The 3rd Workshop on Nonverbal Cues for Human-Robot Cooperative Intelligence

At the Honda Research Institute Japan, we are conducting research on collaborative intelligence, human understanding, and robot systems. In recent years, generative AI has emerged, and AI technology is becoming more familiar in people's lives. For people and machines to coexist 24 hours a day, 365 days a year, it is essential for intelligent machine systems to understand people's feelings and act accordingly. To achieve this, we focus on human social activities, and our research targets are interactions between individuals, groups, and communities. We aim to advance people's happiness by helping them become the way they ought to be and enhancing their fulfillment. Here, we will introduce the research technologies at the Honda Research Institute to make people happy.

Satoshi Shigemi is the President of Honda Research Institute Japan. Since 1987, he has been conducting research on robots and control systems at Honda R&D Co. In 2000, he was the Senior Chief Engineer and project lead for the research and development of ASIMO, the humanoid robot. He then developed a high-altitude survey robot for the Fukushima Daiichi Nuclear Power Plant. He has published many papers about human-robot interaction.

Sound is an often overlooked communication modality in robotics, despite being rather easy to implement and create, especially in the era of generative audio. In this talk, I will present ongoing work on nonverbal communication in Human-Robot interaction. I will show 1) how robots can use sound to communicate intuitively and efficiently in environments where speech is not ideal, such as crowded spaces, and 2) what happens when a robot's nonverbal signals don't match, for example showing one emotion in the visual channel, and another in the audio channel. This is especially relevant since many robots don't have the same degree of expressiveness in all their modalities, affecting how a multimodal emotional expression is perceived.

Ilaria Torre is an Assistant Professor in Human-Robot Interaction at Chalmers University of Technology in Sweden, where she received a prestigious Wallenberg Academy Fellowship in 2025. Her research centers around improving communication between humans and robots. This includes both verbal (e.g. designing appropriate voices for robots) and nonverbal (e.g. designing legible behaviours to communicate information intuitively and effectively) communication. She is also interested in questions revolving around ethics and sustainability, for example using robotics to reduce gender stereotypes in society, and understanding the tricky relationship between environmental costs and social benefits when deploying social robots. Before joining Chalmers, she was a postdoctoral researcher at KTH Royal Institute of Technology, Sweden, and a Marie Skłodowska-Curie postdoctoral research fellow at Trinity College Dublin, Ireland, after obtaining her PhD from the University of Plymouth, UK, in 2017.

This talk discusses how robots can use nonverbal modalities beyond speech, such as weight variation, morphology, thermal expression, and physical contact cues, to foster empathy and cooperation with humans. In particular, we focus on how embodied AI systems can communicate social and emotional information through nonverbal modalities. We also discuss how this research on nonverbal communication advances our latest project, the Agentic Family System.

BSc from Waseda University, MSc and Ph.D. from Tokyo Institute of Technology Sony Corporation, Sony Intelligent Dynamics Laboratories [2003-2008] University of California, San Diego [2004-2007] University of Tokyo [2013-2014], and University of Tsukuba [2008-2013, 2014-now] Program Chair of the ACM/IEEE International Conference on Human-Robot Interaction (HRI 2027) General Chair of the ACM/IEEE International Conference on Human-Robot Interaction (HRI 2028)

Effective human–robot collaboration requires robots to interpret more than explicit commands. Human gaze, motion, interaction forces, and physiological responses can provide important information about intention, attention, comfort, and cognitive state. This talk presents a series of studies investigating how such nonverbal cues can support adaptive robot behavior across physical human–robot interaction and teleoperation. The first part of the talk focuses on active physical human–robot interaction, where the robot’s actions directly influence the human’s physical state. We examine how interaction conditions and robot behavior affect human responses, and how objective and subjective measures can be used to better understand perceived safety, comfort, and trust. The second part considers teleoperated manipulation, including gaze-informed grasp assistance and graph-based approaches for reasoning about objects and their relationships within a scene. These studies show how combining human cues with task and environmental context can help robots infer user intent, provide appropriate assistance, and adapt the level of autonomy during interaction. These two research directions provide opportunities for multimodal human-state estimation and personalized shared autonomy in future collaborative robotic systems.

Dr. Yue Hu is an Assistant Professor in Mechanical and Mechatronics Engineering at the University of Waterloo and director of the Active & Interactive Robotics Lab (AIRLab) since 2021. She holds a PhD in Computer Science from Ruprecht Karl University of Heidelberg, with postdoctoral training at Heidelberg and the Italian Institute of Technology (IIT). Prior to joining Waterloo, she held academic and research positions in Japan, including as a Japan Society for the Promotion of Sciences (JSPS) fellow at the National Institute of Advanced Industrial Science and Technology (AIST) and Assistant Professor at Tokyo University of Agriculture and Technology (TUAT). Her research bridges physical-social human-robot interaction, collaborative and humanoid robotics, and shared control and autonomy, with a recent interest in cybersecurity and privacy in robotic systems. Dr. Hu served as one of the co-chairs of the IEEE-RAS Technical Committee on Model-based Optimization for Robotics until 2025, and currently serves as one of the Associate Vice-Presidents of the IEEE-RAS Members Activities Board (MAB).

Robots should move in ways that actually mean something. Not scripted gestures bolted onto a task, but real, controllable expression, the kind of subtle shift in posture or motion that tells it is paying attention, hesitating, or engaging. This talk shares what we have learned about expressive robot control: how to give robots a body language of their own, and what happens to the way people see and relate to a robot once it has one. Drawing on longitudinal work watching how people's sense of "what this robot even is" changes over weeks of interaction, the talk explores what that tells us about designing expressive motions and how ready we are to accept them.

Gentiane Venture is a Professor of Robotics at the University of Tokyo, where she leads the GVLab. Her research focuses on human-robot interaction, physical AI, and how humans form relationships with and categorize intelligent agents over time. Her work spans robot motion and behavior design, and cooperative intelligence, combining robotics with interdisciplinary perspectives from ethnography and philosophy.