication robots. Our studies demonstrate the potential of conversational technology to reduce negative affect and support mental health-related quality of life.

Keywords: Empathetic AI, Dialogue Systems, Counseling, Mental Well-being

TEMOCHIMORI is a palm-sized living moss object that gently changes its light in response to physiological signals such as heart rate. Rather than displaying bodily data as numbers on a screen, it transforms an invisible human state into an ambient expression carried by a living form. By integrating biosensing, tangible interaction, and nature, we explore new ways for people to become aware of their own condition, share it with others, and feel a deeper connection with the living world.

Keywords: Human–Nature Interaction, Ambient Biofeedback, Physiological Computing, Tangible Interfaces, Well-being

The transition to parenthood can bring major emotional and practical changes to a couple’s relationship, yet early signs of strain are often difficult to recognize. We analyze prosodic and acoustic features of partners’ voices—including pitch, loudness, and spectral dynamics—to investigate how everyday speech may reflect relationship dynamics during pregnancy and after childbirth. Our long-term goal is to develop unobtrusive and privacy-conscious technologies that enable earlier, more personalized support for couples and families before difficulties become severe.

Keywords: Perinatal Well-being, Couple Relationships, Speech Analysis, Affective Computing, Preventive Family Support