Floating Companion: Exploring Design Space for Soft Floating Robots in Indoor Environments | Proceedings of the 2026 Designing Interactive Systems Conference

2.1 Soft Floating Robots: Embodiments and Actuation

Our work builds on the growing interest in materiality within HCI. Researchers have explored soft interfaces [14, 83, 95] to support physical and emotional interactions. In parallel, a substantial body of work has investigated flying robots designed to interact with people in human-drone interaction (HDI) [9, 46, 53, 61, 77, 93, 106], conceptualizing them as flying companions that could enter everyday life [38, 40, 58]. However, most of this research has focused on rigid, propeller-driven drones. In contrast, soft floating robots (SFRs) extend this line of inquiry into mid-air by combining softness and buoyancy, thereby liberating soft interfaces from ground-based constraints and opening new possibilities for aerial interaction.

Figure 2:

Our research focuses on soft floating robots. For this study, we define a soft floating robot (SFR) as an untethered, autonomous, or semi-autonomous agent that can typically hover in the air without active power. An SFR satisfies three core conditions:

This definition explicitly distinguishes SFRs from heavier-than-air soft drones (e.g., [44, 107]). Unlike the drones that rely on high-speed propellers or flapping wings for lift—introducing noise and safety hazards—SFRs rely on buoyancy, enabling close-proximity indoor interaction. Within this definition, the emerging SFR domain exhibits diverse actuation methods (Figure 2). We classify propulsion into two high-level categories: Bio-Inspired Propulsion and Vectored Thrust Propulsion. Bio-inspired Propulsion includes flapping wings [30, 31, 69, 81, 112], oscillating tail fins [25, 64], and jellyfish-like peristaltic motion [32]. These robots mimic animal locomotion, producing organic, life-like movements that appear to “swim” through the air with gentle, rhythmic trajectories and low noise. Vectored Thrust Propulsion relies on engineering-based thrust generation via propeller-driven or propeller-free methods. Propeller-driven type generally offers stronger maneuverability, speed, and payload capacity with mature control models, though their motion often appears more mechanical [60]. A notable Propeller-free SFR is the ultrasonic micro-blower system in [113], which uses vibrating piezoelectric elements to create jet flows. This enables quiet propulsion and safe, close-proximity interaction [91, 113].

Despite this hardware diversity, SFRs share fundamental material properties that separate them from other platforms. Table 1 summarizes representative trade-offs between SFRs and multirotor drones in indoor human interaction. While flapping-wing drones are also a relevant aerial morphology for comparison, prior HDI research has focused predominantly on multirotor platforms, with physical interaction studies on flapping-wing drones remaining limited. We therefore use multirotor drones as the primary comparison baseline here. SFRs uniquely bridge the gap between mobility and safety by combining 3D flight with “safety through compliance.” Yet, physical advantages alone do not guarantee meaningful interaction. To address these gaps, our study moves beyond single-prototype evaluation to synthesize a structured design space that supports the intentional design of interactions with SFRs.

2.2 Technical Capabilities for Human-Robot Interaction

SFRs have demonstrated a growing range of sensing and interaction capabilities. The GT-MAB (Georgia Tech Miniature Autonomous Blimp), equipped with a monocular camera, has achieved face detection and following [96, 114], hand-gesture recognition [88, 115], and interaction through pointing motions [48]. Approaches utilizing Ultra-Wideband (UWB) localization have further enabled user following and mid-air gesture recognition, even supporting dynamic theatrical performances [66]. These advances lay the groundwork for responsive and socially aware floating companions. Given the absence of GPS indoors, researchers have explored a variety of localization and navigation solutions. SFRs can navigate autonomously as single robots [23, 42, 75] or operate in coordinated groups. eMotionSpheres [33] demonstrated collision-free swarm flight using infrared tracking, while installations such as FLOCK OF [13] and Aerobes [54, 55] showed that SFR collectives can move safely around visitors and create immersive spatial experiences.

Semi-autonomous SFRs have also expanded possibilities for remote presence. The Floating Avatar transformed a blimp into a telepresence platform using projected visuals and audio [99], while flapping-wing robotic avatars enabled embodied remote control through a body-tracking system [111]. Other work has introduced autonomous docking for teleoperated airships, enabling continuous operation via automated landing and charging [79].

Together, these studies highlight diverse technical capabilities—from perception to mobility to telepresence—that begin to support the vision of intelligent floating companions. However, most techniques have been developed and tested in isolation. We still lack an understanding of how these capabilities can be integrated and orchestrated to create coherent user experiences across varied contexts. From an HCI perspective, realizing the everyday potential of SFRs requires not only technical progress but also a systematic investigation of their interaction design. Establishing a structured design space for SFRs is therefore a critical first step.

2.3 Experiential Dimensions: Safety, Proximity, and Emotion

Robots increasingly engage with humans in everyday contexts, taking on social roles [21, 26, 49] and offering emotional support [29, 56, 90]. Recent HCI and HRI research emphasizes the need for safe, physically accessible interactions [1, 22, 28, 82, 89]. Across several empirical studies—each centered on a specific SFR prototype—a shared picture has emerged: SFRs hold unique potential for shaping new experiential dimensions in physical and social interaction.

A foundational insight from prior work is the perception of safety and approachability. BalloonBot [105] demonstrated that users view balloon-shaped robots as safe and socially acceptable, forming a positive basis for interaction. Cuddle-Fish [112] extended this by exploring close-proximity interaction; participants not only felt safe around a flapping-wing SFR but also initiated spontaneous touch and affective behaviors. These findings suggest that SFRs can support a progression from safe coexistence to emotionally meaningful engagement. Further work shows that SFR forms and motions—especially flapping wings—can serve as expressive social gestures such as hugging, shoulder patting, or high-fives [110, 111]. Additional studies exploring hand-gesture interaction with GT-MAB [88, 115] and user perceptions of autonomous floating interfaces like Diri [70] further enrich the interaction design landscape for SFRs. Beyond one-to-one interaction, SFRs have also enabled immersive spatial experiences. FLOCK OF [13] presents a school of fish-like robots that “swim” through the air, allowing visitors to walk among—evoking a robotic aquarium. The Aerobes [54, 55] demonstrate long-term autonomous operation in public spaces, with each agent expressing its own behavioral tendencies and responding to crowds via thermal sensing, illustrating the potential for context-aware group behavior.

3.1 Participants

We recruited a diverse group of 12 experts (5 female, 7 male) specializing in HCI, HRI, Design, and Robotics. We engaged these individuals not merely as study participants, but as co-speculators. This ensured a holistic perspective covering both user experience (HCI and Design) and technical feasibility (Robotics). Their experience ranged from 2 to 40 years (M = 10.42, SD = 11.28), and their backgrounds spanned from academic faculty to industry professionals. In addition to professional diversity, experts were based across seven residence locations (Table 2).

3.2 Procedure

We conducted individual semi-structured online sessions. The process was divided into three phases: (a) Briefing: experts were introduced to the study goals and provided consent; (b) Provocation: to ground the speculation, experts viewed video probes of existing SFRs to stimulate creative thinking, employing methods from [92]; (c) Co-Speculation: experts discussed their daily routines and co-created scenarios describing how SFRs could inhabit and support their everyday life. We probed for specific details regarding the robot’s behavior, spatial positioning, and communication modalities. Sessions lasted approximately one hour and were recorded for analysis. The protocol was approved by the university’s ethical review committee.

Figure 3:

3.3 Data Analysis

All interviews were transcribed and checked against the recordings for accuracy. We conducted a thematic analysis [15] to develop the design space and application categories. To establish an initial coding structure, the first and second authors independently open-coded the same subset of transcripts (2 of 12 interviews) at the level of meaning units relevant to the research questions. They then compared their codes, discussed overlaps and discrepancies with reference to the original excerpts, and consolidated them into a preliminary codebook. This preliminary codebook was organized into two broad analytic clusters: codes related to design-space dimensions and codes related to application use cases. Using this preliminary codebook, the same two researchers independently re-coded the shared subset, yielding an inter-rater reliability of Cohen’s kappa = 0.72 and percentage agreement = 85.2% [62]. Disagreements were resolved by revisiting the original transcript segments and refining code definitions. The refined codebook was then used to code the remaining transcripts and was iteratively revised through discussions among the research team. Lower-level codes were regrouped into candidate higher-level themes, which were merged, split, renamed, and clarified until a stable thematic structure was reached. This process resulted in a final codebook of 49 codes⁶, which supported the development of the ten design dimensions and three primary categories of application use cases. We do not claim exhaustive saturation of all possible SFR interactions or use cases; rather, we reached sufficient thematic coverage at the level of the final framework, with no substantially new top-level design dimensions emerging in the final three interviews. All procedures adhered to data protection and privacy requirements and the ethical research policy of the local institutional ethics review board.

4.1 D1: Spatial Mobility Scope

SFRs can operate across multiple spatial scales, from proximal interactions to room-scale and even multi-floor environments, enabled by their buoyant movement (Figure 4). Their low inertia with soft envelopes allows safe interaction through gentle contact with people and surroundings, reducing spatial constraints and expanding the robot’s usable movement space. They can hover quietly near users without generating downwash or noise, and occupy mid-air space to move over obstacles and navigate more freely. As P3 noted, this vertical mobility enables SFRs to inhabit unused “dead space” in domestic environments: “It can float over living-room clutter or drift up the stairs, occupying air that ground robots can’t reach.” This capability allows SFRs to transition across floors via staircases or open atriums, extending their range of operation beyond a single room [27, 84]. This dimension guides designers in selecting appropriate spatial ranges for interaction, helping them align the robot’s mobility with intended use scenarios.

Figure 4:

4.2 D2: Vertical Floating Mobility

This dimension captures an SFR’s ability to modulate its altitude to position itself relative to the user’s body (Figure 5). SFRs can align with users’ postures across vertical zones, enabling flexible spatial interaction. At the ground level, their soft bodies enable safe, animal-like behaviors such as gently nudging a user’s legs [97], supporting natural interaction with children and pets. At body and eye levels, SFRs position themselves for face-to-face interaction. As P9 noted, “It feels respectful […] if I sit down, it lowers itself to meet my gaze.” The overhead zone, typically inaccessible to ground robots and unsuitable for rigid drones, becomes a usable interaction space, where SFRs can hover without occupying floor area. This dimension focuses on where interactions occur relative to the user’s body. It helps designers use height as a design parameter to position interactions across vertical zones, supporting different forms of engagement.

Figure 5:

4.3 D3: Spherical Proxemics

This dimension extends Hall’s proxemics [43] from a 2D planar model to a 3D volumetric one (Figure 6). While D2 focuses on physical positioning in vertical space, D3 addresses how distance is perceived and interpreted in terms of social and emotional proximity. SFRs move across both vertical and horizontal axes, forming a spherical proxemic field that allows them to enter, inhabit, or exit personal space from multiple directions, including overhead. They can hover quietly within peri-personal space and also move into more distant zones to provide ambient awareness while remaining non-intrusive. As P4 described: “It’s like a pet bird landing on your head or shoulder, soft and not noisy, but a drone would feel dangerous.” Rather than treating proxemic zones as fixed categories, we conceptualize them as a continuous 3D spectrum [103]. This dimension helps designers use distance in three dimensions to regulate interaction intensity, shifting from intimate, close-range engagement to more peripheral interaction.

Figure 6:

4.4 D4: Interaction Configuration

This dimension describes the ratio of robots to users, ranging from one-to-one interaction to multi-user and multi-robot configurations (Figure 7). SFRs can flexibly shift between configurations due to their soft structure and low inertia, enabling safe operation in close proximity and higher-density clustering in mid-air. This allows collisions to be experienced as safe interactions rather than failures, supporting new forms of collective engagement. As P7 envisioned: “We can create an air aquarium […] the robots bump and drift like a swarm of jellyfish, and we can walk through the swarm, feeling the touch.” In multi-user settings, this enables shared physical interaction, where users can pass, guide, or play with SFRs. This dimension helps designers choose appropriate interaction configurations by balancing the number of robots and users, enabling both individual interaction and collective experiences.

Figure 7:

4.5 D5: Interaction Targets

This dimension categorizes the entities with which an SFR can interact (Figure 8), including users, the environment, other robots or devices, and the robot itself. Due to their softness and low downwash, SFRs have the potential to engage with a wider range of interaction targets. For example, they may move through plant canopies or hover near pets while generating less airflow disturbance than rotor-based drones. Their soft bodies may also tolerate gentle contact with walls or furniture, enabling designers to consider contact-based interaction rather than treating all collisions as failures.

SFRs can further collaborate with ground robots through spatial complementarity: while ground units operate at the floor level, SFRs provide overhead access, enabling coordinated multi-layer coverage of indoor environments. Experts also identified a unique target category—the self. SFRs can remain in a free-floating idle state, where subtle drifting motion introduces variability in behavior. As P2 noted: “When a vacuum robot docks, it becomes dead weight. But the floating robot drifts and rotates on its own, as if engaging in self-play, it keeps a pulse of life even when it’s doing nothing.” This dimension helps designers determine what the SFR interacts with, supporting the design of interactions with people, environments, other systems, and the robot’s own behavior.

Figure 8:

4.6 D6: Relationship Roles

This dimension describes the social roles an SFR can take in relation to the user (Figure 9). The softness, buoyancy, and gentle motion of SFRs enable a range of relational roles that are difficult to achieve with rigid drones. SFRs can act as guides, leading users from overhead or side-by-side positions while maintaining a consistent position relative to the user, even when moving across levels such as staircases. Their soft, floating presence evokes cultural archetypes such as guardian angels or sprites, making them well-suited for caregiving roles. Their ability to adjust height supports inclusive interaction with users of varied physical conditions, from crawling infants to wheelchair or bedridden users. As assistants, SFRs can operate from an overhead vantage point, providing support without occupying floor or table space. SFRs can also take on more affective roles. For example, their soft bodies allow safe physical contact, enabling forms of touch-based interaction. In telepresence scenarios, SFRs can act as embodied proxies, allowing remote users to interact physically through movement and positioning.

Finally, central to the care-receiver role, the SFR’s inherent clumsiness and balloon-like fragility align with the “weak robot” concept [72] and design approaches that frame vulnerability as a mechanism to elicit care [6]. As P10 noted, “When it gets stuck in a draft, it looks like a clumsy puppy asking for help.” This vulnerability is not a functional failure but a relational resource, fostering empathy and attachment through acts of care. This dimension helps designers map the robot’s physical properties (e.g., height adaptability, softness, limited control) to social roles, supporting interaction design across guidance, assistance, and care-oriented scenarios.

Figure 9:

4.7 D7: Physical Form

Figure 10:

This dimension focuses on the SFR’s physical form, including morphology, materiality, and scale (Figure 10). Unlike rigid robots with fixed shells, SFRs use inflatable envelopes that offer extensive morphological plasticity. Designers can shape them into zoomorphic, geometric, or abstract forms, each conveying different emotional tones. As P7 noted: “You can shape it into a cloud, it can give people a calm, relaxed feeling just by floating there.” Such visual language shapes first impressions and influences perceived animacy and approachability. Materiality is also crucial. Because SFRs enter intimate spatial zones (D3) and invite physical touch, the tactile quality of their surfaces matters. Options range from technical balloon films (e.g., nylon–aluminum film, TPU) to soft, fabric-like coverings (e.g., cotton) that enhance warmth and reduce perceived threat. Scale further expands design possibilities. Inflatable bodies can range from small personal robots (e.g., GT-MAB [115]) to large, architectural-scale installations (e.g., Festo Air_ray [31]). However, designers must balance scale with payload-volume constraints (Sec. 6.1) to ensure flight feasibility. This dimension encourages designers to leverage the SFR’s customizability, using soft forms, tactile materials, and appropriate scale, to align physical embodiment with desired social character and interaction style.

4.8 D8: Communication Modalities

This dimension describes the channels through which an SFR communicates (Figure 11). The softness and floating motion of SFRs enable a unique set of multimodal strategies. Visual communication extends beyond screens or LEDs to include kinetic cues. Bio-inspired SFRs can use rhythmic appendage motions, such as flapping wings or undulating fins, and expressive 3D trajectories to signal intention or emotion. Auditory communication benefits from the SFR’s quiet operation. Unlike drones with disruptive rotor noise, SFRs produce only gentle motor hums that can serve as soft localization cues without being disruptive. They also generate subtle environmental sounds, such as lightly brushing against a wall, to request attention. Haptic communication is enabled by the robot’s deformable, helium-filled body. Users can squeeze or tap the robot as a form of input, while the robot can initiate gentle physical contact. As P3 imagined: “It could remind me not with an annoying beep, but by lightly bumping me, like a pet nudging you.” This dimension helps designers to choreograph multimodal communication by combining visual, auditory, and haptic cues, leveraging soft contact, quiet motion, and subtle environmental interaction.

Figure 11:

4.9 D9: Floating Motion Dynamics

This dimension describes the kinematic qualities of an SFR’s movement (Figure 12). SFR motion is shaped by passive buoyancy, resulting in dynamics that are smooth, inherently slow, and partially indeterminate. Rather than maintaining rigid stillness, SFRs naturally float and often exhibit small drifts influenced by air currents. As P5 observed: “It doesn’t just hang there like a drone, the way it drifts around feels like a living thing breathing.” SFRs tend to follow curvilinear, continuous paths rather than sharp, linear drone-like trajectories. Because SFRs are light and move through air slowly, they accelerate gradually and avoid abrupt movements. This results in smooth, gentle motion. Air currents can also cause small, natural variations, such as slight swaying or drifting.

This dimension encourages designers to embrace, rather than correct, the inherent imprecision, slowness, and environmental drift of SFR motion. Leveraging smooth trajectories and slow velocity profiles can shape a relaxed interaction tempo while maintaining a non-threatening presence.

Figure 12:

4.10 D10: Level of Proactiveness

This dimension describes how the SFR behaves, ranging from passively responding to external forces to actively initiating interaction. (Figure 13). In a passive mode, the SFR is driven by external forces. Users can push, hug, or toss it safely. In a reactive mode, it adjusts its height (D2) or proximity (D3) in response to user actions or environmental changes, for example, lowering itself when a user sits to stay at eye level. In a proactive mode, the SFR initiates interactions based on internal triggers. Unlike rigid robots that depend on audio-visual alerts, SFRs can express proactivity through gentle physical cues. As P4 described: “It doesn’t just beep from a distance, it floats over and gently bumps my shoulder to say ‘hey’.” This is enabled by its soft, low-inertia structure. This dimension helps designers calibrate the SFR’s level of agency across passive, reactive, and proactive behaviors, using physical actions to initiate interaction, shaping how the robot engages with users and is perceived by them.

Figure 13:

4.11 Linkages in the Design Space

The design dimensions presented here serve as foundational elements for developing SFR interactions. Together, they provide a starting point for translating the distinctive qualities of SFRs, such as buoyancy, softness, lightness, and aerial mobility, into design considerations. In this section, we clarify how these dimensions relate to one another while preserving their distinct roles.

5.1 Emotional Support and Companionship

Expert participants frequently highlighted the potential of soft floating robots (SFRs) to provide emotional support and companionship, especially for alleviating loneliness. They described SFRs as envisioned to offer a gentle sense of presence—“a sense of a living being in the room” (P4)—for people working alone or family members at home. Participants envisioned SFRs floating nearby during daily routines, such as integrating with smart speakers to tell children bedtime stories. They also proposed semi-public applications: for example, in waiting areas where, as P8 noted, “the time would feel less boring because I could play with the robot” or in children’s spaces where SFRs could accompany users and reduce caregivers’ burden.

Participants frequently compared SFRs to pets, highlighting their perceived liveliness and capacity for pet-like responses. An SFR might greet its owners at the door or gently nuzzle their leg. As P7 remarked, “it could be like a cat, keeping the distance but still making you happy just by being there. Or like a dog, coming to you for touch and play.” P3 noted the practical advantages: “A robot pet wouldn’t shed hair, and you wouldn’t need to spend time walking and feeding it. It would be nice for people who want a pet but don’t have the time to take care of it, or who are allergic.”

5.2 Social Facilitation and Communication

Participants also envisioned SFRs as mediators of social connection. As telepresence robots, they could integrate with video calls to provide remote companionship with a tangible, tactile presence, or act as physical emojis that perform gestures synchronized with messages. As P3 suggested, “I could send my friends an emoji through the robot near them, like having it roll twice in the air.” Participants also noted that expressive flight patterns could convey emotion [20] and enrich remote or co-located communication.

In workplace settings, SFRs were compared to office dogs that lighten the atmosphere and strengthen collegial bonds. As P7 explained, “When a cute SFR enters the office as an office dog, it can trigger people to care for it. That makes colleagues see each other as kind, caring people.” At social gatherings, SFRs could function as icebreakers, helping spark conversation and reducing social anxiety.

5.3 Ambient Experience and Atmosphere

Compared to rigid robots such as drones, participants emphasized that SFRs are better-suited for shaping dynamic atmospheres. Bioinspired SFRs that “swim” through the air can evoke the calm of an underwater world, helping people relax and reduce anxiety. Participants also imagined SFRs serving as dynamic ambient artifacts: floating through the home in sync with lighting to improve the aesthetic appeal, or moving among guests to enliven a party. As P8 suggested, “In an office lounge, a cloud robot could change its color in response to the environment to enhance the atmosphere.”

Participants also envisioned applications in public spaces. In museums, shopping malls, or immersive theaters, SFRs could appear as ghosts, sprites, or alien-like NPCs that interact softly with visitors. On stage, they could serve as expressive flying props, moving between performers and audiences to enhance connection. In amusement parks or zoos, animal-shaped SFRs could invite tactile play for children. These applications were seen as especially compatible with large indoor environments; as P10 noted, “Theaters and shopping malls often have bigger spaces that could support larger robot sizes.”

5.4 Daily Living Assistance

Participants envisioned SFRs as everyday assistants that support routine tasks through subtle, non-intrusive interactions. Their softness enables gentle wake-up experiences, as P4 described: “The robot could float over and make gentle physical contact to wake me up, like a soft floating alarm clock.” They might also remind desk workers to take breaks, as P9 noted, “Objects flying in the air are more likely to catch people’s attention,” or intervene in gaming overuse by drifting in front of the screen. For meeting reminders, participants suggested soft nudges or wing taps, as P3 noted: “It could remind me by lightly bumping me, or tapping me with its wings.”

Leveraging their 3D mobility, SFRs could help locate misplaced items by hovering above them as visual markers. As P10 suggested, “At home, the SFR could team up with ground robots, the floating one searching from above, the ground one checking under the sofa.” Their reach also makes them useful for ceiling inspections, detecting leaks or cracks, or checking dusty shelves, with potential for light cleaning. Participants further envisioned SFRs as follow-me cameras that “can film users from different angles, distances, and heights” (P3). With near-neutral buoyancy, SFRs exert almost no weight when perched on the user’s body; P8 remarked, “I would place the floating robot on my lap and then put books, tablets, and snacks on it, it would make things more convenient.”

5.5 Public Space Services

Participants envisioned SFRs as mobile guides and information providers in public spaces. Their ability to float alongside users makes them well-suited as museum tour guides or airport assistants. As P2 described, “It could be like Paimon from Genshin Impact, a floating guide always by your side, ready to answer questions and explain things.” They could also serve as shopping assistants in malls or library helpers, aiding users in locating products or books on shelves at different heights. Participants further highlighted the potential of SFRs as flying advertisements, extending existing uses of indoor blimps. With customizable appearances and expressive movement, SFRs can attract attention in malls or restaurants and create interactive experiences beyond static displays. As P8 noted, “This kind of interactive advertisement can fly around the audience and let them touch, making it more engaging than common flat advertisements.”

6.1 Design Within Physical Feasibility Boundaries

While the design dimensions articulate the possibilities, their realization is governed by the specific materiality of SFRs’ helium-filled structures. As illustrated in our Physics-Design Dependency Framework (Figure 15), interaction design is not separate from engineering constraints but is entangled in a rigid Buoyancy-Mass Cycle. For lighter-than-air robots, payload capacity is directly determined by the volume of the helium-filled envelope. Consequently, design decisions such as form, scale, and the weight of on-board hardware are tightly coupled: changing one parameter cascades into constraints across the system. This framework aims to help designers navigate this material entanglement, moving beyond simple trade-offs to holistic design decisions.

Figure 15:

6.2 Design Principles

Building on the ten-dimensional design space and the feasibility strategies outlined above, we distill three core design principles. These principles guide designers and researchers in moving beyond traditional rigid-robot paradigms and fully leveraging the unique affordances of SFRs.

7.1 Applying Design Principle 1: Embracing Soft Physical Contact

Guided by Design Principle 1, these examples illustrate a shift from collision avoidance to purposeful physical contact. They demonstrate how the inherent softness of SFRs enables haptic interactions that would be unsafe or impractical with rigid drones.

7.2 Applying Design Principle 2: Utilizing Volumetric Proxemics

Following Design Principle 2, these examples illustrate how SFRs free interaction from the 2D ground plane, utilizing the z-axis to regulate attention and social presence.

7.3 Applying Design Principle 3: Evoking Animacy and Enchantment

Aligned with Design Principle 3, these examples illustrate how an SFR’s buoyant physics and organic motion can be leveraged to enrich the atmosphere of a space.

8.1 The Essence of Floating: From Physics to Experiential Qualities

Experts consistently emphasized that SFRs’ value lies in ambience creation and social companionship rather than utilitarian efficiency, echoing calls to challenge the assumption that robots must be designed primarily as flawless, productive agents [6]. We position SFRs not as replacements for precision-based rigid robots, but as a complementary morphology specifically suited for low-speed, high-intimacy, and ambient contexts where mechanical silence is paramount. This invites a deeper question: What makes “floating” afford these qualities in ways other platforms do not? We argue that the distinctiveness of SFRs arises from the entanglement of their governing physical laws—buoyancy, inertia, and aerodynamics. These forces produce soft, slow, and naturally drifting behaviors that reshape how users perceive the robot.

8.2 Vulnerability and Benevolent Chaos in Relational Dynamics

While SFRs introduce unique opportunities for interaction, they also embody inherent physical limitations: they are lightweight, susceptible to airflow, and require protection from sharp objects. Rather than treating these limitations as flaws to be hidden, experts suggested reframing them as relational resources that support emotional connection.

8.3 Situating the SFR Design Space within HDI

Situated within broader HDI research, our design space includes both concerns shared with other flying robots and interaction qualities that are especially salient for SFR. Dimensions such as interaction configuration (D4), interaction targets (D5), and levels of proactiveness (D10) are not unique to SFRs; rather, they are also present in prior HDI work on drone roles [37, 108], application contexts [46], and how people interpret flying robots in inhabited environments [9]. What distinguishes SFRs, however, is not the introduction of an entirely separate set of dimensions, but how softness, buoyancy, low inertia, and quietness reconfigure these concerns in close-proximity interaction. In conventional multirotor HDI, noise, exposed rotors, and safety constraints often limit proximity and physical contact [16, 106, 118]. By contrast, SFRs make gentle touch, overhead presence, compliant contact with the environment, and slow drifting motion more available as interaction resources rather than hazards. This is especially visible in D2, D3, D8, and D9, where altitude, proxemics, communication, and movement are shaped not only by flight capability, but also by the material conditions [10, 106]. Our findings also complement recent designerly HDI work that approaches flying robots as experiential, relational, and situated artifacts [19, 38, 39, 40]. Rather than proposing a single interaction concept or scenario, our contribution is to systematize how SFR embodiments reshape interaction design across roles, proxemics, communication, and motion.

8.4 Limitation and Future Work

This work is expert-informed rather than end-user validated. We adopted this approach because soft floating robots (SFRs), as an emerging form, remain technically immature, and early user studies risk conflating hardware instability with judgments of interaction quality. While this strategy enabled us to envision interactions beyond current technical constraints, it captures imagined possibilities rather than lived experience. Accordingly, the proposed design space and use cases should be interpreted as generative starting points rather than evidence of end-user desirability or long-term acceptance. Moreover, the subtle and emergent dynamics of long-term cohabitation—such as how users negotiate space with a floating agent over weeks or months—cannot be adequately captured in short-term sessions. Future research should therefore move beyond speculative exploration toward in-the-wild deployments, examining how SFRs are integrated, appropriated, and sustained in everyday indoor environments over extended periods. Such studies should also investigate diverse cultural contexts to determine whether the design space requires regional adaptations given varying attitudes toward robots, physical touch, and personal space.

Our demonstrations were instantiated on a single flapping-wing platform, which we position as a representative case study rather than a proxy for all SFRs. While this configuration prioritizes quietness and safety, other propulsion types (e.g., propellers or micro-blowers) would introduce different acoustic and kinematic profiles. We acknowledge that the current implementation represents one trajectory within the broader design space. Future work should develop and compare diverse hardware configurations to understand how different physical embodiments (e.g., sound, airflow, texture) influence the perception of animacy and comfort.

Finally, we acknowledge the inherent material challenges of helium-based flight. Issues such as helium leakage, resource scarcity, and the strict buoyancy-mass constraint remain practical barriers for ubiquitous adoption. We acknowledge the ethical tension of using helium, a finite resource. This design space invites us to think of SFRs not as disposable consumer gadgets, but as ephemeral artifacts. Their gradual deflation could be incorporated as part of an interactional lifecycle in appropriate contexts, while practical deployments should prioritize helium retention and reuse. Future engineering efforts should prioritize high-retention envelope materials, ultra-lightweight components, and sustainable solutions—such as gas recapturing [98] and recycling systems or safe and regulation-compliant alternative lift gases. We hope our design space provides the interactional motivation to drive these necessary hardware innovations.