1 Introduction

Soft robots are made of compliant materials with mechanical similarities to soft biological tissue, and their design is often anchored in biomimicry and bioinspiration [1, 2]. Biomimetics proceeds by imitating mechanisms and processes in nature to solve tasks, while bioinspired design seeks to abstract general principles from nature for technological purposes [3, 4]. While these dominating design approaches to soft robotics have succeeded in imbuing soft robots with impressive capabilities, they disregard the aesthetic dimension of robot design. This potentially could be detrimental to achieving practical use of soft robots in human-centered applications, as a robot’s aesthetic design demonstrably impacts human–robot interaction (HRI) on multiple levels [5,6,7]. Moreover, soft robots have a radically different embodiment and appearance than traditional robots, which is plausible to affect, e.g., attributions of sociality, emergent physical interaction, and perceptions of robot agency [8]. Prior work has proposed that soft robots are well-suited for assistive and care robots because they can engage in safe tactile physical interaction [9], are capable of more complex emotion expressions due to their soft materials [10], and, in addition, have a friendlier or more lifelike appearance than traditional robots [11]. The latter is also a highly desirable trait for personal robots that assist or provide entertainment for people in a domestic environment [12, 13]. Although the aesthetic aspects of soft robots have recently been explored in art, design, and architecture (e.g., [14,15,16,17,18,19,20]), no prior work has interrogated what constitutes appropriate aesthetic design criteria for soft personal robots specifically. Prior work has, however, recently called for further study of human perceptions and acceptance of soft robots to enable implementing soft robotics in socially assistive robots [10].

In this article, we introduce the concept of Soft Biomorphism to describe formal attributes of soft robotic objects (e.g., their visual appearance, texture, and movements). Soft biomorphism also denotes an alternative design paradigm for soft robotics centered on enhancing their inherent organic aesthetic qualities. In our work, we aim to inductively uncover appropriate contexts and applications for this design aesthetic empirically. I.e., contrary to most work within robotics, wherein the design is derived from the application of the robot, we aim to utilize this specific robot design aesthetic as a driver for discovering matching types of interactions and applications for soft robotics.

It has been hypothesized that humans are genetically attracted to life and lifelike things [21] and shown that contact with nature can improve human health and well-being [22,23,24,25]. Biologically inspired traits are also already used in several product designs, ranging from everything between cars to furniture, for both practical and aesthetic purposes [26,27,28,29]. Moreover, lifelike traits can improve the likability of robots [30], and organic robot appearances can help trigger social responses from humans in HRI [31, 32]. It can thus be argued that biomorphic robot designs are beneficial for robots aimed at social HRI. Considering these points, soft biomorphism can plausibly prove a desirable design aesthetic for personal robots that feature soft robotics technology.

The present work consists of a first exploratory phase focused on soft biomorphism and a follow-up study specifically addressing human well-being. Initially, our work did not set out to address well-being and HRI. Instead, its focus was to explore soft biomorphism as a design space and a design principle and its potential for soft personal robots more broadly. However, in our first exploratory study, we found close ties between the soft biomorphic design aesthetic and robot applications related to improving human well-being. These findings led us to further investigate soft biomorphism as a potential design aesthetic for embodied agents for well-being. Through design practice and two studies, the present work thus explores soft biomorphism as a design aesthetic, its potential benefits for soft personal robots, and its appropriateness for robots intended to support well-being. Specifically, the article addresses the following research questions:

RQ1

How can the organic and lifelike qualities of silicone-based soft robots be amplified through their aesthetic design?

RQ2

Which ideas about personal robot applications, interactions, and relations do soft robots with enhanced biomorphic features elicit from potential users?

RQ3

Are soft robots with enhanced biomorphic features considered more appealing and more appropriate for applications wherein soft robots are used to support well-being through physical interaction?

2 Related Work

We position this work within research on robot design and embodiment, personal robots, human interaction with soft robots, and soft robots for human well-being.

2.1 Embodiment in HRI

An indispensable property of robots in real-world environments is their physical embodiment [8]. The design of a robot’s body has proved to affect various facets of HRI, such as how users experience, e.g., the social presence of robots [33, 34], their levels of comfort and trust [35, 36], empathy [37], attractiveness [38], and enjoyment [35, 38]. Furthermore, utilizing soft materials in the design of social robot morphologies have been explored for intimate HRI [39], and it has been shown that the inclusion of soft materials increases perceived safety and comfort in close HRI [40]. While one of the most used design approaches for robots is anthropomorphism [41], anthropomorphic robot designs have also been criticized, e.g., because people might wrongfully attribute anthropomorphic robots the same social, intellectual, or emotional capacities as humans, which can lead to a decrease in satisfaction with HRI [42,43,44]. Moreover, it has been argued that designers lose design freedom and aesthetic control when robots are dominated by anthropomorphic or zoomorphic features, which can lead to suboptimal designs [43].

2.2 Personal robots

Following prior definitions of personal service robots [12, 13], we consider a personal robot a robot that operates in a domestic setting, either to help one or more people with practical tasks, provide entertainment, or keep people company. Studies have shown people to have a positive attitude towards living with robots [45,46,47], and previous research has investigated the effects of robot movement [48, 49], interaction modalities [50, 51], capabilities [52, 53], and robot appearance [54,55,56] in relation to personal robots. Due to the expected impact of personal robots in future societies, there has been a call for further research on how a robot’s physical design and its behavior impact human expectations towards it [13].

2.3 Soft robotics and HRI

Only a few studies have investigated various aesthetic designs of soft robots, and the capacities of soft robots for social and physical interaction with humans. Boer and Bewley [57] sought to encourage a broader understanding of possible social robot designs by approaching soft robotics with a focus on their performative qualities and “otherness”. Their designs gave rise to ideas of HRI of a playful, negotiable, and curious character. Zheng and Walker [58] investigated soft robotics artifacts focusing on affective HRI and found that their soft robotic artifacts attracted emotional investment due to their “biomorphic quality”, kinetic forms, and compliant materials. Budak et al. [59] created a soft robotic “breathing” wall with a biomorphic design aesthetic to explore the potential of human interaction with imitated organic life that challenge the distinctions between organic and synthetic, subject and object. Jørgensen et al. [60] questioned the presumption that soft robots are more “natural” than rigid robots, with findings indicating that both soft and rigid robots can evoke social responses and are ascribed social agency, even if not being designed with distinct zoomorphic or anthropomorphic attributes. Moreover, their findings showed that soft robots appear to invite touching and might be perceived as safer. Klausen et al. [61] and Farhadi et al. [62] explored the possibility of conveying emotions and communicative signals to a user by altering a soft robot’s breathing rate, movements and shape, discovering that participants perceived varying levels of arousal and pleasure at different breathing rates.

2.4 Robots and well-being

While numerous studies have explored robots’ abilities to support human health and well-being (for systematic reviews, see [63, 64]), only a few have focused more specifically on soft robots within this context. Asadi et al. [65] investigated whether people would synchronize their breathing rhythm with a soft robot by touching it while performing different tasks. Their results suggest that although participants who touched the robot did not synchronize with the robot’s movement rhythm, touching a “breathing” soft robot can lead to brain patterns indicative of an increased positive emotional valence. Haynes et al. [66] examined soft robotics as a haptic technology for lessening anxiety by developing a pneumatically actuated huggable interface simulating slow breathing. Through an anxiety-inducing mixed-design experiment, the researchers found that the haptic interface effectively reduced pre-test anxiety in a manner similar to guided meditation. In a qualitative study, Hall et al. [67] explored soft robotics technology in a healthcare setting by developing haptic soft robotic prototypes mimicking attributes of affective human touch intended to reduce procedural anxiety in radiotherapy. Feedback from patients, healthcare professionals, and carers suggest that flexible devices, which can be used on the body and provide pulsation sensations, could be used to reduce procedural anxiety and calm down patients in the given context. Sabinson and Green [68] sought to reduce stress and regulate emotional states in confined spaces through a therapeutic wall-mounted soft robotic surface that initiated different inflation patterns. Comparing their soft robotic prototype with a 2D graphic vector, the researchers found that the participants who followed the exercises led by the soft robotic prototype had significantly lower perceived stress levels, and that the prototype’s visual characteristics helped some participants focus and breathe deeper.

2.5 Summary

To summarize, prior work has addressed some of the effects of embodiment on HRI in general and various aspects of personal robot designs, soft robots and HRI, and more specifically, soft robots and well-being. However, research on soft personal robots and their aesthetic design has not yet been conducted. In general, more research is equally needed on the effects of embodiment on HRI and perceived agency with respect to soft robotics specifically, and how their aesthetic design might modulate these [60].

3 Soft Biomorphism

Biomorphism is used as a concept within art, design, and architecture to characterize organic, lifelike, abstract, or curvilinear forms [69] that evoke or refer to living organisms [70]. Similar to [71], we understand “organic” to refer to something that consists of or is derived from living matter. Biomorphic forms are thus “nature-centric” or “biologically-shaped” and relate to natural phenomena without being direct representations of them [72], a form of “nature abstracted into new nature” [73].

In the present work, we use Soft Biomorphism to characterize the aesthetics of a specific class of soft robot designs that emphasize visual, material, and kinetic resemblance with soft natural organisms. Thus, soft biomorphism encompasses both the object’s form as well as its tactility and movement. Drawing on prior descriptions of biomorphic forms [69, 72, 74,75,76], we take soft biomorphic forms to include curvilinear, sweeping, bulbous, convex, or concave, annular, irregular, rugged, arciform, or asymmetrical forms. The added “soft” in “soft biomorphism” refers to physical softness [2] and not symbolic or figurative softness (e.g., as in a “soft voice”), hence soft biomorphic designs also afford sensations similar to those of touching a living soft-bodied organism. Soft biomorphic movements are taken to be akin to the anatomical movements and dynamic movement patterns of living organisms and can be flowing, and smooth, as well as erratic, or gyrated. Soft biomorphic robot designs may vary in their degree of abstraction or realism with respect to their representations of soft natural organisms in a robot’s form, tactility, and movement. While audio can also be included as a factor in soft biomorphism (see, e.g., [14]), the exploration of this aspect lies beyond the scope of this work.

The rationale behind proposing soft biomorphism as a design principle is to enable the design of soft robots that appear lifelike yet unfamiliar to facilitate more open-ended and negotiable human–robot relations, that are not modelled on, for instance, the interactions with an animal or a pet. Prior work has argued that robot designs with forms and behaviors open to interpretation may offer increased opportunities for long-term relations between humans and robots because communication can be more varied and engaging over longer time [77]. An advantage of designing robots after animal models instead of choosing an anthropomorphic design is that a robot’s limited behavioral, cognitive, or perceptual abilities may be reflected clearer through its appearance [78], thereby supporting the establishment of more appropriate social expectations about the robot [79]. Basing designs on familiar animals, such as domestic pets, however, may also lead to people having expectations about a robot’s abilities that cannot be met, potentially resulting in termination of interaction with the robot [78]. It has also been argued that zoomorphic robots designed to be perceived as certain animals are unethical [80], and that robot designs should be based on unfamiliar animals to avoid misconceptions about robot abilities [81]. Thus, where anthropomorphic and zoomorphic robot designs may generate (misleading) expectations about a robot, the abstraction of nature in biomorphic designs can provide familiar cues for intuitive social interactions that may be explained by our inherited inclination towards nature and lifelike things [82]. By eschewing replication of a particular animal or organism, more design freedom is also obtained, while the impression of the robot as a responsive interaction partner can still be conveyed through visual, haptic, and kinetic similarities with natural organisms in general. We believe this can lead to more appropriate robot designs and more interaction possibilities compared to relying on zoomorphic or anthropomorphic designs (e.g., see [77]). In this work we thus explore soft biomorphism as a design principle as well as a design aesthetic and examine which uses and interaction possibilities this design aesthetic suggest to users. In that sense, our work reverses the renowned axiom “form follows function” [83], as we instead use aesthetic design as a starting point and a driver for discovering practical functionality and HRI modes. Hence, our approach acknowledges the inherent interdependence of form and function, aiming to explore the ways in which soft biomorphic appearance can influence and inspire function within HRI.

4 Prototypes Designs

The prototypes were developed through design practice with soft robotics and silicone materials. The focus in the aesthetic practice was not to design visually appealing designs but, instead, to make an uncritical investigation of the breadth of soft biomorphism as a design space. Consequently, some of the prototypes may be deemed odd, unappealing, or eccentric, and radically divergent from traditional robots. We chose to include designs that we ourselves found aesthetically pleasing or beautiful as well as those we found off-putting to allow for contrasts and more specific assessments of their various facets to emerge in discussions of them with potential users.

In two consecutive rounds of design experiments aimed at exploring RQ1, we first created soft surfaces and then static and dynamic three-dimensional objects. The first round of design experiments focused on ways of manipulating silicone rubber’s visual and haptic attributes through texturing and pigmenting. In the second round, both static as well as dynamic prototypes, which could move and change shape by being pneumatically actuated, were explored. An Arduino UNO microcontroller, with a custom motor shield driving three low-noise pumps (MITSUMI R-14 A213) and three solenoid valves (Uxcell Fa0520D 6V NC), was used to simultaneously control the dynamic prototypes. Nine out of a total of 15 soft biomorphic prototypes were selected for inclusion in the studies (see 5 Study 1 and 6 Study 2). The selection was made to capture the variety of the design space, and we additionally chose designs with open and negotiable qualities and identities. The selected prototypes thus feature different affordances [84,85,86] for interaction: flat, smooth surfaces, for instance, may invite touch and stroking, while rounded shapes can encourage holding the objects, and pointy or limb-like shapes can afford grasping. In future work on soft biomorphism, criteria or parameters for choosing representative soft designs could be chosen explicitly, to reduce subjectivity in the process. By systematically assessing and selecting prototypes based on parameters such as form, tactility, movement, or levels of abstraction versus realism, it would be possible to ensure a more representative sampling of the design space and a more transparent and reproducible selection process.

The selected prototypes (see Fig. 1) are presented individually in Table 1. The reader is highly encouraged to watch the supplementary video (https://youtu.be/col9OP5FcPw) to get a better impression of the prototypes.

Fig. 1
figure 1

The selected soft biomorphic prototypes. A human hand is included in the photo to give an impression of their sizes

Full size image
Table 1 The soft biomorphic prototypes selected for the first study and their attributes and design inspirations (The images of prototypes 1–3 are cropped and magnified to show the surface textures, and show squares of the following sizes: 1 displays 5 × 5 cm; 2 displays 10 × 10 cm; 3 displays 20 × 20 cm)
Full size table

5 Study 1

The purpose of the first study was to explore which ideas about personal robot uses, interactions, and relations that would emerge in people’s encounters with prototypes featuring biomorphic traits. Interactions are situations where two or more entities react to each other over a short time span, and relations instead denote how two or more entities behave towards each other, and may be of a longer duration [87, 88]. A design evaluation workshop was held in five renditions, each with two different participants and the first author as facilitator (see Fig. 2). As the study was conducted amidst a COVID-19 lockdown, measures were taken to reduce the spread of disease, and the five sessions were organized in small groups consisting of people who lived together. Inspired by common practice within participatory design and co-design, we sought to facilitate active partaking from participants through this format [89,90,91]. Video recordings of the sessions were made with a camera on a tripod.

5.1 Participants

The 10 included participants (female: n = 5, male: n = 5; age: range 26–53 years, M = 33.10, SD = 10.28) were a convenience sample and had professions in areas including engineering, pensions, publishing, and audiologopedics. 80% (n = 8) stated no familiarity with robots, and the rest (n = 2) that they knew only little about robots. All participants were Danish citizens living in Copenhagen, Denmark. Each workshop session featured one female and one male participant that were in a domestic partnership. It was a coincidence and not a deliberate choice that only heterosexual couples were recruited. The participants did not receive any compensation for their participation.

5.2 Procedure

Each workshop session had a duration of 54 to 66 min. After receiving information about the study and signing consent forms, the two participants were briefed that the research aimed at exploring how to design soft personal robots. They were informed that they would be shown different prototypes, which they were allowed to touch. The prototypes’ names were not disclosed to prevent biasing associations.

The first exercise was a warm-up exercise to initiate thinking about personal robots. The participants were asked to imagine that they had their own unique personal robot at home and asked to describe briefly what this robot could do, how they would use and interact with it, how it looked and what materials it was made from.

The second exercise (“show and tell”) sought to investigate if specific or recurrent visions of interactions and relations were triggered in the meeting with our soft biomorphic prototypes. After revealing the prototypes, the participants were asked to envision them as elements or whole bodies of personal robots, and that all prototypes were capable of movement. The first participant was asked to choose one of the soft biomorphic surfaces (Table 1, prototypes 1–3) and the second was asked to choose one of the static soft biomorphic objects (Table 1, prototypes 4–6). Both participants were then asked to give a detailed description of their chosen object, share their impression of it, and to consider what actions they imagined it to perform and in which contexts.

In the third exercise, the participants were given 10 min. to collaboratively come up with a future scenario that featured a personal robot and took place in a domestic setting. The exercise was inspired by fictional inquiry [92], a design method wherein roleplay is used to interrogate people’s visions of desirable futures. It also drew on the co-constructing stories technique [93], in which storytelling, focused on past experiences and anticipated future experiences, is used for formative design concept feedback and to reveal potential users’ attitudes and visions. The participants were asked to include one of the three dynamic soft biomorphic prototypes (Table 1, prototypes 7–9) in the story and to envision that it took place in the future, and that the story’s characters were very familiar with the robot.

In the fourth and final exercise, the two participants were asked to revisit their responses from the first exercise and reflect on if their encounters with the soft biomorphic prototypes had influenced their ideas of an ideal personal robot, as they might not have been familiar with this technology when formulating these, and if the soft prototypes might be integrated into their ideal personal robots (see Fig. 2).

Fig. 2
figure 2

Selected still images from the workshop

Full size image

5.3 Thematic Analysis

To analyze the collected video data from the workshop, we transcribed the video recordings (speech and actions performed) and conducted a thematic analysis using an inductive coding scheme [94, 95]. Two of the authors read through the transcription and established preliminary codes separately, which they shared with each other and mutually agreed on condensing into 9 applicable codes. They then individually coded 20% of the data and the inter-rater reliability was calculated and found to be high (Cohen’s κ = 0.814). Hence, the first author proceeded to code the remaining data and subsequently developed three themes in dialogue with the corresponding author.

5.4 Results

The following three overarching themes were found to capture the central ideas expressed about personal robot applications, interactions, and relations that the prototypes elicited in the workshop sessions.

5.4.1 Organic Qualities—Both Appealing and Uncanny

Most participants (n = 7) remarked that the prototypes’ forms, surface textures, colors, or movements had organic or natural qualities, when they gave descriptions of the prototypes in the second exercise. Some attributed these to the irregularities and variations in surface texture or color, to others, it was the softness that made the prototypes appear “natural” and “warm”. Half of the participants (n = 5) equally mentioned resemblances to specific natural organisms or body parts including mammalian organs, limbs, insects, aquatic plants, and cephalopods. These organic qualities were variously, sometimes even interchangeably, described as being desirable and uncanny in the proposed and imagined interactions and relations with the prototypes. Some participants (n = 3) felt that these qualities could contribute a sense of emotional comfort in the interaction with a personal robot and enhance the social and emotional connection with it, e.g.:

“It [the robot] must be organic, so it’s like that... In the tasks where it might have to interact with children or have to interact with me in, one way or another, I want it to be something that appears friendly, and that I feel is like ... a mammal in some way. Or, you know, something that is relatable” (P1.10)

Other participants (n = 4) found the organic qualities of the prototypes’ forms, textures, or colors, to be somewhat unappealing or uncanny because they made the prototypes appear very “natural-appearing” (P1.2) or like “alien” versions of organic things. Several (n = 4) in addition associated the prototypes’ organic aesthetics with notions of them being “uncontrollable” or “autonomous”, capable of making their own decisions and having some degree of sentience or agency. Interestingly, some participants (n = 4) also remarked on that the prototypes had both organic qualities and traits from fabrication that revealed they were man-made. These participants appreciated the organic qualities but also found it positive that the prototypes had cues that made them appear “constructed”, “man-made”, or “clinical”, as it made it possible to separate the prototypes from living beings. Elaborating on this, one participant explained he saw potential in

“combin[ing] the ambition to be, like, true to nature with the aesthetic dimension you can add when constructing something. (…) [T]o make the lifelike less creepy and more beautiful in such a completely basic appealing way” (P1.1)

5.4.2 The Sensation of Touching the Prototypes

Nearly all participants (n = 8) explicitly assessed the experience of touching the prototypes during the second exercise. Some found the prototypes very pleasant to touch due to their compliance, smooth surfaces, or “inviting” visual appearances (P1.4), which made participants express ideas of engaging in close physical contact with the prototypes in the third exercise. For example, a participant imagined touch as a pleasing way to interact with a soft personal robot for entertainment purposes:

Then you go over [to the robot] (*P1.5 makes a calm movement with his hand, gently and repeatedly pats Purple Tentacle as if it is a small animal*). It would be very pleasant if it was like that anyway.... You know, not like tapping an app [on a phone]. But it would be very nice if you woke it [the robot] up just like that (*P1.5 caresses Purple Tentacle repeatedly*). (P1.5)

Others described in the second exercise that they found the sensation of touching the soft materials too similar to the sensation of touching something organic which was not appealing to them in a context of interacting with technological artifacts. A number of participants (n = 4) changed their initial statements about the prototypes after having touched them for an extended period. For instance, one participant stated in the first exercise that she did not like the idea of having physical contact with a personal robot. Reflecting on her idea of an ideal personal robot in the fourth exercise, she now mentioned that she would like to engage in physical interaction with a personal soft robot by holding the robot in her hands to decrease stress levels, after having touched the prototypes on and off for more than 30 min. (P1.8).

5.4.3 Soft Biomorphic Robots for Emotional and Physical Well-Being

All participants (n = 10) imagined the prototypes to have specific functions in a domestic setting and to interact with humans. This occurred particularly in the third exercise where the participants envisioned the dynamic soft biomorphic prototype to perform practical tasks, such as cooking, and cleaning, and to take part in social interactions for companionship, entertainment, and care. A surprising finding to emerge from the data was that most participants (n = 7) expressed ideas about physically engaging with the soft biomorphic prototypes or integrating them on the body specifically to enhance emotional or physical well-being in either the second, third, or fourth exercise. Some imagined holding a moving robot for relaxation, and others imagined the prototypes would be fit for wearing on the skin to measure biometric data, or reduce anxiety or stress, while “concealing the technology” (P1.9). Related to this, in the third exercise, the majority (n =6) openly discussed the prototypes’ potential to take part in activities related to detecting emotional state and helping stimulate positive affect, and as something that “(…) brings the good vibes” (P1.4) when you are interacting with it. These participants imagined that soft biomorphic prototypes might help people relax through touch, robot movements, vibrations, or temperature regulation. It was particularly the prototypes’ soft materials and dynamic movements that drove these ideas, but some also coupled it to the prototypes’ visual aesthetics and “lifelike” appearances. Other participants (n = 2) linked the interaction with the prototypes to interactions that are social in the manner of how one interacts or communicates with an animal for enjoyment and pleasure, which is also known to help regulate emotions and improve mood [23, 24].

5.5 Discussion

Study 1 sought to explore which visions of interactions, relations, and personal robot applications encounters with biomorphic soft robots are generative of (RQ2). In accordance with prior work [60], we found that participants were eager to touch and interact with the soft prototypes. One of the most interesting findings was that nearly all participants expressed ideas of using or interacting with the soft biomorphic robots as a means to increase emotional or physical well-being. This could suggest that a biomorphic design aesthetic is well-fitting for personal robots intended to foster well-being and could signal to a person the robot’s ability to take part in comforting, e.g., in applications such as care or therapeutic robots. This novel finding is worthy of further study, as it can have implications for several proposed use cases of soft robots, including recent work on using soft robotics technology for stress and anxiety relief and supporting well-being [65,66,67,68, 96]. Participants also noted how soft biomorphic traits can help to “conceal the technology” and “make it [the technology] appear organic” and they associated the prototypes’ likenesses with living organisms with notions of social interaction or communication. This could also be an indication that a soft biomorphic design aesthetic is appropriate more generally in soft robotics applications that involve social HRI. The findings also suggest that although the prototype designs were not intended to resemble any specific animal or organism, a transference of existing social interaction models used with animals occurred, as some participants imagined interacting with the prototypes as if they were animals. Moreover, some participants argued that the “organic” aesthetic made the prototypes appear autonomous and self-governed, which could indicate that this specific design aesthetic has an effect on perceived agency. If this is the case, a biomorphic design aesthetic might be applicable for applications wherein it is important that the user perceives the soft robot as an autonomous agent.

5.6 Limitations

Study 1 has some limitations. Firstly, only a subset of the biomorphic prototypes was included. If other designs had been selected, the participants’ perspectives on the soft biomorphic design aesthetic may potentially have differed. Moreover, we did not include any soft robots with a traditional geometric design aesthetic to compare with in the study. Hence, the study’s results do not allow us to determine if the findings are driven by the biomorphic design aesthetic specifically or perhaps a more general effect of the soft robotic embodiment. The number of included biomorphic prototypes was limited to nine to avoid overwhelming participants and to allow for an initial in-depth exploration of participants’ qualitative assessments of this aesthetic before subsequently comparing them with traditional designs (see 6 Study 2).

We acknowledge another limitation in asking people potentially unfamiliar with robotics and the physiology of soft organisms to come up with ideas for soft robot use cases. Indeed, in Study 1, the majority of the participants expressed a lack of familiarity with robots, including soft robots, which may have constrained their ability to propose practical functionalities for the prototypes. However, this user-centered design approach allows participants to focus on high-level aspects of use cases rather than practical implementation, and doing so, to potentially also draw on cultural imaginaries about robots (e.g., fictitous robots they are familiar with from movies with abilities that are currently unrealizable in reality). By involving individuals with diverse backgrounds, this approach additionally has the potential to introduce new perspectives, knowledge, experiences, and issues unfamiliar to us as researchers [97]. Thus, we chose this approach to engage participants in imaginative exercises to explore how their knowledge and experience might stimulate creativity in the task and allow them to uncover pressing ideas and insights for further investigation in the development of personal soft robots.

Lastly, we conducted the thematic analysis by collating all data from the workshop, as we wanted to uncover general themes emergent in interaction. Hence, we did not differentiate between the different exercises of the workshop where participants performed different tasks. Thematic analyses of data from each stage of the workshop might have yielded more nuance, however at the cost of generalizability. Likewise, the decision to evaluate all prototypes together allows us to assess the ideas about applications, interactions, and relations the biomorphic design aesthetic produces overall, but not the more fine-grained determinants of these, e.g., specific formal traits such as shapes, colors, or surface textures.

6 Study 2

Motivated by the findings of Study 1 and to overcome the above-mentioned limitations, we carried out a second study to investigate if personal soft robots with enhanced biomorphic qualities are considered more appealing and more appropriate for applications wherein soft robots are used to support well-being through physical interaction than other soft robotics designs (RQ3). Secondly, we wanted to explore whether the soft robotic prototypes were considered more appealing following physical interaction with them, than at first sight, as indicated by Study 1 and prior work [60].

We designed the interaction scenario of Study 2 to emulate a use situation, wherein the moving soft robotic prototypes are used to support well-being through bodily contact. This proposed application was based on the idea that “breathing” soft robots might be used to facilitate relaxation (see 2 Related Work) through entrainment. Entrainment refers to the activity in which independent rhythmical biological systems interact [98], and one system temporally entrains the other system’s frequency [99]. Prior work on HRI has shown this principle to be applicable to technologies that use tactile stimuli to reduce anxiety [100] or regulate the heart rate to improve mental health [101].

We selected two of the dynamic soft biomorphic prototypes, Gray Tube and Green Oval, to use for the study. We considered these the most appropriate designs for physical handling, holding in the hands, and bodily contact as they are three-dimensional forms capable of supporting their own weight without buckling and are compact and coherent morphologies. In addition, we designed two new simplified geometric versions of the selected prototypes, Geometric Tube and Geometric Oval (see Fig. 3). We chose a simplified geometric design with monochrome coloring for these prototypes, as a contrasting alternative to the biomorphic aesthetic of the original prototypes. The geometric prototypes were made from the same materials and have the same overall shapes, sizes, colors, and movements as the two soft biomorphic robotic prototypes on which they are based, but lack the intricate details, surface textures, and polychromatic coloring of the biomorphic designs. Comparing these prototypes with the biomorphic designs thus allows for assessing whether the findings in Study 1 can be considered effects of the original prototypes’ biomorphic design aesthetic.

Fig. 3
figure 3

The four soft robot prototypes. From left to right: Gray Tube, Geometric Tube, Green Oval, and Geometric Oval

Full size image

We programmed a pulsing movement pattern for the prototypes, meant to resemble relaxed breathing in domestic animals (15 breaths per minute (BPM), following [102]), in which the prototypes calmly inflated and subsequently deflated. Although all prototypes used the same pulsing frequency and identical hardware for actuation, the variation in size and configuration of the air chambers resulted in different internal pressures and associated force transfers. Consequently, the prototypes with smaller air chambers (Gray Tube and Geometric Tube) exhibited pulsing experienced as more powerful compared to those with larger air chambers (Green Oval and Geometric Oval).

6.1 Hypotheses

We hypothesized that:

H1

Participants will deem the soft biomorphic prototypes more appealing than the geometric prototypes.

Following results from Study 1, we anticipated that the soft biomorphic prototypes would receive higher appeal ratings than the geometric prototypes would. This hypothesis was further motivated by the biophilia hypothesis [21] and recent research in soft robots and HRI that shows a preference for biophilic robot designs [68].

H2

Participants will deem the soft biomorphic prototypes more appropriate than the geometric prototypes for interactions intended to support well-being.

Based findings from Study 1, we expected that the soft biomorphic prototypes would be rated more appropriate for the application of improving well-being than the geometric prototypes.

H3

All included prototypes will be rated more appealing after physical interaction than at first sight.

We predicted that all four prototypes would receive a higher appeal rating post-interaction, as we observed in Study 1 that some participants changed their attitude towards the prototypes after having interacted with them for a while. Furthermore, this was motivated by prior work on soft robotics in HRI [60].

6.2 Participants

A total of 32 participants (female: n = 15; non-binary: n = 1; male: n = 16) were recruited through convenience sampling at the University of Southern Denmark’s main campus in Odense, Denmark. The participants’ ages ranged between 19–45 years (M = 24.5, SD = 5.86), and they self-reported their nationalities as Danish (n = 23), German (n = 2), Kurdish (n = 1), Nepalese (n = 1), Polish (n = 1), Mexican (n = 1), Greek (n = 1), Bosnian (n = 1), and Spanish (n = 1). All participants were university students enrolled in the following study programs: psychology, computer science, physiotherapy, law, biology, media science, engineering, applied mathematics, political science, biomedicine, global management and manufacturing, English, sports, robot technology, biochemistry, economics, and pharmacy. All participants were asked to indicate to which extent they agreed with the statement “I generally consider my knowledge and skills in the field of technology/robots to be high” on a 7-point Likert scale ranging from “1 – does not apply at all” to “7 – applies fully” (M = 3.5, SD = 1.87). No compensation was given to the participants.

6.3 Procedure

All participants took part in the study experiment individually with the first author as facilitator. Before the experiment, participants were briefly informed that the purpose of the project was to generate knowledge about which robot designs are suitable in contexts related to improving well-being by means of robotics technology. They were told that we wanted their feedback on four robot prototypes we had designed, and that they would interact with all prototypes shortly.

Upon having signed an informed consent form, the participant was taken into a classroom (see Fig. 4). The participant was seated near a table where the four prototypes were laid out. They were told that the prototypes in front of them were examples of soft robotics and were intended to provide relaxation and improve well-being. The facilitator informed them that this would occur by holding the prototypes in contact with one’s body. A demonstration sequence, in which each prototype moved one by one was shown, each moved for 10 s in the relaxed pulsing pattern. Subsequently, participants filled in the pre-interaction questionnaire. For each prototype they were asked to “Indicate your overall impression of the robot pictured above on the scale below from 1 to 7” with a 7-point Likert scale ranging from “1 – unappealing” to “7 – appealing”.

Fig. 4
figure 4

Experimental setup in a classroom at the University of Southern Denmark’s main campus in Odense

Full size image

Participants were then asked to pick up a prototype and gently hold it with both hands in contact with their chest until the prototype stopped moving. These movements were identical with how the prototypes moved pre-interaction, the only difference being that each movement pattern now lasted 30 s. Participants then filled out the post-interaction questionnaire that contained the same questions as the pre-interaction questionnaire and in addition asked for each prototype to “Indicate how appropriate you find the robot’s design for its intended use on the scale below from 1 to 7” on a 7-point Likert scale ranging between “1 – inappropriate” to “7 – appropriate”. For each rating, participants were also given the opportunity to elaborate why they had chosen a specific value. Lastly, after providing general comments and demographic data, all participants were given the opportunity to ask questions.

6.4 Statistical Analysis and Tabulation of Comments

Statistical analyses of participants’ ratings were run in IBM SPSS. Initially, a Shapiro–Wilk test of normality was conducted (p > 0.05), which showed that, except for the three ratings for Geometric Tube (pre-interaction appeal, p = 0.16; post-interaction appeal, p = 0.06; post-interaction appropriateness, p = 0.053), the data was not normally distributed. Visual inspections of histograms confirmed these results. Subsequently, the data was analyzed using Wilcoxon Signed Rank Tests and Friedman Tests as appropriate (see 6.5. Results).

Participants’ comments concerning the prototypes’ appealing, unappealing, appropriate, and inappropriate qualities were gathered in a table to allow comparison across the different classes of prototypes (see Table 3).

6.5 Results

6.5.1 Pre-and Post-Interaction Appeal Ratings

To investigate whether the prototypes were rated more appealing after physical interaction than at first sight, four separate Wilcoxon Signed Rank Tests comparing pre- and post-interaction appeal ratings for the four prototypes were conducted (see Table 2). For Gray Tube and Geometric Oval, no significant difference in appeal rating was found. For Geometric Tube, a statistically significant increase in appeal post-interaction, z = 2.11, n = 32, p = 0.04, with a small effect size (r = 0.26) was found. For Green Oval, a statistically significant increase in appeal post-interaction, z = 2.43, n = 32, p = 0.002, with a medium effect size (r = 0.30) was found.

Table 2 Median values for all ratings and statistics for pre-interaction and post-interaction appeal ratings comparison
Full size table

6.5.2 Overall Appeal Ratings for all Four Prototypes

Friedman Tests were conducted to compare the four prototypes’ appeal ratings using data from ratings pre- and post-interaction, to assess if any of the prototypes were rated as more appealing than others.

6.5.2.1 Pre-Interaction

Pre-interaction, there was a statistically significant difference between appeal ratings for the four prototypes (χ2 (3, n = 32) = 15.11, p = 0.002). Inspection of median values showed the following ranking of the four prototypes: Geometric Oval (Md = 5.5), Gray Tube (Md = 5), Geometric Tube (Md = 4), and Green Oval (Md = 3) (see Table 2). Post-hoc tests (including Bonferroni correction for multiple tests) showed that there was a statistically significant difference between ratings for Geometric Tube and Geometric Oval (Geometric Tube – Geometric Oval: p = 0.002). No significant differences were found between other prototypes.

6.5.2.2 Post-Interaction

For post-interaction ratings, no statistically significant difference was found between overall appeal ratings for the four prototypes (χ2 (3, n = 32) = 6.72, p = 0.081).

6.5.3 Appropriateness Ratings for all Four Prototypes

A Friedman Test was conducted to compare appropriateness ratings for the four prototypes. No statistically significant difference was found (χ2 (3, n = 32) = 6.32, p = 0.097).

6.5.4 Participant Comments in Post-Interaction Questionnaires

The participants’ comments were explored through categorization to gain insight into which formal traits of the four prototypes they found appealing, unappealing, appropriate, or inappropriate. A total of 125 comments were provided in the questionnaires. Comments made under ratings for appeal and appropriateness were categorized as being either positive, negative, both positive and negative, neutral, or ambiguous. Neutral and ambiguous comments were excluded. Positive comments concerning overall appeal were tabulated under the prototype to which they belonged under the column Appealing Qualities (n = 22). Similarly, negative comments were tabulated under Unappealing Qualities (n = 14). Likewise, positive comments concerning appropriateness were tabulated under Appropriate Qualities (n = 18) and negative comments under Inappropriate Qualities (n = 22). Comments containing both positive and negative parts were separated into their negative and positive constituent parts and tabulated as above (Appealing Qualities (n = 6), Unappealing Qualities (n = 6), Appropriate Qualities (n = 3), Inappropriate Qualities (n = 3)). General comments that mentioned specific qualities of the prototypes (n = 7) were tabulated in a separate cell of the table (nonspecific general comments (n = 4) were excluded).

Table 3 Participants’ comments that elaborate on qualities that are considered appealing, unappealing, appropriate, or inappropriate as well as general comments about the prototypes’ qualities
Full size table

Comparing the comments made for each of the soft biomorphic protypes and their geometric equivalents, contained in Table 3, reveals that both overlapping and distinct traits were mentioned as being appealing/unappealing and appropriate/inappropriate aspects of the biomorphic and geometric design aesthetics respectively.

For both Gray Tube and Geometric Tube, participants found it appealing that they were comfortable to hold and had pleasant inflations. In addition, Gray Tube was considered appealing due to its surface texture being pleasant to touch, its “hum” (P2.21), and its “organic” visual appearance (P2.17).

In terms of unappealing qualities, participants found that Gray Tube’s "pulse frequency is much too high” (P2.11) and that its movements could not be felt clearly, which was experienced as if it was “dying” and had “difficulties breathing” (P2.23). For Geometric Tube, participants found it unappealing that its shape was boring, that it looked “too hard/inorganic for the purpose” (P2.25), and that it was not pleasant to touch.

Participants who found Gray Tube and Geometric Tube appropriate mentioned size, stating that both prototypes fitted well in one’s hand. Additionally, Geometric Tube had an appropriate shape and a “very neutral design that makes you want to interact with it” (P2.8).

Both prototypes were deemed inappropriate because they were not pleasant to hold. Geometric Tube was also deemed inappropriate due to its angular and “strict” shape (P2.24) or its “dull” color and shape (P2.10). Some also found Gray Tube inappropriate because its surface had too many “bumps” (P2.11), because it “looks like a brick” (P2.14), and because its vibration was deemed “the least calming” of the four prototypes (P2.24).

For Green Oval and Geometric Oval, participants found both prototypes appealing because of the pleasant sensation when touching and holding them, their visual appearances, and their calming movements.

Participants that elaborated on unappealing qualities mentioned for both Green Oval and Geometric Oval that they had unappealing pulse frequencies and visual appearances. Additionally, a few found Green Oval’s vibrations and how it felt to hold it unappealing. For Geometric Oval, one participant found it unappealing because it was “too clunky” and did not have a good effect on her (P2.1), while another highlighted touching it as being unappealing.

As for appropriate qualities, participants commented that Green Oval “had the best vibration” (P2.24), that it was nice to hold and touch due to its texture, size, and shape. Geometric Oval was deemed appropriate because of its “very neutral design” (P2.8) and due to its color and round edges, because it fitted well in the hand and was pleasant to hold.

Green Oval was considered inappropriate because it looked “nasty” (P2.8), “weird” (P2.11), “chaotic” (P2.14), or “lightly repulsive” (P2.24), and due to its color and shape. For Geometric Oval, participants instead mentioned that the form was “clumsy” (P2.1, P2.7), that its size was too difficult to hold, and that a more “rounded/organic” shape would look more calming (P2.25).

Interestingly, participants also compared how well-aligned the prototypes’ visual appearances, tactile qualities, and movements were in some instances (n = 9). In these cases, participants mentioned discrepancies between the prototypes’ formal attributes. For instance, one participant argued that Gray Tube “looks nicer than the others” but that she “couldn’t feel it as much as the others” (P2.25), and another that Geometric Tube was “boring aesthetically” but “pleasant to interact with” (P2.17). Another participant mentioned how Geometric Oval’s “(…) form is not good” but that “its function is” (P2.11), while a different participant argued that Green Oval “(…) doesn't look super pleasant” but that “it's nice to touch it” (P2.28).

6.6 Discussion

Study 2 sought to investigate if soft robots with enhanced biomorphic qualities are considered more appealing and appropriate for applications wherein soft robots are used to support well-being by means of physical interaction (RQ3) and, additionally, whether soft robotic prototypes in general are rated as more appealing following physical interaction than at first sight.

We hypothesized that the soft biomorphic prototypes would be considered more appealing than the geometric prototypes (H1). The results did not support this hypothesis, as the only statistically significant difference found in appeal ratings for the four prototypes was between the two geometric prototypes (Geometric Tube and Geometric Oval, pre-interaction data).

Based on Study 1, we hypothesized that participants would find the soft biomorphic prototypes more appropriate than the geometric prototypes for supporting well-being through HRI (H2). We found no significant differences between the ratings for the four prototypes, and the results did therefore not support this hypothesis. The results from Study 2 suggest that in Study 1 it might have been the soft robots’ soft materials and not the soft biomorphic aesthetic per se that made people suggest applications for them in improving well-being. A participant in Study 2 also alludes to this in a general comment, stating that “[s]oft material[s] can help calm you down” (P2.21). However, categorization and analysis of comments provided indication that it was not solely their soft materiality that was taken into account, when considering their appropriateness. Participants also mentioned, e.g., visual appearance, stating that Geometric Tube looked “(…) too hard/inorganic for the purpose” (P2.25), and that Green Oval looked “(…) a bit nasty and [it] doesn’t make you want to touch it” (P2.8). But, also, more specifically form and color, mentioning being “(…) not too fond of the color and shape” of Green Oval (P2.17) or explaining that Geometric Tube’s “(…) color and shape are relatively ‘dull’” (P2.10). Others focused on movement, expressing that Green Oval “(…) had the best vibration” (P2.24), or the sensation of touching the prototypes, mentioning how Geometric Oval “(…) was pleasant to hold” because it “(…) [f]itted the hand well” (P2.15). As participants mentioned the prototypes’ visual appearances, forms, colors, movements, and tactility as affecting their evaluations of appeal and appropriateness, it is plausible that these design aspects interact in a more complex and intricate way in these assessments, than what is captured in the distinction between a biomorphic and geometric design aesthetic. However, contrasting the participants’ comments about the biomorphic and geometric prototypes shows that there are distinctions between what words participants use to describe the two styles, particularly their visual appearance and overall form. For instance, where biomorphic prototypes’ visual appearances and forms are described as “chaotic” (P2.14), “organic” (P2.17), “repulsive” (P2.24), and “weird” (P2.11)”, the geometric prototypes are designated as “neutral” (P2.8), “boring” (P2.12), “angular” (P2.7), “strict” (P2.24), and “dull” (P2.10). This may suggest that the two styles’ visual appearances and overall forms are perceived differently and maybe even as contraries, with the biomorphic design aesthetic being “organic” and the geometric design aesthetic being “angular”. Thus, although no quantitative statistically significant differences were found between the appropriateness ratings, the qualitative data indicates that the two groups of prototypes are perceived differently from one another in terms of their visual appearances and overall forms.

Based on results of Study 1 and prior work, we hypothesized that interaction with the soft robotic prototypes would lead to an increase in appeal rating (H3). We found partial support for this hypothesis, as appeal ratings for Geometric Tube and Green Oval were statistically significantly higher post-interaction. This finding indicates that physical interaction with soft robots can have a positive impact on how appealing people find them to be, but that this is not always the case. It is possible that this effect is dependent on interaction time or the context and framing of the interaction, which were different in Study 2 than in Study 1, and therefore we did not manage to produce the effect consistently. To understand this finding better, we compared the change in pre- and post-interaction appeal rating for each participant for all prototypes. We found that post-interaction, several participants had changed their rating of a prototype one step or more ( ±) on the 7-point Likert scale (Gray Tube—62.5% (n = 20); Geometric Tube – 62.5% (n = 20); Green Oval – 59.4% (n = 19); Geometric Oval – 78.1% (n = 25)). In fact, a larger proportion had changed their rating two or more steps ( ±) (Gray Tube—28.1% (n = 9); Geometric Tube – 25% (n = 8); Green Oval – 25% (n = 8); Geometric Oval – 31.3% (n = 10)). This could indicate a polarizing rather than unidirectional effect of the interaction, i.e., that post-interaction some participants found them more appealing and others less, hence we did not find any statistically significant difference with the Wilcoxon test. However, this hypothesis does not generalize easily as a statistically significant higher post-interaction appeal rating was indeed found for both Geometric Tube and Green Oval.

6.7 Limitations

Study 2 is subject to certain limitations. Only two out of the nine soft biomorphic prototypes from Study 1 were included. Hence, the chosen designs may not have possessed the full set of qualities that drove the findings of Study 1. Furthermore, the participants did not evaluate a fully functional biomorphic robot validated to improve well-being by means of physical interaction, but only prototypes framed as possessing this ability. Another limitation of the study consists in the self-report measure used. Only two rating items for each prototype, namely appeal and appropriateness, were used to ensure a fast completion time, as participants were not financially compensated. Other validated self-reporting measurement instruments that include more sub-items for each construct may have been more suited to detect a difference in how participants assessed the two categories of prototypes. Instead of measuring appeal we might, for instance, have used the likeability scale from the Godspeed Questionnaire series [103] or the warmth scale from the Robotic Social Attributes Scale (RoSAS) [104]. Instead of appropriateness, we could have used intention to use from Heerink et al.'s [105] extension of the Unified Theory of Acceptance and Use of Technology (UTAUT) measure.

7 General Discussion

The present work has sought to explore how specific design features of silicone-based soft robots can increase their organic qualities (RQ1); which visions of interactions, relations, and personal robot applications encounters with such biomorphic soft robots are generative of (RQ2); and if soft robots with enhanced biomorphic characteristics are considered more appealing and appropriate for supporting well-being through physical interaction (RQ3). To address RQ1, we formulated a soft biomorphic design aesthetic aimed at enhancing the inherent organic qualities through visual, haptic, and kinetic added attributes, and explored it through the construction of a set of prototypes. RQ2 and RQ3 were addressed in the first and second study, respectively. Combined, this work allows us to identify some central issues and tentative guidelines concerning the enhancement of biomorphic traits in soft personal robots.

A common theme emerging from both studies is the negotiation between how organic appearing versus geometric or “constructed” a soft personal robot should be. While some participants in both studies argued that the soft biomorphic prototypes’ organic aesthetic made them more appealing and increased their desire to interact with the prototypes, others argued that the organic features of some prototypes made the prototypes appear “alien”, “uncontrollable”, or “chaotic”, and decreased their desire to interact with them. This division would appear to mirror the often-cited uncanny valley hypothesis [106], which states that if a robot’s appearance becomes more human-like, it will have a positive effect on observers’ emotional responses until a threshold whereafter it becomes eerie. The associated recommendation of Mori [106] that designers should strive for a restrained level of human likeness and only a “considerable sense of affinity” also bears similarities with comments by participants from the first study who appreciated that the biomorphic prototypes possessed both biomorphic and “man-made” or “constructed” traits. Comments from participants in Study 2 (Table 3) showed that although many people found the geometric prototypes’ simple and “clinical” aesthetic appealing, multiple participants did not find their sharp-edged designs appealing or appropriate for HRI to support well-being. Based on these insights, we propose to consider the biomorphic and geometric design aesthetics as two poles within the design space of soft robotics. Mixing or mediating between these two styles could potentially provide a shortcut to discovering designs for personal soft robots that are considered appealing, as they embody both organic and man-made qualities. The suggestion to mix the two styles aligns with findings of prior work showing that robots mixing biomorphic and product-like aesthetics were favored over strictly biomorphic or strictly device-like embodiments [107]. Our findings suggest that this principle might be extended to soft robotics technology as well. Thus, designers need to find a suitable balance to suit the specific use of the personal robot and its intended tasks and interaction modes. For instance, personal robot designs intended for close physical HRI might benefit from increasing the biomorphic characteristics over the geometric characteristics to aim for softer and more organic curves that afford bodily interactions. Based on the findings from Study 1 and 2, we summarize the formal attributes that are assessed as making the prototypes appealing/unappealing (Table 4). We recommend designers to use this overview for inspiration when designing soft personal robots intended for close physical HRI.

Table 4 Summary of formal attributes in soft personal robot designs that were assessed as appealing and unappealing respectively
Full size table

As evident from Table 4, the perception of softness can be assessed both as appealing and unappealing, even when utilizing the same physical material. This intricacy needs careful attending to when designing soft robots. It also highlights the need for further research into the perception of different shore hardness silicones in personal soft robot designs. Moreover, it is crucial to investigate how other underlying design aspects may contribute to the varied perception of the same material as appealing or unappealing in different prototypes. Future studies should thus explore additional design dimensions to comprehensively account for the factors influencing the perceived appeal of soft robot designs.

Prior research that would appear to challenge the recommendation to mix biomorphic and geometric design aesthetics, however, also exists. Löffler et al. [78] found that the uncanny valley effect extends to zoomorphic robots as a quadratic function, where robots that are either high or low in animal-likeness are preferred over designs with medium animal-likeness. They therefore argued that robot designers should maintain consistency in the level of animal-likeness being either low or high, rather than incorporating a combination of realistic and unrealistic features in robot designs. According to Löffler et al.’s [78] findings, the best predictors for animal-likeness in robot designs were natural color schemes and proportions, the use of plastic or metal, joint visibility, and the presence of a snout. Comparing these findings with our soft biomorphic prototypes (see 4 Prototypes Designs), it can be seen how many of these characteristics are contained in our designs. For instance, all prototypes are made from flexible soft silicone and, thus, avoids the use of metal and plastics; all prototypes follow natural color schemes; most of the prototypes do not have joint visibilities (e.g., Claw, Tuberous Form, Green Oval); and some of the designs contain natural proportions (e.g., Rugged Surface and Skinlike Surface). Thus, as the biomorphic prototypes are predominantly “animal-like” following these characteristics, it can be argued that a biomorphic design aesthetic can be mixed with a geometric style and still contain appropriate levels of biomorphic traits to ensure that the soft robot designs are not deemed uncanny. Furthermore, similar to our recommendation that personal robot designs must balance between biomorphic and geometric traits to achieve appealing and appropriate designs, Löffler et al. [78] argued that robot designers must balance between imitating living creatures and conveying that a robot is a thing and not a living creature to ensure usable robots for HRI.

Another discovery of the present work concerns the relationship between a prototype’s visual appearance, tactile qualities, and movements. Qualitative data from Study 2 shows how a few participants compared the consistency between different formal attributes when evaluating a prototype’s appeal or appropriateness for supporting well-being in HRI. In multiple cases, participants mentioned discrepancies between the prototypes’ formal attributes. For instance, one participant expressed that Green Oval was “(…) lightly repulsive to look at, however, it had the best vibration” (P2.24), and another mentioned that Gray Tube “(…) looks nicer than the others (more organic and calming) but I couldn’t feel it as much” (P2.25). Similarly, a different participant explained in a general comment that the robots “(…) who looked attractive were not as pleasant to interact with as those who looked less attractive” (P2.2). A design recommendation would therefore be to strive for coherence between a robot’s visual, tactile, and kinetic qualities to make the most appealing and appropriate designs for personal robots in HRI. This recommendation bears similarities to arguments put forth in [108], that a robot’s appearance and overall form should match its movements, and in [13], that a robot’s capabilities should match its form and behavior. To achieve this coherence between visual appearance, tactile qualities, and movements, we suggest also aligning the design with its intended use context and user group by prioritizing user-centered design and iterative prototyping. By emphasizing the intended users’ ideas, needs, expectations, and cultural norms, their aesthetic, tactile, and kinetic expectations and preferences can be determined. Through user involvement, testing, and identification of inconsistencies between the different attributes, designers must iteratively refine the design to ensure appealing and appropriate robots for the specific use case scenario.

8 Conclusion

In this article, we proposed a novel design paradigm for soft robotics, through the concept of Soft Biomorphism, aimed at attaining aesthetic similarities with soft-bodied organisms while avoiding representing (an abstraction of) a specific organism. We designed a range of soft biomorphic prototypes and conducted two empirical studies. Study 1 sought to interrogate which personal robot applications and notions of human–robot interactions and relations the biomorphic design aesthetic is generative of. We found that most participants associated this aesthetic with soft personal robots aimed at fostering physical and emotional well-being. Based on these findings, we conducted a second study (Study 2) to examine whether soft robots with enhanced biomorphic qualities would be considered more appealing and appropriate for applications wherein soft robots are used to support well-being through physical HRI. While the results did not show a statistically significant preference for soft biomorphic prototypes over geometric prototypes, qualitative data indicated that participants perceived the two aesthetic styles differently from each other in terms of visual appearance and overall form. Moreover, we found significant differences in appeal ratings post-interaction for certain prototypes, indicating that physical interaction with soft robots can affect perceived appeal. Based on our findings, we highlighted central issues to take into consideration when using biomorphic aesthetics in soft personal robot designs and gave tentative design recommendations to combine biomorphic and geometric traits as well as an overview of appealing and unappealing formal attributes. Finally, we recommended to align the robot’s formal attributes (visual appearance, tactility, and movement) to ensure appropriate designs.

9 Future Work

The present work has taken first steps in exploring soft biomorphism as a design aesthetic for soft robotics and assessing its benefits and effects on HRI. Future work should aim to elucidate the findings of this largely exploratory study with respect to specific types of soft personal robots. The contribution of specific formal attributes of biomorphic soft robots to the high-level impression of the robot and the experience of the HRI should be examined in more detail as well as their interaction. Furthermore, it needs to be established what level of soft biomorphism is appropriate for specific applications and different types of HRI with soft robots.