ELEGNT: Expressive and Functional Movement Design for

Non-anthropomorphic Robot
Yuhan Hu Peide Huang
Apple Apple
Cupertino, United States Cupertino, United States
[email protected] [email protected]

Mouli Sivapurapu Jian Zhang

Apple Apple
arXiv:2501.12493v1 [cs.RO] 21 Jan 2025

Cupertino, United States Cupertino, United States

[email protected] [email protected]

Figure 1: Overview of our research hypothesis: robots should not only move to fulfill functional goals and constraints, i.e.,
robot moving from the initial state to goal state through a shortest, feasible trajectory (function-driven trajectory), but also use
movements to express its internal states to human counterparts during the interaction, i.e., via expression-driven trajectory to
express robot’s intention, attention, attitude, and emotions.
Abstract emotions—alongside traditional functional considerations like task
Nonverbal behaviors such as posture, gestures, and gaze are es- fulfillment, spatial constraints, and time efficiency. In this paper,
sential for conveying internal states, both consciously and un- we present the design and prototyping of a lamp-like robot that
consciously, in human interaction. For robots to interact more explores the interplay between functional and expressive objectives
naturally with humans, robot movement design should likewise in movement design. Using a research-through-design methodol-
integrate expressive qualities—such as intention, attention, and ogy, we document the hardware design process, define expressive
movement primitives, and outline a set of interaction scenario story-
Permission to make digital or hard copies of all or part of this work for personal or boards. We propose a framework that incorporates both functional
classroom use is granted without fee provided that copies are not made or distributed and expressive utilities during movement generation, and imple-
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. Copyrights for components of this work owned by others than the ment the robot behavior sequences in different function- and social-
author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or oriented tasks. Through a user study comparing expression-driven
republish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from [email protected].
versus function-driven movements across six task scenarios, our
Conference DIS’25, July 05–09, 2025, Funchal, Madeira findings indicate that expression-driven movements significantly
© 2018 Copyright held by the owner/author(s). Publication rights licensed to ACM. enhance user engagement and perceived robot qualities. This effect
ACM ISBN 978-1-4503-XXXX-X/18/06 is especially pronounced in social-oriented tasks.
https://doi.org/XXXXXXX.XXXXXXX
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Keywords expressive utility emphasizes the trajectories taken to achieve these

human-robot interaction, non-anthropomorphic robot, expression, goals. The latter incorporates considerations for expressing and
research-through-design, robot movement, theory of mind communicating the robot’s intention, attention, attitude, and emo-
tional state to the user, as illustrated in figure 1. We elaborate the
ACM Reference Format: building blocks for expressive movements with kinesics and prox-
Yuhan Hu, Peide Huang, Mouli Sivapurapu, and Jian Zhang. 2018. ELEGNT:
emics primitives. Through video prototyping and storyboarding,
Expressive and Functional Movement Design for Non-anthropomorphic
we demonstrate a range of use cases and task scenarios for the lamp
Robot. In Proceedings of Designing Interactive Systems (Conference DIS’25).
ACM, New York, NY, USA, 13 pages. https://doi.org/XXXXXXX.XXXXXXX robot in home environments, organized along the dimensions of
robot agency and the social versus functional nature of tasks. Our
work aims to offer design inspiration and a framework for future
1 Introduction integration of expressive robots into daily life.
In this paper, we present ELEGNT, a framework of expressive To evaluate the benefits of incorporating expressive movements
and functional movement design for non-anthropomorphic robot. and comparing the outcome between expressive and functional
We argue that robots should not only move to fulfill functional utilities, we conducted a user study comparing expression-driven
purposes and constraints but also move “elegantly” - using their movements with function-driven movements across various task
movements to express intentions, attention, and emotions to their scenarios. Participants (n=21) were tasked to watch human-robot
human counterparts during human-robot interactions (HRI). We interaction videos in six different tasks, each with two robot vari-
present our practice of designing movements incorporating func- ations. After each video, they evaluated the perception along the
tional and expressive utilities and a user research to understand metrics of engagement, intelligence, human-likeness, willingness
the effect of expressive movements. to interact, sense of connection, and robot character.
Robots are increasingly entering homes as assistants and com- The results reveal that movements incorporating expressive
panions, making it essential to understand how they coexist with utilities significantly increased user ratings, compared to move-
humans, interact with people, and fulfill functional and social roles ments only driven by functions. Perceptions varied across tasks,
in everyday life. Like most animals, humans are highly sensitive with expressive movements particularly benefiting social-oriented
to motion and subtle changes in movement. Existing research on tasks, such as entertainment and social conversations. Results also
robotics suggests that a robot’s movements can not only perform suggested demographic effects from participants’ age and profes-
practical functions but also convey the robot’s purpose, intent, state, sional backgrounds. Qualitative analysis revealed additional in-
character, attention, and capabilities [17]. sights into perceived robot characteristics and how users infer
While much of the research separates pragmatic robotics—such the robot’s state from its movements, suggesting the potential to
as robotic arms performing household tasks—from social robotics, customize these movements to individual preferences and align
like therapeutic robots providing emotional support, we argue that them with other interaction modalities, including voice and light.
any robot interacting with humans, even if designed primarily for We hope the proposed framework and our study outcomes will
practical functions, embodies social value and should have its design inspire future research on expressive movement generation for
and behaviors shaped accordingly. For instance, in a collaborative non-anthropomorphic robots.
manipulation task with a human teammate, a robot should not only
consider functional actions, such as picking up and placing objects, 2 Related Work
but also employ expressive movements that convey its intentions,
In this section, we review related work in the domain of non-
state, and even character traits. These expressive cues can help
anthropomorphic robots for human-robot interactions, with an
human collaborators anticipate the robot’s actions, build trust, and
emphasis on robot expression and movement-centric design. Addi-
foster a sense of comfort and enjoyment in the collaborative process.
tionally, we discussed related work from animation and character
Our research addresses several questions: How do we design
design, which highly inspired this work.
expressive movements alongside functional actions for robots in-
teracting with humans? What are the design spaces and movement
primitives? How do users perceive robots employing expressive 2.1 Non-anthropomorphic Robots for HRI
movements versus purely functional ones? The form and appearance of a robot impacts how people perceive
In this paper, we present our practice on designing an non- it, interact with it, and build long-term relationships with it [6].
anthropomorphic robot in the form of a lamp, featuring a 6-DOF Existing robotic form can be categorized with anthropomorphic
arm and a head equipped with a light and a projector. As a common (human-like) [22], zoomorphic (animal-like), and appliance-like,
household form factor, the lamp-like robot offers a rich design and as illustrated in figure 2. While it is beneficial to have robots with
interaction space to engage with both the environment and users anthropomorphic design that have a positive impact on acceptance
through lights and movements—for example, directing a user’s [11], research also suggested that user preferences of robot forms
attention by illuminating specific spaces or objects. were task and context dependent [15]. The appearance of robots
We use research-through-design (RtD) [13] methodology to it- should match its capabilities and user expectations.
erate the design of the robot’s form, movement, and interaction Anthropomorphic robots use human-like gestures and expres-
scenarios. We formulate movement objectives with both functional sions, such as body pose and facial expressions, mapped from
and expressive utilities - whereas functional utility brings the ro- humans’ behaviors to convey a various of internal states. Non-
bot’s from initial to goal states within the physical and task space, anthropomorphic robots does not have an explicit vocabulary or
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

mapping for expression. Existing research suggest multiple expres-

sive channels for non-anthropomorphic robots, including move-
ments [7, 17, 27, 32, 37], light/color [30, 33], sound [28], tactile
expression [8, 19, 20], and so on. For instance, Shimon [18], a mu-
sical improvisation robot, incorporates a socially expressive head
to communicate its internal states including rhythm, emotional
content, intensity, as well as manage turn-taking and attention
between the robot and human musicians, supporting joint musi-
cal attention. “The Greeting Machine” [3], embodies a small ball
rolling on a larger dome, designed to communicate positive and
negative social cues in the context of opening encounters. Existing
studies suggest that even abstract and simplistic movements, such
as Approach and Avoid, are effective in expressing robot intentions,
evoke positive and negative experiences of the users.
This work is highly inspired by the paper ”Designing Robots
with Movement in Mind” [17], which present technics of movement
centric design, including character sketches, video prototyping,
and Wizard of Oz studies. They illustrate the approach and de-
sign strategy with the design of non-anthropomorphic robots and
robotic objects. They suggest movements as robot dynamic affor- Figure 2: Existing robot form spectrum: Anthropomorphic,
dance, which help cues users on potential actions and interactions Zoomorphic, and Appliance-like.
that the robot is capable of. The robot’s expressive movements
were considered early on in the design process, and may co-evolve
in the design iterations with robot hardware appearance and use
relationships, and emotions. These works highlight the power of
cases. Many recent research [14, 25] shift the robot design focus
motion, even with simple geometry, is effective in storytelling and
from the production of life-like forms to the process of movement
expression.
and kinesthetic creation. It is important to consider the expressive
Research also shows that movements do not need to mimic hu-
power of the robot’s movement - design movements that express
man forms in detail to be perceived as intentional or expressive.
the robot’s purpose, intent, state, mood, personality, attention, re-
Humans are adept at interpreting the movements of abstract shapes,
sponsiveness, intelligence, and capabilities [34]. In this work, we
as demonstrated by studies on point-light displays [29], where par-
take the idea of movement-centric design further with illustrated
ticipants could classify activities and recognize individuals from
movement designs grounded in real-world interaction scenarios
minimal visual cues. Beyond recognition, humans often attribute
and conduct a user study to evaluate the effects of expression-
internal states, characters, and intentions to abstract movements,
driven movements versus purely functional ones. Additionally, our
as exemplified by the Heider-Simmel Illusion [1], where simple geo-
work follows the common practices of research-through-design ap-
metric shapes moving in suggestive ways are perceived as having
proaches, combining artefact-centered research [9] and speculative
purpose or personality. This phenomenon is closely tied to the The-
design explorations [2, 21].
ory of Mind [5], which describes humans’ ability to infer mental
states and intentions from observed behaviors.
2.2 Movement Design for Expression Drawing inspiration from animation principles and leveraging
humans’ innate sensitivity and projection to motion, we aim to
Movement plays a fundamental role in how humans perceive and
design and program physical robots with movements that effec-
interact with the world. Humans, like many animals, are highly
tively convey expressive and intentional behaviors. These principles
sensitive to motions [16]. Movements are essential in the coordina-
form the foundation of our work, combining the expressiveness
tion and performance of joint activities, serving to communicate
of motion with functional considerations to create engaging and
intentions and refer to objects of shared attention [10].
meaningful interactions.
Insights into expressive movement design can be drawn from do-
mains beyond robotics, such as animation, behavioral science, and
performing arts [31]. In these fields, movement is used as a medium 3 Methodology
for communication, enabling objects, characters, and forms to con- In this section, we outline the design process of a lamp-like robot,
vey emotions, intentions, and narratives. For example, in character developed by a team of HRI researchers, roboticists, and anima-
animation, abstract forms like dots, lines, and shapes are brought tors. Through iterative brainstorming, sketching, storyboarding,
to life through motion, timing, and staging. A classic example is the and both hardware and software prototyping, we explored a range
animated short "The Dot and the Line" [24], where all expressions of design considerations, including form selection, movement de-
are conveyed through motion with minimal visual elements. The sign, and potential use cases. Rather than providing an exhaustive
Pixar’s iconic animation "Luxo Jr." [23], also serves as the primary taxonomy of design spaces, our aim is to highlight key design oppor-
inspiration of the lamp form, featuring two desk lamp characters, tunities and primitives that can inspire and guide future research
demonstrating how simple movements can communicate narratives, and practice.
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

the potential for mobility, it also introduces challenges such as in-

creased control complexity and a higher risk of physical collisions
during interactions. In contrast, ceiling-mounted lamps minimize
these risks but are limited in their interaction capabilities, as they
can only provide light from a top-down angle. Additionally, ceiling
robot movements are tend to be neglected during interactions since
they fall outside the user’s line of sight. Beyond spatial coverage,
we aim for the robot to use kinesthetic movements for expressive
purposes, such as nodding, shaking its head, or leaning forward and
backward. Therefore, the motors must be strategically positioned
to accommodate these motion ranges.
We also explored other non-anthropomorphic forms with simi-
lar range of motions, including designs inspired by a flower and a
giraffe, and abstract forms like sculptures and art pieces, as shown
in figure 3. While this paper focuses primarily on the tabletop lamp
integration, we envision that certain design principles may trans-
late across different embodiments. For example, movement speed,
pauses, and proxemic relationships could be applied universally.
However, some design patterns, like a nodding gesture, may map
differently depending on the embodiment. In forms with a clear
head-neck relationship, the gesture might be immediately recog-
nizable, whereas in more abstract forms, it could invite broader
interpretations.
Through iterative rapid prototyping, we integrate a robotic hard-
ware platform for further testing and deployment, as illustrated in
figure 4 (left). The robot is composed of a re-purposed 6-DOF robotic
Figure 3: Sketching ideas of non-anthropomorphic robots arm [36], a 3D-printed plastic lamp head with an embedded LED
with different form factors, sizes, and placements. light, a laser projector, and an internal camera, plus a downward-
facing external camera. Additionally, it is equipped with a voice
system to listen and speak to the user.

3.1 Designing Robot Form 3.2 Generating Robot Movements

There have been numerous explorations into home robots, such Existing research has shown that when communicating stories and
as vacuum robots [38], table-top robotic assistants [35], robot pets evoking emotions, movement often plays a more important role
[12], and humanoids [26]. These robots often take on anthropo- than form. Characters and emotions can be conveyed through the
morphic, zoomorphic, or appliance-like forms, as illustrated in timing and quality of a movement. For example, the Heider-Simmel
figure 2. Existing HRI research indicates that a robot’s form can Illusion [1] demonstrates that even simple geometric shapes can be
shape user expectations and influence interaction affordance. For perceived as human figures if they appear to move on their own
instance, users may expect a humanoid robot to interpret facial accord. Humans naturally project character and metaphorical states
expressions and gestures, whereas a vacuum robot might invite less onto moving objects. By intentionally designing these movements,
social engagement. Aligning a robot’s form with user expectations we can instill character perceptions into robots, creating social
and functional capabilities is a critical design consideration. bonding and tolerance between robot and human, and make the
Inspired by the characters in “Luxo Jr.” [23], we adopt the form interactions more enjoyable.
factor of a desk lamp. Although primarily appliance-like, it incorpo-
rates subtle anthropomorphic elements—such as the lamp head and 3.2.1 Framework Formulation. To consider the problem of gen-
the arm connecting the head to the stand—that evoke the appear- erating movements considering both functional and expressive
ance of a head and neck. The lamp’s light and camera can also be objectives, we present the high-level formulation of the problem
mapped to robot “eyes”, providing a design opportunity to convey to guide the low-level trajectory design and integration. We math-
the robot’s attention and purpose. ematically formulate the robot movement problem as a Markov
We consider aesthetic, expressive, and pragmatic aspects when Decision Process (MDP) defined by a tuple (𝑆, 𝐴, 𝑃, 𝑅). At timestep
prototyping the hardware of the robot. From a pragmatic perspec- 𝑡 ∈ 0, 1, . . . ,𝑇 , the state 𝑠𝑡 ∈ 𝑆 consists of the robot joint angles, the
tive, we aim for the robot to have a wide range of motion, allowing tool states, and the environment states. For example, the tool states
it to cover a reasonable interaction space—for example, transition- would include turning the light on and off, as well as projecting
ing from illuminating a table to lighting up a couch. We explored images. The environment states include the perceived state of the
various placements and configurations for the lamp robot, including users and other object of interests in the environment. The action
ceiling-mounted, tabletop, and floor lamp designs, as illustrated in 𝑎𝑡 ∈ 𝐴 consists of the change in joint angles and the tool event. The
Figure 3. While a floor lamp offers broader spatial coverage and transition function then defines 𝑠𝑡 +1 = 𝑃 (𝑠𝑡 , 𝑎𝑡 ). For simplicity, we
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Figure 4: Hardware composition of the lamp robot (left); Interaction modalities between human and robot, including gesturing,
verbal communication, light and projection display, and touch interaction (right).

also denote the trajectory 𝜏 = (𝑠 0, 𝑠 1, . . . , 𝑠𝑇 ). The reward function the goal state, such as whether it moved to the desired position,
𝑅 consists of two parts: functional utility 𝐹 and expressive utility 𝐸. turned on the light, and projected the accurate information.
Functional Utility 𝐹 defines the function-driven utility of reaching Expressive utility 𝐸, on the other hand, motivates the actions
certain states: aimed at communicating the robot’s traits, states, and intents to its
𝑇
∑︁ human interaction partners. For example, the robot may increase
𝐹 (𝜏) = 𝑓 (𝑠𝑡 ) (1) expressive utility by looking toward a book before moving to it or
𝑡 =0
displaying curiosity through head tilts. Expressive utility can be
Without loss of generality, we assume there is only one goal state
measured by users’ perceptions of the robot, including perceived
𝑠𝑔 . In this case, 𝑓 (𝑠𝑡 ) = 1 (𝑠𝑡 = 𝑠𝑔 ) , where 1 (·) is the indicator
intelligence, intuitiveness, interaction quality, trust, engagement,
function.
sense of connection, and willingness to use the robot. Drawing on
Expressive Utility 𝐸 defines the expression-oriented utility of
Theory of Mind (ToM)—the human cognitive ability to attribute
reaching certain states:
mental states such as beliefs, desires, emotions, and intentions
𝑇
∑︁ to others—we incorporate the following expression categories to
𝐸 (𝜏) = 𝑒 (𝑠𝑡 ) (2) capture expressive utilities in the design of our expressive motion
𝑡 =0
library.
In this work, we draw on design research methods to define 𝑒 (·)
along the expressive dimensions of attention, emotion, intention, Intention. Intention refers to the purpose behind the robot’s ac-
and attitudes, as discussed in Section 3.2.2. tions and the anticipation of its upcoming movements. For instance,
Finally, the objective is to maximize the total utility: when a robot extends its hand, the user can identify which object
the robot intends to pick up and what it plans to do with it, enabling
max 𝐹 (𝜏) + 𝛾𝐸 (𝜏) (3)
𝑎 0 ,...,𝑎𝑇 −1 cooperation, supervision, or intervention as needed. In the case of
where 𝛾 is the weight for the expressive utility, which could vary a lamp robot, it might briefly turn its head toward a target before
with different task and user. In Section 4, we present a user study to moving to reach or interact with it. This behavior signals the robot’s
evaluate the perception difference between the robot movements intention, indicates a shift in attention, and cues the user about the
when taking 𝛾 > 0 versus when 𝛾 = 0. next action.
3.2.2 Functional and Expressive Utility. In the context of the lamp Attention. Attention refers to where the robot’s focus is directed,
robot, Functional utility 𝐹 drives motions that aim at achieving a with gaze serving as a strong indicator of that focus. For instance,
physical goal state, such as taking the initial state of user’s reading when a robot looks at an object, it may be analyzing it or preparing
activity or an explicit verbal request, the robot moves to face the for upcoming actions. In the context of a lamp robot—where camera
book and turns the light on, as well as projects assistive information and light act as metaphoric “eyes”—we design gaze behaviors such
such as a visualization of a content in the book. The functional that looking toward the user can signal attention, for example when
utility is measured based on the level of completion of the task in the user is speaking. Similarly, a robot can exhibit joint attention by
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

robots can use kinesics—expressive body movements—to commu-

nicate information and convey mental states or attitudes [4]. Ki-
nesics encompass both spatial (pose-related) and temporal features
as design primitives. For spatial features, robots can incorporate
metaphorical gestures to represent various states. For instance, a
lamp-like robot with a head-and-neck configuration might nod
or shake its “head” to display attitudes, or lower it to convey sad-
ness. The lamp’s long arm joint could also be imagined as a lower
body, enabling gestures like “tail wagging” to signify excitement
or “sitting down” to imply relaxation. For temporal features, robots
can adjust parameters such as speed, pauses, and acceleration (or
jerkiness) to communicate attitudes and emotions. For example,
adding pauses and jerky movements might suggest hesitation or a
lack of confidence. Varying movement speed can signal different
levels of emotional arousal: quick, sharp movements may indicate
high-arousal states like excitement or fear, while slower, smoother
motions might convey calmness or sadness.
Similar to humans, robots can use proxemics—the management
of spatial distance—to express their relationship with the environ-
ment and the people around them. This helps set expectations, estab-
lish communication channels, create boundaries, and signal context.
Proxemics in robots can involve both static and dynamic motion
primitives. For static primitives, robots could position themselves
relative to an object or person to convey attention and intention.
For example, directing their gaze at an object and using light or
projection to highlight it can signal focus or communicate context.
Pointing their “head” away from an object might suggest ignorance
or disinterest. Close proximity—such as touching an object—can
indicate affection or interest. For dynamic primitives, robots can
Figure 5: Illustration of the design space for expressive robot use movement to express attitudes or intentions. Approaching or
movements, including kinesics and proxemics movement avoiding an object may reflect a robot’s stance toward it, while
primitives. shifting direction between objects or events can indicate changing
attention. Dynamic behaviors can also incorporate the use of light
accompanied with movements to guide user attention or emphasize
gazing at or illuminating the same object or event as the user. For
a point, such as for reminders or persuasive cues.
example, while a user operates an object, the robot might watch
In sketching out this design space, we aimed to illustrate how
the user’s hand and the object being manipulated.
kinesics and proxemics can serve as motion primitives for gener-
Attitude. Attitude refers to the robot’s stance toward a person, ating robot expressions. Rather than offering an exhaustive list of
object, or event. For instance, the robot may express agreement design parameters or options, this framework is intended to inspire
or disagreement through motions such as nodding or shaking its and guide further exploration and idea generation.
head. It can also convey attitudes or confidence in response to an
instruction or its own action by varying its movement profile—for
example, pausing to indicate hesitation or moving quickly and 3.3 Interaction Scenarios
decisively to show confidence. By design, the primary function of the lamp robot is to illuminate
spaces and support user activities. Equipped with a projector, the
Emotion. While robots do not experience emotions as humans do,
robot can extend this functionality by creating in-situ projections
their ability to simulate emotional expressions is crucial for creating
on walls, desks, and other surfaces. This capability enables the ro-
intuitive, engaging interactions. For instance, a robot might use
bot to project assistive information—either at the user’s request or
light, bouncy movements to convey happiness, slow movements to
proactively—to remind or support ongoing activities. For instance,
suggest a relaxed state, lower its head to indicate sadness, or employ
it could project a tutorial video to guide a task or display a creative
sudden, jerky motions to signal fear or other negative emotions.
drawing for inspiration. By projecting onto objects in the envi-
3.2.3 Building Blocks: Expressive Movement Design Primitives. Through ronment, the robot can also convey intention or offer contextual
a collaborative effort by animators and robot designers, we devel- information, such as displaying a water icon near a plant to remind
oped a design space with several primitives for creating expressive the user to water it.
motions, as illustrated in Figure 5. Drawing inspiration from human We envision the lamp robot engaging in both social- and function-
and animal nonverbal behaviors, we designed motions to express oriented tasks. Figure 6 illustrates some design outcomes of inter-
intention, attention, attitudes, and emotions. Similar to humans, action scenarios and task designs through iterative storyboarding
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Figure 6: Illustration of at-home interaction scenarios, organized by the robot’s agency (proactive vs. reactive) and task context
(function-oriented vs. social-oriented).

and video prototyping. On the x-axis, we consider the primary goal skills according to the task requirements. A high-level task man-
of the human-robot interaction. In function-oriented tasks, the ager interprets the lamp’s initial placement, the environment, and
lamp robot serves as an assistant or tool—providing information contextual information to determine and activate the appropriate
displays, offering desired lighting for user activities, adapting bed- state spaces during initialization. Figure 4 (right) illustrates the var-
time lighting, and reminding users of schedules or activities. In ious modalities the robot may respond to, including user activities
contrast, social-oriented tasks position the lamp robot more like and instructive gestures, speech commands, and touch interactions.
a friend or pet, emphasizing companionship and entertainment. The robot leverages torque sensing in its joints and can potentially
Examples include suggesting creative ideas, introducing the room integrate touch sensors on its surface, enabling it to detect tactile
to visitors, engaging in playful social interactions, playing music, input and switch to compliant modes when needed.
and projecting atmospheric lighting to enhance the overall user Through the iterative design process, we selected six task scenar-
experience. ios for further implementation of function-driven and expression-
The second dimension (y-axis) reflects the robot’s agency in driven robot movements for a user study. This selection covers all
human-robot interaction, distinguishing between proactive and four sectors of the representative space, comprising three function-
reactive roles depending on the task. In robot-proactive tasks, the oriented and three social-oriented tasks, as detailed in Section 4.2.
robot initiates the interaction. Examples include sending reminders,
nudging the user to build habits, or offering creative suggestions. In
robot-reactive tasks, the robot responds to user requests or actions. 4 User Study
For instance, in a photography lighting task, the robot activates Our research question is whether movements driven by expres-
the light based on the user’s verbal instructions and adjusts its sive utility can enhance users’ perceptions of the robot and their
position in response to pointing gestures. Similarly, a sleep light experience in human-robot interaction. To investigate this, we
might switch on or off in response to the user’s movements or verbal compare two robot conditions: one employs only function-driven
commands—activating a nightlight when asked or upon detecting movements (𝛾 = 0 in Equation 3), while the other incorporates
that the user has gotten out of bed. expression-driven movements in addition to function-driven ones,
To accommodate a wide range of tasks, the robot employs mul- achieving the same goal states but through different trajectories
tiple modalities and activates different input/output channels and (𝛾 > 0). Our objective is to determine whether—and to what ex-
tent—incorporating expression design into the robot’s movements
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

influences user interaction outcomes, and how these effects may the limit, and look back at the user and shake head before sending
vary according to the context of the tasks. a verbal reaction.
Remind Water. Robot interrupts use activity and sends out a
4.1 Research Questions and Hypothesis reminder to drink a cup of water. F: Move to point toward the water
RQ1: To what extent does adding expression-driven movements, cup, light up, and send a verbal reminder; E: Move to the goal pose
in addition to function-driven movements, influence users’ percep- described in F, push the cup toward the user, and gaze toward the
tions of the robot? user before sending the verbal reminder.
H1: Users will perceive a robot that combines expression-driven
and function-driven movements as more engaging, human-like, Social Conversation. Robot takes the role as a social companion,
and intelligent than one solely incorporates function-driven move- engages with the user in a social conversation about daily activ-
ments. ities. F: Respond to user’s speech verbally; E: Use movements as
RQ2: Does the task context affect movement preferences? nonverbal cues in accordance with verbal texts, including gazing
H2: Users’ perceptions will vary by task context—expression- at the user, pointing to refer to the object in context of speech, use
driven behaviors will be less favorable for function-oriented tasks kinesthetic gestures to display emotions of excitement (dancing
and more favorable for social-oriented tasks. movement) and sadness (lowering the head).
Play Music. Robot plays music entertainment accompanying
4.2 Method user’s daily activities. F: Play music with no movement; E: Play mu-
We used a within-subject study design in which each participant sic while perform dance movements, align the movement rhythm
viewed videos of the robot completing six different tasks, presented with music tempo.
in a randomized order. After watching each video, participants
rated their perception of the robot and its interaction with the 4.3 Measure
human shown in the video. They were also encouraged to explain We include six dimension of quantitative metrics to measure the
the reasoning behind their ratings, providing insights into which perception toward the robot (human-likeliness, perceived intelli-
specific robot behaviors influenced their preferences. gence, perceived emotion/character), the quality of the interaction
To create the video demonstrations, a team of human-robot in- (interaction engagement, the sense of connection), and willingness
teraction researchers and animation designers iteratively designed to use the robot in real life. Specifically, participants rate their per-
and refined pre-recorded robot movement trajectories using the de- ception on a scale from 0 to 100 to indicate their agreement to
sign primitatives proposed in section 3.2.3. These trajectories were six statements, measuring the above-mentioned aspects. Besides,
then implemented using the off-the-shelf WidowX arm controller we collected demographic data of participants including their gen-
to ensure smooth interactions. The videos used in this study are der, age, background regarding robotics, background regarding
included in the supplementary materials. expression design (such as performing arts, psychology, animation,
We designed and implemented six scenarios, each presented in communication), general level of empathy (“I find it easy to express
two conditions: empathy and understanding towards others.), general acceptance
F: A robot with function-driven movements only. toward robot (“I feel comfortable interacting with a robotic compan-
E: A robot with both function-driven and expression-driven ion”). After each video, we collect qualitative feedback of the video
movements. by asking “how would you describe the robot in the video? What do
Details of the six task scenarios and robot movement description you like or dislike about the robot?” This allow us to gather insights
are provided below. on participants reasoning of their choices and explore open-ended
Photograph Light. Robot responds to user’s hand gestures to ideas on the perception that we did not cover in the quantitative
move and offer desired lighting conditions for photography. F: Move metrics.
in response to user gesture and object position; E: Move to express
the curiosity toward the object by leaning forward, movements 4.4 Participants
incorporating robot’s attention to user command by looking back We recruited 30 participants using emails and announcements dis-
toward the user when detecting an instructive gesture. tributed within the organization1 . Responses were filtered based
on the time taken to complete the task, excluding those that took
Project Assistance. Robot observes user task and provides a cor- less than ten minutes, as well as any incomplete responses. This
responding video projection to guide the task. F: Move to a tar- process resulted in 21 valid participants (𝑁 = 21). Among them,
get position for projection, and project a corresponding video; E: eight are female, twelve are male, and one participant did not dis-
Show curiosity toward the user activity and display joint attention close their gender. The participants’ ages range from 26 to 51 years.
through gaze direction. In terms of ethnicity, ten participants self-identified as Asian, nine
self-identified as White or Caucasian, and two preferred not to
Failure Indication. User instructs a goal position for the robot
disclose their ethnicity.
which is out of reach, the robot displays the error message back to
the user. F: Attempt to move toward the goal direction, reach the 1 The study is exempt under the organization’s Human Study Review Board criteria.
limit, and verbally output the error; E: Pause to display hesitation This study fits under the research involving benign behavioral interventions and
before moving, stretch the body to display efforts when reaching collection of information from adults with their agreement (CFR 46. 104 (d) (3))
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Figure 7: Quantitative Results: Comparing perception ratings between expression-driven (blue) and function-driven (pink)
robot movements across six different task scenarios.

5 Results
This section presents the results through both quantitative and
qualitative lens to uncover the perception differences between the
two robot conditions across different tasks.

5.1 Quantitative Results

To test H1, we compared the average ratings across different met-
rics between the two robot conditions, averaged across different
tasks. The robots with expression-driven movements are rated
much higher (𝑀 = 56.16, 𝑠𝑡𝑑 = 27.15) than robots incorporat- Figure 8: P-values of T-tests on user perception scores com-
ing only function-driven movements (𝑀 = 28.77, 𝑠𝑡𝑑 = 27.15). paring expression-driven and function-driven movements.
Welch’s t-test revealed a statistical significance in the difference,
𝑡 = 19.85, 𝑝 < 0.0001. The biggest difference lies in the metrics
of Perceived character (𝑡 = 10.58), followed by Human-likeliness perceived significantly better compared to function-oriented
(𝑡 = 9.32), Engagement (𝑡 = 8.80), Sense of Connection (𝑡 = 8.50), tasks (toward the left side).
then Willingness to Interact (𝑡 = 7.37), and Perceived Intelligence To further investigate these differences, we conducted statisti-
(𝑡 = 5.22), 𝑝 < 0.001 for all of the individual metric, which indicate cal tests (Welch’s t-test) for each task and metric to compare the
statistical significance in the differences. Thus, H1 is supported. expression-driven and function-driven robot movements. The re-
Figure 7 depicts the average ratings (from 0 to 100, the higher, sulting p-values are displayed in Figure 8, where the dark color
the better) for expression-driven robots (blue) and function-driven indicates statistical significance (𝑝 < 0.05). The table shows that
robots (pink) across various tasks and evaluation metrics. The x- for social-oriented tasks (play music, conversation, remind water),
axis represents different tasks, arranged based on the purpose of the expression-driven robots significantly outperform function-driven
task—ranging from function-oriented tasks, such as photography robots across all metrics. However, for function-oriented tasks (pho-
lighting, projecting information, or displaying error messages, to tograph light, project assistance, failure indication), the two robots
socially-oriented tasks, such as music entertainment, social conver- show no significant differences in metrics such as Perceived in-
sation, and habit nudging. The results reveal that expression-driven telligence, Willingness to interact, and Engagement. Thus, H2 is
robots (blue) outperform function-driven robots (pink) across most supported.
of the tasks. The trend indicates that for social-oriented tasks To understand the effect of participants’ backgrounds on their
(toward the right side of the x-axis), expression-driven robots are perception of robots, we conducted a linear regression analysis to
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Figure 9: Effect of participants’ demographics and backgrounds on average perception ratings

examine the correlations between perception metrics (average rat- matching her energy, dancing bigger because she was dancing bigger.
ings of perception) and background variables, including gender, age, It makes me want to join. There is such power in the synergy!” (P12)
general level of empathy, general acceptance of robots, backgrounds In the qualitative reasoning, participants attributed the robot with
related to robotics and character design, as shown in Figure 9. Our expressive movements characteristics of human or pets, with its
findings indicate that age significantly influenced perceptions of ex- own drives and needs. Many participants said the expressive robot
pressive robots, with older participants showing less preference for remind them of a “puppy” or “child” . In the failure indication
expressive robots (𝑝 < 0.001). Additionally, we observed a trend in scenario, P12 described the expressive robot has “a resilient spirit”.
empathy levels affecting perception differences between functional In the playing music scenario, P1 said “it looks like it’s having fun.”
and expressive robots: participants with self-rated low empathy On the contrary, participants found it difficult to attribute human
perceived a stronger increase in robot likability after the integration characteristics in robots with little movement and described the
of expressive movements. In contrast, participants who self-rated robot as a “tool” or compared it to existing home devices. Partici-
as having high empathy were less influenced by the integration of pants commented the robot with only functional movements as
expressive movement in the robot. We also found a positive corre- “boring”, “too machine-like”, “not engaging”, “emotionless”, and may
lation between robot acceptance and perception scores. However, arouse negative feelings, especially during social conversations
these correlations did not reach statistical significance. Besides, we and playing music scenarios. P3 is confused about its motivation
conducted t-tests to compare the perception difference between of asking a social-oriented question during the conversation, as
gender groups, robotic backgrounds, and groups who have or do they assumed the robot did not have its own needs or emotions.
not have backgrounds related to character and expression design, P7 also commented that the robot looked “kinda creepy as if it
including animation, psychology, performing arts, etc. Gender did was intently staring” during the social conversation due to little
not have a significant impact on perception (𝑝 = 0.2). Robotic back- movements. Participants also noted the unnaturalness and lack
ground is a significant predictor of perception, with non-roboticist of social connection with only functional movements. During the
rating robots higher than roboticist (𝑝 = 0.006). Background related scenario of reminding human to drink water, P1 commented that
to expressive character design is another strong predictor, with “It [the functional robot] does not seem to care whether or not the
experienced character designers and artists rating robots signifi- human drinks the water.” P15 mentioned that without the robot
cantly lower than others. For all the groups above, they rated the “looking at the user like how humans engage with each other, there
expressive robots higher than the functional ones. was a lack of connection.”

5.2 Qualitative Results

We conducted a qualitative thematic analysis of participants’ feed- 5.2.2 Inference of Robot State. Participants perceived and dis-
back on individual robot behaviors to gain deeper insights into the cussed the robot’s intention, attention, attitudes and emotion be-
reasoning behind their ratings. We identify the perception reason- hind the expressive movements in the questionnaires. For instance,
ing behind expressive and functional movements, as well as the in the social conversation scenario, P1 mapped the robot’s move-
interaction with task context and other interaction modalities, as ment toward the window as if “checking the weather outside” [intention],
visualized in figure 10. and the moving around as excitement, hanging its head as sadness
[emotions]. During the projection task, P14 perceived the leaning
5.2.1 Perception of Robot Characteristics. Participants com- forward and tilting head movement as robot’s displaying curios-
mented the robot with expressive movements as more engaging, ity, saying “The robot first seems to be interested in the human task
lively, harmless, embodied a “sense of humor”, and were “fun to [attitude]. I liked that. It seems to be happy to help.” In the task of
watch”. P4 noted that the expressive robot showed more infor- photograph light, the expressive robot tilts the head back toward
mation of robot internal states, such as emotions, which weren’t the the human when the human gestures. Several participants were
apparent in function-driven only movements. Several participants able to perceive it as the robot paying attention to the human
said the expressive motions were affective throughout the inter- instruction. P7 said, “I liked when it looked at the person for feedback,
action of robot playing music and dancing: “... that appeared to be as if saying "is this good?" ”
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

Figure 10: Summary of qualitative results - clusters and implications.

Even for robots with only functional movements, some par- The preferences of adding expressive movements vary across
ticipants still perceive the robot’s movement with theory of mind, tasks. For tasks that include little functional movements, and for
such as projecting robot’s attention and intention. For example, tasks that were social-oriented and less sensitive to efficiency, ex-
during the task of failure indication, the robot stretches its arm pressive movements were more appreciated. For instance, in the
toward the note before displaying the failure message, participants scenario of playing music for entertainment, P21 noted that “I really
interpreted the stretch as “the robot seemed to struggle” (P6). P7 also liked this application for robot engagement! No fast responses were
noted that the robot’s facing direction clearly indicated its attention: necessary, so having an engaging dancing motion made the robot more
“the robot and the person are looking at the same note”. engaging. ” On the other hand, for the tasks that inherently have
clear function-driven movements and are more function-oriented,
adding expressive movements could be confusing to some individu-
5.2.3 What expressive movements were valued and what
als and preferences vary. For instance, in the scenario of photograph
were not. While adding expressive movements proved to benefit
light, participants thought the expressive movements made the ro-
the interactions, some participants found them unnecessary and
bot seem less “predictable”(P5), less “steady”(P14). P18 wished the
inefficient. P15 noted for the failure indication task “... there needs
robot to “stick to only the task relevant motion which is angle and
to be a balance between engagement through motion and speed com-
the light”. Even without expressive movements, in such function-
pletion of the task being given, otherwise the human might grow
oriented tasks, participants rated highly of the robots as long as the
impatient. It might be OK the first time with the novelty factor but
robot were able to move in correspondence to the context and user
will quickly fade out.” While some participants enjoyed the expres-
request.
sive motions, others noted that some expressive behaviors could be
It is important to note that many participants were less accept-
too exaggerated, thus distracting or disturbing. Some participants
able to robot taking proactive roles than reactive roles, such as
mentioned they disliked the robot to move all the time, especially
reminding the user to drink water. For instance, P20 noted that “...
the movement for no apparent reasons, which may imply “a lack
I don’t like my life to be controlled by a robot. If I’m in the middle
of attention on the robot’s part” (P5, conversation) Most participants
of some exciting readings, I don’t want to be disrupted by a robot’s
appreciated the information that were “quick and easy to inter-
command.” Adding the expressive movements such as with a play-
pret”, while for the subtle movements, participants had different
ful character could increase the acceptance of robot behavior, P8
preferences.
noted that “Without the playfulness, I might find this type of
Participants reported negative perception when there was a mis-
interaction with a robot annoying rather than welcome and
match between robot’s movement and its perceived capabilities.
engaging.”
For the failure indication function, P14 noted that “I did not like
that it tried to get momentum with an impulse as it seems fake.” P20 5.2.4 The effect of voice and light. Participants repeatedly com-
thought the robot did not have a camera on the head thus the mented on the alignment between movements and other modalities,
“looking at notes” action seemed fake. such as the robot’s sound and light. Several participants felt there
Conference DIS’25, July 05–09, 2025, Funchal, Madeira

was a mismatch between robot’s speech cadence and the ex- interaction, increases user engagement, and can even convey an
pressive motions - while the expressive motions were “endearing, additional layer of information. On the other hand, for tasks that
showing a character”, the tone from the robot was very “automated”, are function-driven—such as adjusting lighting angles or shifting
“stiff”, and “took away from how friendly the interaction felt”. P7 and between projection spaces—adding expressive motions can disrupt
P15 noted that the timing of the voice need to align with the the robot’s primary function and potentially cause confusion or an-
timing of the movement, to make it feel more natural. P12 found noyance for users. This implies that expression-driven movements
the sound from the motors disturbing, and may only “punctuate need to complement function-driven movements by adjusting both
with smaller slower movements”. the amount and timing of expressions to enrich—rather than con-
The coordination between movement and lighting can influence flict with—the original motions. Future research should balance the
the comfort of interaction. Some participants mentioned preferring trade-off between task efficiency and characterfulness in human-
the robot to remain steady while maintaining the light. In such robot interaction, while also considering individual preferences
cases, expressive movements might interfere with the primary light- through personalized behaviors. For instance, although some users
ing function, as the robot’s motion could distract when displaying enjoyed a more animated robot, others disliked constant movement,
attention or curiosity. Additionally, the proximity of the light also particularly when it occurred without a clear or explicit reason.
impacted perceptions of disruption, as noted by P21. P7 appreciated The design and integration of expressive motions also need to
that “the robot turned out its light when looking at the person” during align with the robot’s embodiment and capabilities. For instance,
assistive projection. gaze behaviors should be co-designed with the placement of the
robot’s “eyes” (cameras) and “head”. While this may be intuitive
for humanoid robots, robots with non-anthropomorphic features
6 Discussion rely on appearance design and movement patterns to suggest a life-
In this paper, we conducted design research to explore how adding like embodiment that can be intuitive to humans and even other
expressive movements on top of purely functional ones affects species. It is equally important to match these movements with
human-robot interactions. Both quantitative and qualitative find- the robot’s other modalities—in this case, its voice and the lamp’s
ings show that expressive movements, compared to strictly func- light or display. As many participants noted, the speech content,
tional ones, enhance the overall interaction experience and improve tone, and timing during movement sequences all play a key role
perceived robot qualities. Participants were more likely to recognize in shaping the perceived quality of the robot’s behaviors. Future
the robot’s state of mind, projecting intentions, attention, emotions, research needs to consider more extensive alignment among these
and attitudes throughout their interactions. For instance, partici- different modalities to further enhance human-robot interaction.
pants recognized the robot’s “gaze” as a marker of joint attention,
suggesting a stronger bond between human and robot. Addition-
ally, participants more frequently described the expressive robot 7 Conclusion
as a living being—for example, a “pet,” a “child,” or a “friend.” In In this paper, we present ELEGNT, a framework for designing
some tasks, adding expressive movements made the experience expressive and functional movements for non-anthropomorphic
more engaging and playful. In particular, when the robot initiated robots in daily interactions. The framework integrates function-
an interaction or nudged participants, its expressive movements driven and expression-driven utilities, where the former focuses
made those interruptions feel more acceptable, such as in the case on finding an optimal path to achieve a physical goal state, and
of interrupting the user during the reading and nudging the user the latter motivates the robot to take paths that convey its internal
to drink water. This may be due to that participants possess more states—such as intention, attention, attitude, and emotion—during
empathy towards the robot with expressive movements, as they human-robot interactions. We use a lamp-shaped robot to illus-
remind them of living beings, just like the pets making the mess trate the design space for functional and expressive movements in
in the home; and thus the initially disturbing behaviors transfer various interaction scenarios, ranging from function-oriented to
into a playful social interaction. This highlighted the benefit of social-oriented tasks, and involving reactive versus proactive robot
adding expressive and characterful motions for robot-initiated task roles. We conduct a user study to compare perceptions of the robot
scenarios. when using expressive movements versus only functional move-
The quantitative results reveal differences in perception across ments across six different task scenarios. Our results indicate that
various tasks. For social-oriented tasks—such as playing music, incorporating expressive movements significantly increases user
engaging in social conversations, and nudging water—adding ex- likability toward the robot and enhances interaction engagement.
pressive movements was significantly more preferred. The qualita- The perception varies across tasks, with social-oriented tasks that
tive reasoning further illuminates this trend: in these tasks, users require minimal function-driven movements benefiting particularly
prioritize engagement and entertainment over task efficiency. Con- from the addition of expression-driven movements. Qualitative
sequently, adding expressive movements enhances the robot’s play- analysis further elaborates on users’ differing perceptions of the
fulness and character. Moreover, social-oriented tasks in the study robot’s characteristics and the perceived robots’ mental models.
generally entail minimal function-driven motions. For instance, The findings also highlight the importance of aligning movement
when playing music or holding social conversations, the robot pri- with other robot modalities, such as voice and light. Future work
marily responds verbally, and the functional output does not involve will integrate these design insights into a generative framework for
any physical movement. In these contexts, incorporating expressive creating context-aware robotic movements that effectively express
movements aligned with the social and task setting enriches the intentions in non-anthropomorphic robots.
ELEGNT: Expressive and Functional Movement Design for Non-anthropomorphic Robot Conference DIS’25, July 05–09, 2025, Funchal, Madeira

References conference on intelligent robots and systems, Vol. 3. IEEE, 2431–2436.

[1] Ahmad Abu-Akel and Simone Shamay-Tsoory. 2011. Neuroanatomical and [27] Amy Koike and Bilge Mutlu. 2023. Exploring the Design Space of Extra-Linguistic
neurochemical bases of theory of mind. Neuropsychologia 49, 11 (2011), 2971– Expression for Robots. In Proceedings of the 2023 ACM Designing Interactive
2984. Systems Conference. 2689–2706.
[2] Patrícia Alves-Oliveira, Maria Luce Lupetti, Michal Luria, Diana Löffler, Mafalda [28] Hideki Kozima, Marek P Michalowski, and Cocoro Nakagawa. 2009. Keepon: A
Gamboa, Lea Albaugh, Waki Kamino, Anastasia K. Ostrowski, David Puljiz, Pedro playful robot for research, therapy, and entertainment. International Journal of
Reynolds-Cuéllar, et al. 2021. Collection of metaphors for human-robot interac- social robotics 1 (2009), 3–18.
tion. In Proceedings of the 2021 ACM Designing Interactive Systems Conference. [29] Lynn T Kozlowski and James E Cutting. 1977. Recognizing the sex of a walker
1366–1379. from a dynamic point-light display. Perception & psychophysics 21 (1977), 575–
[3] Lucy Anderson-Bashan, Benny Megidish, Hadas Erel, Iddo Wald, Guy Hoffman, 580.
Oren Zuckerman, and Andrey Grishko. 2018. The greeting machine: an abstract [30] Diana Löffler, Nina Schmidt, and Robert Tscharn. 2018. Multimodal expression of
robotic object for opening encounters. In 2018 27th IEEE International Symposium artificial emotion in social robots using color, motion and sound. In Proceedings
on Robot and Human Interactive Communication (RO-MAN). IEEE, 595–602. of the 2018 ACM/IEEE International Conference on Human-Robot Interaction. 334–
[4] Michael Argyle. 2013. Bodily communication. Routledge. 343.
[5] Simon Baron-Cohen. 1991. Precursors to a theory of mind: Understanding atten- [31] Lian Loke and Toni Robertson. 2013. Moving and making strange: An embodied
tion in others. Natural theories of mind: Evolution, development and simulation of approach to movement-based interaction design. ACM Transactions on Computer-
everyday mindreading 1, 233-251 (1991), 1. Human Interaction (TOCHI) 20, 1 (2013), 1–25.
[6] Christoph Bartneck and Jodi Forlizzi. 2004. Shaping human-robot interaction: un- [32] Michal Luria, Guy Hoffman, Benny Megidish, Oren Zuckerman, and Sung Park.
derstanding the social aspects of intelligent robotic products. In CHI’04 Extended 2016. Designing Vyo, a robotic Smart Home assistant: Bridging the gap between
Abstracts on Human Factors in Computing Systems. 1731–1732. device and social agent. In 2016 25th IEEE International Symposium on Robot and
[7] Mason Bretan, Guy Hoffman, and Gil Weinberg. 2015. Emotionally expressive Human Interactive Communication (RO-MAN). IEEE, 1019–1025.
dynamic physical behaviors in robots. International Journal of Human-Computer [33] Michal Luria, Guy Hoffman, and Oren Zuckerman. 2017. Comparing social robot,
Studies 78 (2015), 1–16. screen and voice interfaces for smart-home control. In Proceedings of the 2017
[8] Paul Bucci, Xi Laura Cang, Anasazi Valair, David Marino, Lucia Tseng, Merel CHI conference on human factors in computing systems. 580–628.
Jung, Jussi Rantala, Oliver S Schneider, and Karon E MacLean. 2017. Sketching [34] Toru Nakata, Tomomasa Sato, Taketoshi Mori, and Hiroshi Mizoguchi. 1998.
cuddlebits: coupled prototyping of body and behaviour for an affective robot Expression of emotion and intention by robot body movement. In Proceedings of
pet. In Proceedings of the 2017 CHI Conference on Human Factors in Computing the 5th international conference on autonomous systems.
Systems. 3681–3692. [35] Pranav Rane, Varun Mhatre, and Lakshmi Kurup. 2014. Study of a home robot:
[9] Nazli Cila, Cristina Zaga, and Maria Luce Lupetti. 2021. Learning from robotic Jibo. International journal of engineering research and technology 3, 10 (2014),
artefacts: A quest for strong concepts in Human-Robot Interaction. In Proceedings 490–493.
of the 2021 ACM Designing Interactive Systems Conference. 1356–1365. [36] Trossen Robotics. [n. d.]. WidowX 250S. https://www.trossenrobotics.com/
[10] Herbert H Clark. 2005. Coordinating with each other in a material world. Dis- widowx-250
course studies 7, 4-5 (2005), 507–525. [37] Leila Takayama, Doug Dooley, and Wendy Ju. 2011. Expressing thought: im-
[11] Brian R Duffy. 2003. Anthropomorphism and the social robot. Robotics and proving robot readability with animation principles. In Proceedings of the 6th
autonomous systems 42, 3-4 (2003), 177–190. international conference on Human-robot interaction. 69–76.
[12] Masahiro Fujita. 2004. On activating human communications with pet-type robot [38] Ben Tribelhorn and Zachary Dodds. 2007. Evaluating the Roomba: A low-cost,
AIBO. Proc. IEEE 92, 11 (2004), 1804–1813. ubiquitous platform for robotics research and education. In Proceedings 2007 IEEE
[13] William Gaver. 2012. What should we expect from research through design?. International Conference on Robotics and Automation. IEEE, 1393–1399.
In Proceedings of the SIGCHI conference on human factors in computing systems.
937–946.
[14] Petra Gemeinboeck and Rob Saunders. 2017. Movement matters: How a robot
becomes body. In Proceedings of the 4th international conference on movement
Computing. 1–8.
[15] Jennifer Goetz, Sara Kiesler, and Aaron Powers. 2003. Matching robot appearance
and behavior to tasks to improve human-robot cooperation. In The 12th IEEE
International Workshop on Robot and Human Interactive Communication, 2003.
Proceedings. ROMAN 2003. Ieee, 55–60.
[16] Edward T Hall. 1966. The hidden dimension. Garden City (1966).
[17] Guy Hoffman and Wendy Ju. 2014. Designing robots with movement in mind.
Journal of Human-Robot Interaction 3, 1 (2014), 91–122.
[18] Guy Hoffman and Gil Weinberg. 2010. Shimon: an interactive improvisational
robotic marimba player. In CHI’10 Extended Abstracts on Human Factors in
Computing Systems. 3097–3102.
[19] Yuhan Hu and Guy Hoffman. 2019. Using skin texture change to design emotion
expression in social robots. In 2019 14th ACM/IEEE International Conference on
Human-Robot Interaction (HRI). IEEE, 2–10.
[20] Yuhan Hu and Guy Hoffman. 2023. What Can a Robot’s Skin Be? Designing
Texture-changing Skin for Human–Robot Social Interaction. ACM Transactions
on Human-Robot Interaction 12, 2 (2023), 1–19.
[21] Yuhan Hu, Jasmine Lu, Nathan Scinto-Madonich, Miguel Alfonso Pineros, Pedro
Lopes, and Guy Hoffman. 2024. Designing plant-driven actuators for robots
to grow, age, and decay. In Proceedings of the 2024 ACM Designing Interactive
Systems Conference. 2481–2496.
[22] Peide Huang, Yuhan Hu, Nataliya Nechyporenko, Daehwa Kim, Walter Talbott,
and Jian Zhang. 2024. EMOTION: Expressive Motion Sequence Generation for
Humanoid Robots with In-Context Learning. arXiv preprint arXiv:2410.23234
(2024).
[23] Lasseter J. 1986. Luxo Jr. http://www.pixar.com/short_films/Theatrical-Shorts/
Luxo-Jr
[24] C. Jones. 1965. The dot and the line. https://archive.org/details/thedotandtheline
[25] Peter H Kahn, Nathan G Freier, Takayuki Kanda, Hiroshi Ishiguro, Jolina H
Ruckert, Rachel L Severson, and Shaun K Kane. 2008. Design patterns for sociality
in human-robot interaction. In Proceedings of the 3rd ACM/IEEE international
conference on Human robot interaction. 97–104.
[26] Kenji Kaneko, Fumio Kanehiro, Shuuji Kajita, Kazuhiko Yokoyama, Kazuhiko
Akachi, Toshikazu Kawasaki, Shigehiko Ota, and Takakatsu Isozumi. 2002. Design
of prototype humanoid robotics platform for HRP. In IEEE/RSJ international