Overview

Background and Problem Introduction

Brief eye contact with strangers can foster connection, belonging, and positive affect, yet such moments are often scarce in public spaces.

Brief eye contact between strangers can bring about positive emotions: a sense of connection and belonging. However, such moments rarely occur in reality.

• Brief, positive eye contact is scarce but holds social value.

Research Objective

This paper investigates how a spatially situated robot can reshape the visual field of a shared space to influence how strangers notice and respond to one another.

The authors explore whether a robot can influence how strangers notice and respond to each other by altering the "visual relationships" within a shared space.

• visual field: what people can see in a shared space, whom they see, and how sightlines are organized.
• spatially situated robot: a robot placed within a specific spatial context, acting as a member of the space rather than an abstract algorithmic system.

Research Method / What the Authors Did

We present MirrorBot, a mobile robot equipped with two actuated mirrors that dynamically redirect reflections to reshape sightlines between people.

The authors developed a mobile robot equipped with two actuated mirrors that dynamically alter the optical paths of sightlines between people.

• sightlines: The meaning is line of sight, referring to what can be seen from a certain position in a space, who can be seen, and whether the line of sight is guided or blocked.
• Through this robot (controlling the orientation of the mirror), it changes how two strangers enter each other's field of vision.
• Manipulating visibility between people.

Research Process Design and Main Phenomenological Results

In a study with 32 strangers in 16 pairs in a waiting‑room setting, MirrorBot elicited patterns such as low-stakes icebreaking, nonverbal synchrony, joint sensemaking, asymmetric engagement, and avoidance.

• low-stakes icebreaking: Low-pressure icebreaking, very low-cost small interactions.
• nonverbal synchrony: Nonverbal synchrony, laughing together, matching movements, etc.
• joint sensemaking: Joint guessing/understanding, "What is this robot doing?".
• asymmetric engagement: Asymmetric engagement, one person is very engaged, the other is very cold.
• avoidance: Avoidance.
• This robot indeed elicited various patterns of stranger interactions, both positive and failed.
• The word 'elicit' is used very well, mainly to draw out certain patterns.

Further Refinement of Results

Participants also attributed multiple roles to the robot, such as mediator, observer, magnifier, or disrupter, revealing that its social meaning was fluid and co‑constructed.

Participants did not consistently understand the robot as a fixed identity.

• mediator: Something that helps both parties establish a connection.
• observer: Observer.
• magnifier: Amplifier, amplifying a certain feeling or attention.
• disrupter: Disrupter.
• Its social meaning is fluid, co-assigned by participants, not fixed.
• MirrorBot is not a social entity with a fixed role; its role is co-defined within the context.

Full Article Contribution Summary

Our work extends HRI by showing that robots can act not only as conversational partners but also as spatial mediators, curating opportunities for human–human connection through the reconfiguration of spatial relationships.

Robots don't necessarily have to be "objects that talk to people"; they can also serve as "spatial mediators." By rearranging spatial relationships, they create opportunities for connections between people.

• spatial mediators: spatial mediators.
• curating opportunities: planning, arranging, not forcing to happen.
• What this paper likely aims to leave behind is: robots can be "spatial choreographers" or "relationship mediators."

Overall, this paper focuses on a new approach to human-robot interaction: not direct interaction, but embedding robots into the interactive relationships between people. Robots influence human-to-human social interactions by affecting the social media between people. In applying social media, this robot specifically focuses on space and optics.

• Social media: too broad; more accurately, it's about reconstructing "visibility" and "line-of-sight relationships" between people.
• Regarding the focus: what the authors truly care about is not physical reflection itself, but how reflection changes social relationships.

Revised Summary

This paper explores a relatively new perspective in HRI. It's not concerned with how robots directly communicate with people, but whether robots can embed themselves into the conditions of interaction between people to indirectly influence social interactions among strangers. Specifically, the authors use robots with movable mirrors to alter line-of-sight relationships and mutual visibility in shared spaces, thereby creating lower-pressure mutual gaze and social starting points.

Why It's a New HRI Perspective

Traditionally, human-robot interaction has focused on robots' roles as conversational partners, social agents, task collaborators, companions, etc. However, in this paper, the authors do not approach the role of robots in human society from this angle.

How Exactly Do the Authors Address the Contradiction of "People Don't Naturally Interact and Don't Want to Be Forced to Interact"?

1. State between strangers in ordinary situations: Sitting close, theoretically able to interact. However, staring, directly initiating conversation, etc., can seem abrupt and intrusive. The author mentions maintaining a state of "civil inattention," roughly meaning "polite mutual non-interference." (Immersed in phones)
2. Author's task: To find low-pressure, testable, and exit-friendly ways of connecting.
3. Author's strategies:
   • Indirect eye contact.
   • Gradual escalation (corresponding to the design of four states → gradual warming up).
   • Built-in "right to exit" (design mechanisms for opt-out, etc.).
4. Summary: More like an invitation. Focuses more on connections that might have happened but didn't. Without being intrusive, creates an intermediate layer where strangers have the opportunity to notice and test each other, namely MirrorBot.

Introduction

Task of the introduction: The problem is worth researching, and the approach is reasonable. How does the author frame the problem?

1. First, focus on eye contact. It is the strongest non-verbal signal. Everything later revolves around this, as the author aims to emphasize "how to create indirect mutual gaze."
2. The author emphasizes that the research object is to give social sparks a chance to emerge.
3. Social fabric: There are very subtle yet important connective structures between people, and this is meaningful and valuable.
4. Contradiction: Although meaningful, such connections rarely occur in current public spaces. The author explores reasons: looking at phones, social norms...
   • Up to this point, tension is established: valuable connections are difficult to occur naturally.
5. Paradoxical environment: In some public settings, people are physically placed together but socially separated.
6. Theoretical approach: Spatial relationships. How spatial relationships affect social interactions, and in turn, how they are shaped by social interactions.
7. Whether strangers notice each other largely depends on spatial layout and line-of-sight relationships. However, much past HRI has focused more on the robot's own expressive behaviors rather than influencing the environment itself.
8. Why mirrors: Three characteristics of mirrors (design philosophy):
   1. Invites self-awareness: Makes people notice themselves first.
   2. Enables indirect observation: Allows indirect observation of others.
   3. Without the intensity and pressure of direct mutual gaze, it can gently connect people.
9. Formally posing the research question: What happens when mirrors are no longer fixed but movable, tiltable, and decide who sees whom? That is, how can robot-driven mirrors create and shape such moments among strangers?

Why does the author think this problem is worth researching? Why choose this approach?

In real-world social interactions among strangers, there exists a paradox: physical proximity provides the premise for social engagement, yet it is constrained by certain factors (such as "social norms"), leading to the suppression of potential interactions. Public spaces bring strangers together, but the polite avoidance of disturbance and the pressure of direct eye contact make it difficult for low-threshold mutual attention and lightweight interactions, which could otherwise occur naturally, to emerge. These social moments, though small, hold significant value.

The author does not aim to completely resolve the tension in stranger interactions, but rather seeks a low-pressure, exit-friendly mediating mechanism that makes it easier for certain mutual attention and interactions, which would otherwise be hard to initiate, to occur.

The author believes that spatial relationships are a promising entry point, focusing specifically on sightlines and visibility. This is because they influence human social behavior while being tangibly present. Specifically, using mirrors as an approach is advantageous because mirrors can better leverage spatial relationships and help robots alter the environment, aligning with the design philosophy.

The key reason for choosing mirrors is not merely their ability to participate in spatial reconfiguration, but that they allow individuals to start with self-reflection and then, through indirect reflection, probe the presence and reactions of others, thereby reducing the social pressure associated with direct eye contact. In other words, the author attempts to use robot interaction to manipulate spatial relationships in an effort to address this paradox.

Interim Summary

1. What are the two layers of the paper's contribution?
   • An attempt to alleviate the social "paradox."
   • A new understanding of human-robot interaction: robots can interact with humans not directly, but by influencing the environment.
2. Refined answer:
   • Proposes and preliminarily tests a design pathway that uses mediated mutual gaze to facilitate low-pressure interactions among strangers.
   • Offers a more concrete new perspective for HRI: robots are not only direct interaction agents but can also serve as spatial mediators, indirectly shaping the conditions for human-human interactions by reconfiguring sightlines and proxemic relationships in shared spaces.

How Exactly Is It Done

Mirror Design

The author experimented with five mirror configurations and ultimately found that what they truly needed was not a flashy reflective device, but a structure that could clearly, smoothly, and with low pressure organize the three relationships: "self / other / both appearing simultaneously."

Excluded Misguided Directions

1. Back-to-back mirrors?
   • Reflections are not easily caught in peripheral vision; there is no clear sense of relationship between the two mirrors, making them more suitable for individual mirror-gazing.
2. Small mirror embedded in a large mirror
   • This would make "I check myself" overshadow "I notice others," pulling the focus back to "looking at oneself."
3. Mirrors that open and close like a book
   • The author explored this idea to see if the opening and closing of mirrors could itself become an action that reveals relationships. However, this design has poor predictability, and the motion itself steals the spotlight.
4. Intersecting mirror strips
   • This would fragment the face, which is quite problematic.
5. Mirrors distributed at random angles in space
   • Randomly angled mirrors feel too much like a complex device, akin to "guess what I can reflect," rather than "it's making me and another person form some kind of relationship."

The Right Direction

1. Parallel configuration, like eyes.
2. Equal scale, ensuring both individuals appear with similar size and status in the mirror.
3. Smooth transitions.

Excluding designs that pull users back into self-inspection, blur relational intent, or disrupt face/gaze continuity, the approach converges on a mirror organization that allows smooth switching between self and other with equal scale.

Four Interaction States

Four states constitute the social choreography, progressing from curiosity and self-awareness to moments of mutual recognition.

Voids

First, let the mirrors "exist" without establishing a relationship — a low-stimulus opening.

In the Voids state, participants see only the surrounding environment in the mirror; this state is primarily used during MirrorBot navigation or when gently swaying the mirror to attract attention.

1. Establish the sense of the device before the sense of relationship.
2. Low-salience probing, with very slight mirror nods or sways.
3. Leave room for an "exit right," not escalating unless clear reciprocal cues are detected.
4. Spark curiosity.

Dual Self

First, draw "me" in.

In the Dual Self state, one participant sees their own reflection in both mirrors, while the other still mainly sees the environment. The robot alternates this state, allowing both individuals to first notice their own reflections and thereby be invited into the interaction.

1. First see oneself, quickly drawing the person into the interactive system. The author found during revisions that this step is essential, serving as a "pressure relief layer."
2. Why see oneself in both mirrors? To prevent ambiguity, clearly indicating that "I" am establishing a relationship with the mirror, enhancing readability.
3. Only draw one person at a time, more gently, avoiding an instant increase in social pressure, first establishing a connection between the individual and the system.
4. Build self-awareness; users will understand that "this system organizes visual relationships."
5. Potential risk: causing users to focus too much on themselves, so it must be a transitional state.

Dual Other

The arrival of "the other person."

In the Dual Other state, two people will see each other simultaneously in both mirrors. This creates a mutual awareness, even a simulated eye contact.

1. The other person enters.
2. Why see the other in both mirrors? To be as clear as possible, avoiding ambiguity, ensuring the current relational focus is "the other."
3. This is already closest to the state of "strangers beginning to acknowledge each other," introducing some social pressure.

Self + Other

Simultaneously handling the relationship between "me" and "you."

In the Self + Other state, each participant sees themselves in one mirror and the other person in the other mirror. Moreover, the system alternates this configuration between the two people, ensuring both have equal opportunities to experience it. The author states this creates a balanced, hybrid view.

1. To some extent, it reduces social pressure, providing half self-space.
2. This state alternates between participants to ensure both have equal opportunities.
3. It satisfies self-awareness, observation of others, a degree of mutuality, and avoids direct face-to-face social pressure.

Implied critique: Users need to understand the mirror logic, placing high demands on readability.

Simulation

Assume two people: t and z.

1. Initial State: t and z are in a shared space that is physically together but socially separated.
2. Approach: The robot approaches, but only as an environmental/installation presence.
3. Attraction: Corresponds to the Voids state (mirror subtly nods, sways). It doesn't matter if t and z notice or not; if they do, they only observe slight changes in the mirror's reflection of the environment.
4. Increasing Curiosity: The robot gradually reveals the reflection, and both t and z enter Dual Self mode. At this point, t and z may realize the mirror has some directionality and prepare themselves accordingly.
5. Establishing Connection: t and z enter Dual Other mode, and if the response is appropriate, they transition to Self + Other, forming a connection.
6. Exchange/Continue/Withdraw: t and z may continue with light interaction (the robot lingers longer) or directly communicate (the robot withdraws); if t and z fail to engage in light interaction (one or both do not respond), the robot attempts a few times and withdraws if unsuccessful.

Figure 4 illustrates a progressive social intervention path: the robot first enters the space, then uses low-stimulus methods to attract attention, followed by allowing participants to see themselves separately, then each other, and finally entering a mediated mutual gaze where self and other are juxtaposed. If the person accepts the invitation, the interaction is briefly extended; if the person is already naturally communicating or simply doesn't want to interact, the robot withdraws.

Qualitative Laboratory Study

Describes how the authors transformed the earlier design into a research scenario capable of producing credible observations. MirrorBot serves as a design probe, aiming to extract descriptive, phenomenological insights into how people notice, interpret, and respond to such visual and social changes.

• design probe: A probe used by the authors to observe what social phenomena emerge in reality (how do people understand it? What interactions are elicited? Invitation or intrusion?).
• cover story: Employed a fake task to make reactions more natural.
• Strangers.
• Small sample size, primarily the authors observed some possible reactions (recurring patterns).
• Campus sample implies "limited ecological validity."

Process

• Stage 1: Enhancing credibility with dummy tasks.
• Stage 2: Core observation, allowing free sitting, standing, chatting, and resting. The robot enters one minute later. The experiment follows a "first half standardized + second half adaptive observation" design. However, due to researchers manually controlling it from behind, the system's scalability and automation capabilities were not genuinely demonstrated in this paper.
• Stage 3: Conducting semi-structured interviews separately.
• Multiple people watched videos + thematic analysis of interviews, ultimately summarizing the findings.

Key Points

1. Phenomenon exploration experiment, not an effectiveness verification experiment.
2. Simultaneously focusing on "behavior" + "subjective interpretation."
3. Drawbacks: Limited sample representativeness, limited spatial representativeness, automation not tested, etc.

Findings

MirrorBot does not create social interaction but regulates the threshold for initiating social interaction (lowering it).

Catching First Glances

Most groups' first genuinely socially meaningful contact occurred in the mirror. That is, the mirror indeed changed the rules of gaze management. What MirrorBot first brings is a lower-pressure mutual attention mechanism.

Testing the Waters

After noticing each other, they didn't immediately start chatting but began testing with body/movements: body calibration (If I move like this, how will the system respond? How will the other person respond?); and non-verbal synchronization: not only can it promote social relationships but also gauge willingness. What MirrorBot creates is not direct social outcomes; what emerges first is an interactive space that can be jointly adjusted.

Speculating Together

A connection already exists: joint speculation. Here, ambiguity can also serve as a design resource.

Asymmetric Engagement

This is a case of failure: it can only amplify, not create (willingness) from scratch.

Discomfort, Avoidance

Invitations must not become disturbances. Careful design is a core design condition.

Fading into the Ambience

The robot does not necessarily continuously dominate the interaction. Sometimes it naturally withdraws; other times, it means the connection itself did not develop.

Roles Attributed to MirrorBot

Different people perceive it as different things:

• mediator, intermediary.
• peripheral presence, peripheral existence.
• quiet listener, silent listener.
• pet‑like / childlike, resembling a pet or child.
• magnifier of awareness, something that amplifies mutual awareness.
• disruptor, disturber.

Indeed, the social significance of MirrorBot is not fixed but is co-constructed through interaction.

Implicit critique: joint sensemaking—how much comes from the mirror structure itself, and how much from "there's a weird robot appearing"?

Discussion

Rather than the robot itself, the authors likely aim to discuss "how should we reinterpret the position of robots in human society."

What exactly has the robot's role become?

Participants did not provide a unified understanding. However, this paper challenges a default premise: that robots must rely on expressions, speech, simulated social behaviors, etc., to gain social significance. Robots can also produce social effects by altering the conditions under which interactions occur.

Here, the authors also mention ambiguity as a design resource, meaning that appropriate ambiguity leaves room for interpretation, prompting people to co-construct meaning.

Amplifying possibilities rather than creating intentions out of thin air

I think sometimes this might also generate some intention?

Design must carefully consider exit options. And it imposes requirements on space; for example, hospital waiting areas, spaces with high cognitive load, or fragile states can be dangerous.

Space as an intermediary

The robot actively restructures spatial relationships, thereby influencing how people see, approach, and notice each other.

Shifting from a bilateral "human-robot" relationship to "the robot pulling space itself into the interaction mechanism." Regarding space: sightlines, proximity, visibility specifically refer to the reorganized visibility between people.

Limitations

1. The scenario is too controlled.
2. Only one spatial layout was observed, over too short a time.
3. The system has not been proven reliable for automation.

Strong in raising new questions and demonstrating phenomena, but not well-performing in ecological extrapolation or automation implementation.

Personal understanding

What problem is this paper actually studying?

This paper studies a new robot interaction method: by influencing spatial (sightline) structures to attempt eliciting human social interactions under certain spatial conditions (physical proximity, social isolation scenarios).

Why is this problem worth studying?

First, this type of robot interaction has been rarely and superficially studied before; second, the scenario of physical proximity and social isolation inherently presents a contradiction, which is not conducive to promoting human social interaction, so researching this issue can provide some inspiration for alleviating this contradiction; third, this robot interaction approach may offer entirely new perspectives for studying robot interactions.

What is the core mechanism proposed by the authors? How does MirrorBot actually work?

The core mechanism is to use gaze relationships to guide human social interaction. How it works:

1. By using motors to drive the mirror, the mirror alters gaze relationships.
2. Four states are designed to progressively guide people into social interaction with others.
3. Emphasizes social pressure: low pressure → high pressure → reduced pressure.
4. Strongly emphasizes an exit mechanism, with a focus on social boundaries throughout the process.

How was the experiment conducted, and why was it designed this way?

Specifically, the experiment recruited a small sample. The sample size was small, the setting was singular, and the duration was short. It emphasized that all participants were strangers. The author team used a cover story to ensure more natural reactions and more accurate findings.

The author team carefully designed the timing of people and robot entry, their permissions, etc. They utilized video recordings and semi-structured interviews to organize and summarize findings.

What key phenomena did the authors observe?

The mirror indeed served as an initial connection point for many people. After the connection began, it wasn't just about chatting; instead, people first tested the waters with body language/actions/non-verbal cues, using appropriately ambiguous intentions to allow many to build shared assumptions. Sometimes it failed—an invitation should not become a disturbance. The robot did not necessarily remain continuously involved in the interaction. The robot's role was established on shared understanding and was not uniform.

Paper Contributions

1. Contributed a new robot interaction approach and perspective: influencing human social interaction by altering visual relationships in space, making the robot a choreographer of relationships.
2. Provided some inspiration for alleviating the contradiction between physical proximity and social isolation.

Below are optimizations:

One-sentence summary This paper investigates whether a robot can use movable mirrors to reconfigure gaze relationships in shared spaces, creating a low-pressure, tentative, and opt-out mutual recognition in scenarios like waiting rooms, where strangers are physically co-present but socially separated.

What the paper truly aims to solve The authors identify a social tension in public spaces: strangers sit close but, due to phones, polite non-interference, and the pressure of direct eye contact, light interactions that might naturally occur rarely happen. They argue these micro-encounters hold genuine social value, making them worth studying.

What is the core mechanism The authors propose MirrorBot, a mobile robot with two movable mirrors. Its key innovation is not just that the mirrors move, but that it dynamically orchestrates "who sees whom and when." The system gradually alters visual relationships between people through four states: first Voids, showing only the environment; then Dual Self, letting each person see themselves; followed by Dual Other, allowing two people to see each other; and finally Self + Other, where each sees both themselves and the other. The design emphasizes making social pressure incremental, rather than forcing strangers into direct eye contact immediately. It also includes opt-out logic to prevent invitations from becoming intrusions.

How the experiment was conducted The authors conducted a qualitative lab study, using MirrorBot as a design probe to observe phenomena rather than for rigorous validation. They recruited 16 pairs of strangers, using a memory task as a cover story in a waiting room setting, where two people first entered a normal waiting state before MirrorBot intervened. Data primarily came from video recordings and post-study semi-structured interviews, allowing the authors to observe participants' behaviors and understand how they interpreted the experience.

What the authors observed MirrorBot often enabled strangers' first socially meaningful contact to occur in the mirror, not face-to-face. Subsequently, interactions could evolve into non-verbal synchrony, such as tilting heads together, laughing together, or testing the mirror's responses; or into joint sensemaking, where the pair speculated about the robot's intentions. However, the mechanism was not always effective. Sometimes only one person engaged while the other remained indifferent; some felt stared at or disturbed and avoided interaction; and in some groups, after the novelty wore off, the robot faded into the background. Participants also had varied interpretations of the robot's role: some saw it as a mediator, others as an observer, pet, amplifier, or disruptor.

Contributions of this paper The paper's first contribution is proposing and demonstrating MirrorBot as a research prototype, using "movable mirrors + dynamic gaze orchestration" to mediate mutual gaze. The second contribution is analyzing, through qualitative study, the types of interaction trajectories this mechanism can trigger among strangers, including initiation, synchrony, joint understanding, failure, and avoidance. The deepest contribution is expanding HRI's understanding of robot roles: robots need not only be direct interaction partners but can also act as spatial mediators, influencing human-human engagement by reconfiguring sightlines and proxemic relationships in shared spaces.