Proxemics

This zone is reserved for close, personal interactions such as whispering, touching, or embracing. Intrusion by a non-intimate agent (like a robot) is almost universally perceived as a severe violation.

• HRI Design Implication: Robots should almost never enter this zone autonomously. Entry is only permissible for explicit, consensual tasks (e.g., a healthcare robot administering a patch) or during physical collaboration where contact is the goal.
• Key Challenge: Requires extremely high-confidence intent recognition and explicit user permission protocols.

The space for conversations with friends, family, and close acquaintances. This is the primary zone for collaborative work and most direct HRI.

• HRI Design Implication: This is the target zone for a robot engaging in conversation, handing over tools, or working side-by-side with a human. Approaching within this zone signals an intent to interact.
• Key Consideration: The robot must approach and orient itself appropriately (e.g., frontal orientation for conversation) and may need to adjust within the zone based on cultural norms and individual preferences.

The distance maintained for formal, impersonal, or group interactions, such as with co-workers or service personnel. Communication in this zone is more formal.

• HRI Design Implication: This is the typical zone for a robot's initial approach or for non-intrusive observation. A warehouse AMR might navigate primarily in this zone when not directly interacting. It is also the zone for public-facing robots (e.g., information kiosks).
• Navigation Link: Adherence to this zone is a core component of socially compliant navigation algorithms.

The distance used for public speaking or performances. Individuals at this distance are not considered part of an immediate interaction.

• HRI Design Implication: A robot in this zone is generally not in active interaction. Its presence may be noted, but it is not expected to engage. Path planning for mobile robots can treat humans in the public zone more like dynamic obstacles without complex social modeling.
• Exception: The robot may monitor this zone for activity recognition to anticipate future interactions (e.g., a human walking toward it).

The exact boundaries of each zone are not fixed. They are dynamically influenced by:

• Cultural Background: Norms vary significantly between cultures (e.g., Mediterranean vs. Nordic).
• Context: A crowded elevator compresses all zones, while an open park expands them.
• Individual Factors: Personality, gender, and familiarity between individuals.
• Robot Appearance: A large industrial arm may require a larger personal zone than a small, toy-like robot.

HRI systems must therefore implement adaptive models that can sense and respond to these contextual cues.

Proxemic zones directly inform the cost functions and constraints in a robot's motion planning system. Instead of just avoiding collisions, the planner must:

• Minimize intrusion into intimate/personal zones without invitation.
• Shape paths to travel within social/public zones when passing by humans.
• Predict human motion to avoid cutting through a person's projected personal space.
• Use orientation: Facing a human directly while in their personal zone is engaging; a tangential orientation is less intrusive.

This transforms basic obstacle avoidance into socially aware navigation, a key research area in Human-Robot Teaming for shared spaces.

This is the primary application of proxemics for mobile robots (AMRs, delivery bots, social robots). Algorithms use proxemic zones to plan paths that avoid intruding into a human's personal space unless necessary for a task. Key implementations include:

• Path planning that maintains a minimum social distance, often modeled as an elliptical or dynamic zone around a moving person.
• Velocity modulation where the robot slows or stops as it approaches the boundary of a personal zone.
• Overtaking behavior that uses the public zone for passing, mimicking human sidewalk etiquette.

Example: A hospital delivery robot will navigate corridors in the social zone (~1.2-3.6m), slow when approaching a nurse in a hallway, and only enter the personal zone (<1.2m) to hand off a package at a designated 'handover point'.

This governs how a robot initiates and maintains a stationary interaction. The robot must compute an optimal approach vector and final stopping position relative to a human. Engineering considerations include:

• Frontal vs. side approaches: Most cultures prefer a frontal approach for intentional interaction, while a side approach is less intrusive.
• Stopping distance: Determined by interaction type. A kiosk robot may stop at the social zone boundary, while a collaborative manipulator may need to enter the personal zone for handover.
• Orientation: Aligning the robot's 'face' or primary display toward the user to signal engagement.

This is critical for collaborative robots (cobots) on assembly lines and socially assistive robots (SAR) in care settings, where improper positioning can cause discomfort or disrupt workflow.

Proxemic zones are not fixed radii but dynamic, context-sensitive fields. Engineers model them as functions of multiple variables, allowing robot behavior to adapt in real-time. Influencing factors include:

• Cultural and personal preferences: System parameters may be adjusted for different deployment regions or individual user profiles.
• Activity context: A human focused on a task may have a larger effective personal zone. A robot should approach more cautiously or wait.
• Social relationship: A familiar user (e.g., a factory worker who uses the cobot daily) may tolerate a smaller interaction distance than a first-time visitor.
• Environmental constraints: Crowded spaces compress acceptable distances; the robot must recognize this and adjust its norms accordingly.

This modeling is often implemented using probabilistic filters or neural networks that take multi-modal sensor input (vision, audio) to estimate comfortable distance.

This addresses the complex scenario where a robot interacts with or navigates among groups of people. The robot must understand group dynamics and spatial formations. Key concepts include:

• F-formations: The spatial patterning of interacting groups (e.g., a circle for conversation, a side-by-side lineup). A robot should not break through the o-space (the central space of a group).
• Proxemic fading: In a dense group, individual zones overlap; the robot must infer the collective social space.
• Addressing individuals within a group: The robot may need to slightly enter a personal zone to deliver an item to one person, while acknowledging the presence of others in the social zone.

This is essential for robots in retail, museums, or public spaces, where interrupting a conversation is seen as rude and disruptive.

In Physical Human-Robot Interaction (pHRI), proxemics directly informs safety standards and comfort metrics. It provides a quantitative basis for defining safe operational volumes around a robot.

• Separation monitoring: Safety laser scanners or vision systems can define warning and stop zones that correspond to public and personal spaces, triggering Safety-Rated Monitored Stops.
• Comfort vs. safety: A contact may be within biomechanical force and power limiting (PFL) safety limits but still be a proxemic violation, causing startle or stress. Systems should aim to avoid all unplanned contact, not just injurious contact.
• Hand guiding and kinesthetic teaching: These collaborative operations require the human to enter the robot's intimate space. Proxemic principles guide the design of the interaction to make this forced intimacy feel controlled and intentional, using predictable, slow movements.

A robot can use its own positioning and movement as a non-verbal communication channel to signal its intentions to humans, increasing predictability and trust.

• Legible motion: A robot approaching on a direct, smooth path into the personal zone signals an intent to interact. A curved, hesitant path signals caution or uncertainty.
• Hesitation and gaze: Briefly pausing at the edge of a personal zone can act as a 'request for permission' to enter, similar to a human pausing before a conversation.
• Withdrawal: Moving from a personal to a social zone can signal the end of an interaction or that the robot is ceding space.

This turns proxemics from a passive constraint into an active human-robot communication protocol, crucial for fluent human-robot teaming.

An algorithmic approach for mobile robots that extends proxemic principles to motion. Instead of just avoiding collisions, the robot plans paths that respect social norms, such as:

• Maintaining appropriate personal space zones
• Yielding right-of-way
• Moving at predictable, human-comfortable speeds
• Avoiding cutting between interacting people This is essential for robots operating in dynamic environments like hospitals, offices, or retail spaces.

The subfield focused on direct physical contact and force exchange between a human and a robot. While proxemics governs non-contact distances, pHRI deals with the safety and control required when contact is intentional (e.g., hand-guiding) or incidental. Key requirements include:

• Force and torque sensing
• Compliant control algorithms (e.g., impedance control)
• Hardware designed for safe contact (rounded edges, force-limiting joints) It is the engineering foundation for collaborative robots (cobots).

A robot specifically designed with inherent safety features to operate alongside humans in a shared workspace without traditional safety cages. Cobots implement proxemic and pHRI principles through:

• Power and Force Limiting (PFL) actuators
• Safety-rated monitored stop sensors
• Hand-guiding capabilities for direct teaching They are defined by international safety standards like ISO/TS 15066, which specifies biomechanical limits for human contact.

The computational process by which a robot infers a human's goals or planned actions from observed signals. This allows a robot to anticipate needs and adjust its proxemic behavior proactively. Input modalities include:

• Gaze tracking and head pose
• Gesture recognition
• Motion trajectory prediction
• Physiological data (in advanced systems) For example, recognizing a human reaching for a tool allows a robot to move aside or hand over the item, optimizing the interaction flow.

A robot's computational ability to attribute mental states—such as beliefs, knowledge, and intentions—to its human partner. A robot with a ToM model can reason about what the human sees, knows, or intends to do, enabling more nuanced proxemic adjustments. For instance:

• Understanding a human is focused on a task may cause the robot to increase its approach distance to avoid startling them.
• Inferring a human's need for assistance may cause the robot to enter the personal zone appropriately. It represents a high-level, cognitive layer above basic distance maintenance.