ABSTRACT. Social robots designed for older adults living in senior living communities may detect falls, changes in routine, missed medications, or other conditions that warrant attention. An alert acquires practical and ethical meaning only after a system determines what happened, who should know, and what that person must do. This qualitative study examined how residents, employees, and family caregivers at one continuing care retirement community understood those decisions in relation to an artificial intelligence-enabled robotic assistant. The broader study recruited 42 residents, 23 employees, and 5 family caregivers and consisted of 21 focus-group sessions. This focused synthesis examined existing code structures representing 10 resident and 9 employee group transcripts. An analytic memo derived from 2 family-caregiver groups provided preliminary sensitizing evidence. Across groups, monitoring via the robot assistant held positive salience when it addressed a recognizable care need, reached a legitimate recipient, followed a defined response pathway, and preserved resident control over access. It held negative salience when ordinary life events could be continuously cataloged or information might reach staff or family without clear permission and purpose. Stakeholders also distinguished routine information from safety exceptions and emphasized correct attribution, understandable alert reasons, and human follow-through. We integrate these findings into a multi-stakeholder alert-governance framework that connects detection to authorization, routing, response, and review. The framework positions alert governance as a sociotechnical design requirement and a practical mechanism of accountability.

ABSTRACT. This paper presents a design inquiry into human–robot–environment interaction (HREI) through speculative co-design workshops that centered plants and ecological relations. Building on a prior empirical study, the focus here shifts to the methodological apparatus: a bespoke toolkit of more-than-human card decks, creative writing prompts, and soft prototyping materials, structured by a participant booklet. The workshops were analyzed using four units of analysis, yielding nine themes clustered into three meta-themes: Self–Other Entanglements, Knowledge as Instrumental Mediation, and Reconfiguring Relations. From these, the paper proposes three relational dimensions of HREI and design principles. Together, these contributions show how design inquiry and methods can support ecological responsibility, and offer an early set of design principles for designing robots as mediators of multispecies and place-based relations.

ABSTRACT. With the rapid rise of ageing societies, the global elderly population is steadily increasing. Consequently, there is a growing demand to provide elderly people with companionship and assistance in their daily lives. One potential solution to this issue is the use of Socially Assistive Robots (SARs), which support users through social interaction. However, they typically feature limited interaction, and are regarded as a tool rather than a partner. To address this, our aim is to make SARs feel more partner-like by developing a module that enables them to engage in card-playing activities with users. In this paper, we present an overview of the module system that operates as a player in the koi-koi game, a traditional Japanese card game played with hanafuda playing cards. Additionally, we also present some potential benefit of the module as well as ethical and social concerns surrounding SARs in elderly cares and how the inclusion of card playing ability could affect them.

ABSTRACT. Rising rates of stress and anxiety, combined with the limited personalization of meditation applications and the high cost of human instruction, motivate the development of social robots that can guide mindfulness practice. This paper presents the system design, an experimental validation plan, and an ethical framework for an autonomous robotic instructor for guided mindfulness meditation. The system pairs a three-layer sensing-AI-output architecture with an explicit treatment of the robot’s face and body design, and introduces a rotating, mandala-based visual focus interface developed through three iterations of a double diamond process. We further describe a planned multimodal validation protocol–combining electroencephalography, respiration, and posture sensing–across three meditation traditions, and conclude with a discussion of the ethical, legal, and social implications of embedding biometric sensing and cultural practice.

ABSTRACT. Research and exploratory use of social robots in ASD therapy have been active recently. We introduced a minimally structured robot with limited expressive functions into a therapeutic daycare center for preschoolers with ASD and conducted longitudinal observations of child-robot interactions. The predictable interactions with the robot served as a reassuring "sanctuary," successfully eliciting spontaneous and rich social engagement from the children. However, our findings also suggested a critical dilemma: being in this comfortable zone carries the risk of inducing over-dependence on the robot and may prevent the generalization of social skills to human-to-human interactions. Currently, we are using generative AI to learn from teleoperation logs, extract the implicit therapeutic skills of expert therapists, and achieve robot autonomy. Crucially, this process has highlighted an ethical threat—that naive, full automation may unintentionally reinforce children's dependency on the sanctuary and alienate human caregivers from the caring ecosystem. To deal with this issue, we position the robot as a "transitional object" toward human-to-human interactions and propose the principles of "Relational Handoff Design." By systematically introducing uncertainty in the social context, this scheme invites the child into triadic interactions involving human caregivers and peer children. This prevents the child from being siloed in a closed dyadic relationship, fosters a shared understanding of the child among caregivers, and facilitates a smooth transfer of relationships to human others—paving the way for ethical and responsible use of robots in the therapeutic fields.

ABSTRACT. The rise of eating alone issue has led to social isolation and loneliness, often resulting in health issues such as decreased appetite and malnutrition. Inspired by this problem, our study aims to explore the applications of social robots as co-eating companions, which create social connections through informational support and emotional support. In this study, we developed a robotic co-eating system with eating companion conversation module and eating behavior recognition module. The system assisted in introducing food information, chatting, regulating eating pace, and encouraging meal completion. We conducted pilot study with twelve adults, assigned them to eating with robot twice and investigated their dining experiences. Current results demonstrate the ability of social robots to socially facilitate eating: all participants completed their meals and reported markedly enhanced perceptions of meal flavor and richness. Furthermore, participants exhibited positive attitudes toward the co-eating robot and the whole experience. Overall, this study highlights the potential of integrating co-eating conversation and recognition modules into robots as an eating companion, laying a foundation for future applications of social robots in co-eating scenarios.

ABSTRACT. The rapid development and deployment of socially embedded robots create ethical, social, and environmental challenges that existing governance mechanisms often struggle to address. Existing responses to this issue, such as anticipatory governance, remain limited by the pace of technological change and by the difficulty of identifying harms before technologies become established. This article therefore turns to anticipatory ethics within the design process. We argue that conventional design thinking, with its focus on immediate users and short-term outcomes, does not adequately account for the wider social and ecological systems affected by robotic technologies. Drawing on life-centred design, more-than-human design, and transition design, we develop Deep Design Thinking (DDT). DDT reworks design thinking’s five stages by expanding what design accounts for along two dimensions: who is affected, and over what timescale. We show how DDT applies at each stage using the case of sidewalk delivery robots.

ABSTRACT. Social robots shape the behaviour of the people they interact with. These interactions carry hazards, above all the erosion of user autonomy. That hazard is usually stated only qualitatively, leaving development teams with no threshold to test a design against. In this paper, we make the autonomy hazard quantitative by defining an operational measure of a user's residual self-determination within a robot-user dyad as conditional mutual information computed directly from interaction traces. This measure is not sufficient on its own. A robot that perfectly mirrors its user also drives the measure to zero, which makes an over-accommodating system indistinguishable from a dictating one. Pairing it with a transfer entropy from the robot's actions to the user's behaviour resolves the ambiguity. It yields a planar diagnostic in which the dependency hazard occupies a single identifiable quadrant. Designs agreeing to one percent on self-determination can differ on influence. For design practice, the pair turns a vague hazard into an acceptance criterion a prototype could fail. We simulate a medication-adherence robot and show that sub-maximal proactivity retains substantially more measured self-determination at a modest cost in adherence. Unexpectedly, deeper retained history raises self-determination rather than lowering it, because averaging over a longer window dilutes the robot's coupling to the user. The cost of personalization therefore depends on how the robot uses retained history, not on how much it keeps.

ABSTRACT. Research on socially aware navigation becomes much popular these days thanks to the growing trend of robot emergence in human society. It prioritizes human safety and comfort when selecting movement paths, in contrast to their industrial counterparts. Meanwhile, medical service robots have attracted growing attention for their ability to perform repetitive tasks autonomously, reducing direct human contact and easing healthcare workload. As hospital environments involve pedestrians with distinct roles—patients, staff, and visitors—whose social interaction requirements vary according to their context, we thus propose a role-oriented dynamic socially-aware navigation system that classifies pedestrian roles, detects pedestrian roles, and dynamically selects social rules according to pedestrians and interaction context. Simulations and also real-world experiments conducted in the NYCU intelligent ward demonstrate that the system can reliably detect pedestrians, generate appropriate social behaviors, and accomplish navigation tasks.

ABSTRACT. Robodolphins are increasingly proposed as replacements for captive dolphins, but recent work warns that they risk objectifying dolphins and alienating humans from living animals. Existing responses address this by classifying the purposes for which robodolphins may be deployed, permitting education and research while discouraging public display and entertainment. This paper argues that such classifications are unstable, because purposes cannot be clearly separated, because they treat "human" as a single undifferentiated party, and because they leave the institution of the zoo itself unexamined. Drawing on Donna Haraway's concept of companion species, the paper proposes replacing the question of purpose with a different one: whether response is passing among humans, robodolphins, and dolphins. Because this question can be answered only within concrete relationships, it distributes the problem into tasks that robotics, education, and animal care can each take up from their own position, and it aligns with the iterative orientation of design thinking. The paper illustrates this shift through an educational program and through cases in which the form of the zoo is itself being reworked.

ABSTRACT. The rapid advancement of artificial intelligence (AI) is transforming nearly every aspect of human life. While AI offers unprecedented opportunities to enhance human well-being and spiritual development, it also presents significant risks if its development is driven solely by technological capability and material interests. Therefore, beyond incorporating fundamental principles of ethical AI design, designers should explore ways of embedding positive spiritual values into everyday products so that design can assume a broader social responsibility in fostering compassion, altruism, and human flourishing. Given the widespread influence of Buddhism in Taiwan, this study applies the Design Thinking approach to develop an interactive tabletop game that integrates Buddhist-inspired benevolent design with interactive technologies. Drawing upon the Kṣitigarbha Bodhisattva Pūrvapraṇidhāna Sūtra, the game invites players to undertake a symbolic tabletop journey through the Three Evil Realms: the Hell Realm, the Hungry Ghost Realm, and the Animal Realm, where they experience Kṣitigarbha Bodhisattva's compassion for relieving the suffering of sentient beings. Through collaborative gameplay, participants are encouraged to cultivate compassion, altruism, and a sense of fulfillment derived from helping others. Although such experiences may not immediately transform users' values or behaviors, repeated engagement with the game may gradually nurture benevolent intentions and ethical reflection, allowing positive values to accumulate over time. The findings of this study suggest that integrating spiritual values into interactive product design represents a promising direction for compassion-oriented AI and product design, contributing to the advancement of humanity's spiritual civilization.