Embodiment
A physical presence that acts in the environment: head, base, arms, screen, expression, lights, sound or other actuators.
Manifesto · Social Robotics
From machine
to relational
presence
Social robotics is the shift from robotics as a tool to robotics as a presence — embodied, situated, understandable and safe enough to occupy real human spaces.
01
What social robotics is
A robot, according to the International Federation of Robotics (ISO 8373), is an actuated, programmable mechanism with some degree of autonomy for locomotion, manipulation or positioning. A service robot is used in a personal or professional context to perform tasks that are useful to people or equipment.
Social robotics is both a subset and a cross-cutting layer of robotics: it appears in service, healthcare, education, retail, home, entertainment, telepresence and assistive robots.
Social robotics is the development of embodied robots with enough perception, action and autonomy to interact with people according to social signals, norms and expectations — with the aim of informing, assisting, caring, teaching, accompanying, collaborating, entertaining or mediating human relationships.
This definition separates three things that are often confused: social robot, humanoid robot and conversational assistant. A humanoid may not be social if it only performs industrial movements. A chatbot may be socially competent, but without an actuated body it is not a robot in the robotic sense. A hospital delivery robot, on the other hand, gains a social layer when it understands human presence, negotiates passage and interacts appropriately with patients.
02
What makes a robot social
The criterion is not having a face, arms or a human voice. The criterion is: social interaction is an essential part of the robot’s function. Fong, Nourbakhsh and Dautenhahn popularized the idea of “socially interactive robots” as robots for which social human–robot interaction is central; Feil-Seifer and Matarić consolidated the notion of socially assistive robotics.
In practice, a social robot needs six minimum attributes:
Situated autonomy
It is not a remote control; it can perceive the current state and make decisions within defined limits.
Perception of people
It detects relevant human presence, speech, distance, attention, gestures, context or actions.
Social communication
It uses speech, gaze, gesture, posture, expressions, sounds, text or movement to make its intent understandable.
Defined social role
It acts as a tutor, guide, assistant caregiver, receptionist, colleague, mediator, companion or service agent.
Safety and governance
It protects the person’s body, data, privacy and autonomy; it knows how to fail, ask for help and transfer control.
03
Why the field exists
The field exists to resolve a tension: robots need to enter environments made for people, but those environments are ambiguous, cultural, emotional and unpredictable. A robot therefore cannot be merely efficient — it needs to be legible, predictable, respectful and cooperative.
Human assistance
Social robots can support older adults, children, patients, neurodivergent people, people in rehabilitation or users who need guidance, reminders, motivation or companionship. Socially assistive robotics sits at the intersection of assistance and social interaction.
Mediation with complex systems
In hospitals, schools, hotels, museums, banks and airports, the robot serves as a physical interface for information, triage, navigation, welcoming and referral.
Scaling human services
The mature goal is not to replace human care or teaching, but to absorb repetitive tasks, maintain continuity and free professionals for higher-value relationships. Recent health reviews point to potential in emotional well-being, therapy, loneliness and cognition — while emphasizing the need for longitudinal evidence and ethical care.
Humanizing automation
As robots enter public and domestic spaces, the social layer becomes a form of “embodied UX”: the robot needs to show what it knows, what it does not know, what it will do and when it needs help.
04
What a social robot needs to do
There is no universal list of mandatory sensors or actions. The requirement is functional: the robot needs to perceive people, interpret context, decide within limits and act in a socially understandable way.
Social perception
The robot needs to perceive human presence, position, distance, speech, gestures, objects and context. An important caution: inferring human emotion from face or voice is technically fragile and regulatory sensitive. The serious trend is to speak in terms of observable signals of state and engagement.
Multimodal communication
Voice, text, screen, gaze, LEDs, expressions, movement, gestures, sounds or posture. Good social communication is not only about “responding”: it is knowing how to greet, wait one’s turn, interrupt carefully, ask for consent, explain an action, show uncertainty and end the interaction.
Social planning
Beyond planning movement, the robot needs to plan behavior: when to approach, retreat, wait, guide, insist, stop and ask for human help.
Personalization and memory
Interactions improve when the robot remembers preferences, language, routine and history. But memory in social robotics is a delicate power: it requires consent, data minimization, explainability and the ability to delete records.
05
Perception proportional to the role
A social robot does not need every possible sensor. It needs the set appropriate to its role. The practical rule is: social sensors should be proportional to social benefit.
| Sensor | Social function |
|---|---|
| Microphones / array | Capture speech, locate the speaker, detect interruptions, noise and turn-taking. |
| RGB cameras | Detect people, objects, gestures, faces, visual context and approximate attention. |
| RGB-D / ToF cameras | Measure distance, posture, body position and navigation safety. |
| LiDAR / ultrasound | Navigate human environments and maintain a safe social distance. |
| Tactile sensors | Detect touch, affection, accidental contact or physical interaction. |
| Force / torque | Limit force, detect collision and manipulate objects safely. |
| Buttons / e-stop | Give the person direct control and ensure a safe stop. |
| Environmental sensors | Light, temperature, air quality, presence and institutional context. |
07
A three-wave trajectory
The social robotics market does not appear as a clean category in the statistics. In World Robotics 2025, the IFR reported around 199,000 professional service robots sold in 2024, up 9%; 16,700 medical robots, up 91%; and 20.1 million consumer service robots. Home-care robots still represent a niche — 536 sales recorded in the sample.
Wave 1 — B2B and institutions
Hospitals, hotels, retail, museums, schools and laboratories buy when there is clear ROI, controlled operation and professional maintenance.
Wave 2 — B2B2C and hybrid care
Health plans, home-care providers, senior residences and education networks offer robots as a service, with subscriptions and integrated human support.
Wave 3 — Home and consumers
Plausible when cost, autonomy, privacy, reliability and maintenance are solved. Until then: niches for older adults, children and telepresence.
The Robotics-as-a-Service model is especially relevant because it reduces upfront investment and turns the robot into a monthly service with support, maintenance and continuous updates.
08
Why the field is difficult
The field is technically and ethically difficult because people are not passive “environments”. They interpret intent, form bonds, change moods, have culture, memory, fear, shame and expectations.
Social ambiguity
A smile can mean happiness, nervousness or politeness. Silence can mean refusal, doubt or shyness. The robot needs to handle uncertainty without pretending to be certain.
Long-term engagement
Many robots enchant people during the first week and become furniture afterward. The challenge is to sustain real value, not novelty.
Privacy
Social robots capture speech, images, routines and behavior in intimate spaces. This demands a very careful data architecture.
Emotional dependence
Companion robots can create a poor substitute for human bonds when they are designed badly to maximize engagement.
Regulation
The EU takes a risk-based approach to AI, focusing on safety, fundamental rights and transparency — strongly affecting social robots.
Evidence of effectiveness
In healthcare and education, it is not enough to seem interesting — outcomes, safety and real cost-effectiveness must be demonstrated.
09
A family of robots with clear roles
Social robotics tends to stop being a separate category and become an essential layer of any robot operating near people. Just as every piece of software needs a good interface, every robot in a human space will need minimum social behavior: communicate intent, respect distance, ask for consent, explain failures, adapt language and know when to call a person.
The major change in the coming years is the fusion of social robotics with multimodal models. Language, vision-language and vision-language-action models promise better understanding of context, dialogue, planning and generalization. But LLMs do not guarantee physical safety, reliable perception, privacy or appropriate social behavior — safety and governance remain central challenges.
Assistant caregivers
Expand autonomy and continuity without replacing human care.
Tutors
Increase practice, motivation and personalization in education.
Guides
Organize flows in hospitals, airports, stores and public buildings.
Specialized companions
For loneliness, routines, cognitive engagement and family connection.
Colleagues
Collaborate in work environments with safety and transparency.
Characters
Create cultural, therapeutic, educational and brand experiences.
In summary: social robotics is the shift from robotics as a machine to robotics as a controlled relational presence. It does not require the robot to be human or to pretend to feel emotions. It requires the robot to perceive people, act clearly, respect boundaries, contribute to a human goal and be safe enough to occupy real social spaces.
Blog
Recent publications
Market
What the World Robotics 2025 numbers reveal about care and service
The IFR recorded 199,000 professional service robots sold in 2024, up 9%. The most revealing data point, however, is what has not happened yet.
Ethics
Emotional dependence in companion robotics: what the research says
Companion robots can reduce loneliness in specific contexts. But the same design that creates a bond can create an impoverished substitute for human relationships.
Technology
LLMs in social robotics: real opportunities and underestimated risks
Language models dramatically improve dialogue and planning in social robots. But trust, bias and appropriate social behavior remain central challenges.