Published papers

This study investigated the effects of a method of anthropomorphizing a target space using attachable robotic human parts. As a result of this method, users may perceive a space as having lifelike characteristics and may accept a virtual body image for various objects. We developed robotic human parts to evaluate our method and conducted an experiment in which subjects positioned objects according to instructions given within an anthropomorphic space to determine their acceptance of the virtual body images presented and the anthropomorphic representation of the space. The results showed that subjects' positioning of objects changed according to the position of the human parts. We found that our method successfully generated a virtual body image for a particular region that would not normally be recognized in this way by users.

We propose a '' display robot '' that directly anthropomorphizes objects using body parts that are like those of humans. It is constructed of devices that look like eyes and arms. It anthropomorphizes objects according to the places they are located at or the situations they are in, and this increases their subjectivity and their virtual body image. The display robot enables users to accept the body images of objects and communicate smoothly with them. We assessed user evaluations of the display robot in an exhibition space and in this paper we discuss the effect of anthropomorphization achieved by these devices.

In this study, we propose anthropomorphization framework that determines object's body image by attaching it to human-like body parts. Our purpose is that an object acquires subjective representation by anthropomorphization and it communicates with humans. Although, there are many studies on anthropomorphic agents to communicate with humans via an object, it's unnatural because these agents are unrelated to the object directly. In our method, the object is directly anthropomorphized so it can conduct task-related communication using its intrinsic body image. We designed anthropomorphization framework that changes body image by attaching places of body parts. We also conducted an experiment to evaluate this framework. The result indicates that an attached place of anthropomorphize device influences human's perception for body image.

This paper proposes a new communication model based on physical world objects. In usual human-robot interaction, a human looks on the robot's behaviors without responding to them and seldom immerses him/herself in the interaction. The communication model achieves human's immersion in interaction with two types of behaviors. The first draw's human's attention to a physical world object or a physical world state. The second stimulates a human's feeling under the attention. The novel idea of the model is to employ the actual physical objects to mediate the human-robot interaction. In particular, since the second behavior relates directly to a sort of human's sensory perception, it is easy for a human to become aware of the referent of the robot's behavior. As a result of the two behaviors, he/she begins to infer robot's intentions and immerses him/herself in the interaction.

We held our 6th annual AIWolf international contest to automatically play the Werewolf game “Mafia”, where players try finding liars via conversations, aiming at promoting developments in creating agents of more natural conversations in higher level, such as longer contexts, personal relationships, semantics, pragmatics, and logics, revealing the capabilities and limits of the generative AIs. In our Natural Language Division of the contest, we had eight Japanese speaking agent teams, and five English speaking agents, to mutually run games. By using the game logs, we performed human subjective evaluations, win rates, and detailed log analysis. We found that the entire system performance has largely improved over the previous year, due to the recent advantages of the LLMs. There are several new ideas to improve the way using LLMs such as the summarization, characterization, and the logics outside LLMs, etc. However, it is not perfect at all yet; the generated talks are sometimes inconsistent with the game actions. Our future work includes to reveal the capability of the LLMs, whether they can make the duality of the “liar”, in other words, holding a “true” and a “false” circumstances of the agent at the same time, even holding what these circumstances look like from other agents.

Learning-by-teaching has been proposed as an effective method for enhancing learners' motivation by enabling them to deepen their understanding by teaching others. Prior research has demonstrated that this effect extends to scenarios in which learners teach virtual agents instead of humans, underscoring the educational potential of virtual agents. Consequently, designing agent-based environments that effectively support learners' motivation has emerged as a key challenge. This work focuses on the number of teachable agents as an environmental factor influencing learners' experiences. We developed a virtual classroom in which participants delivered lessons to either 1- or 20-agent students. Drawing on cognitive appraisal theory, we examined whether the stress induced by an increased number of agents functions as eustress - a positive form of stress that enhances performance - or distress - a negative form that impairs it. We further investigated how these types of stress influence learners' psychological states and behavioral outcomes. An experiment was conducted with 16 university students under two conditions: teaching a single agent or teaching 20 agents. The participants' levels of eustress and distress were measured via standardized questionnaires, and learning outcomes were evaluated via subjective reports, test scores, and teaching durations.

This paper examines how coordination emerges when multiple users share control of a cybernetic avatar under constrained nonverbal channels. Across three observational studies in stylized testbeds, we identify two recurring regularities: the Maxim of Symbolization, whereby users compress task-relevant intentions into compact, reusable signals, and the Maxim of Hesitation, whereby users avoid direct overrides and often prefer indirect or delayed correction, especially with unfamiliar partners. Rather than claiming ecological representativeness of the tasks, we use these studies to derive design implications for multi-user teleoperation and cybernetic avatar systems. We conclude by specifying the boundary conditions under which these implications are most likely to apply.

Cybernetic avatars (CAs) play a key role in Japan's vision for Society 5.0 - a society where cyberspace and the physical world are closely intertwined to address social challenges. In-body cybernetic avatars, a sub-category, are remote-controlled micro-/nano-scale avatars or robots operating inside organisms, such as the human body. To explore future applications and implications of interactions with these technologies, we conducted an expert workshop (N = 16) in Japan. Experts co-created visions of a desirable future related to in-body CAs, identifying types of interactions, benefits, and potential risks. In a second study, we engaged an international student audience (N = 174) with three expert-developed scenarios to assess non-expert perceptions beyond Japan. We find that non-experts responded positively to all three types of in-body CA interactions. In addition, our study contributes to a better understanding of potential benefits and risks of in-body CAs from experts' perspectives and sheds light on which factors are perceived as most impactful by non-experts.