SiR - Lecture 2

What is a social robot?

A robot designed for social interaction, combining autonomy + communication abilities.

Cultural origins of social robots

Inspired by myths, automata, and sci-fi, shaping expectations of companionship.

Technological origins of social robots

Developed through advances in AI, robotics, sensors, and machine learning.

Anthropomorphism

Humans naturally attribute human traits to non-human things (incl. robots). For example: talking to your robot like a person, naming your Roomba, thinking your car is “angry,” or feeling bad for a robot that falls over

Social Robot Paradox

Humans respond socially to robots, but robots lack real social competence.

Uncanny Valley

Robots that look almost human but not fully → cause discomfort.

Core capability: Multimodal Communication

Robots communicate via voice, gesture, gaze, lights, sound.

Core capability: Affective Expression

Robots show emotions (smile, tone, lights) to improve interaction.

Core capability: Personality Traits

Robots maintain a consistent style (friendly, calm, energetic).

Core capability: Learning & Social Modeling

Robots adapt based on feedback and observe human behavior.

Core capability: Relationship Management

Robots handle long-term interactions: memory, continuity, trust.

Social Capability Spectrum (from simple → advanced…)

1. Socially evocative

2. Communication robots

3. Affective robots

4. Learning robots

5. Socially competent robot

Shared attention (meaning)

Human and robot focus on the same thing → increases social connection.

Cognitive modeling

Robot predicts human actions/intents using basic Theory of Mind.

Appearance types: Functional

Looks like a tool; design focused on utility.

Appearance types: Artifact-shaped

Looks like objects (lamps, toys) that talk or express.

Appearance types: Bio-inspired

Humanoid or animal-like (zoomorphism); follows living-being design cues.

Proxemics

Social rules about personal space → robots must respect distances.

Context awareness

Robot adapts behavior to location, culture, situation.

Relational role: Robot for you

Robot acts as a tool/helper.

Relational role: Robot as you

Robot acts as human’s proxy (= something or someone that does a task on your behalf)

Relational role: Robot with you

Robot is a teammate/collaborator.

Relational role: Robot around you

Robot shares space but doesn’t collaborate.

Relational role: Robot as if you

Robot mimics human behavior.

Relational role: Robot part of you

Robot becomes a body extension (prosthetics, exoskeleton).

Human-centered design

Design focuses on user needs, comfort, trust.

Robot-centered design

Design focuses on robot performance, humans adapt to robot.

Social perception

Robot recognizes human faces, gestures, gaze, emotion.

Intelligence (definition)

Ability to achieve goals under uncertainty.

Autonomy (definition)

Robot can act independently without external control.

Measuring autonomy

Through perception, planning, learning, action execution, modeling.

3 Proximity types Interaction

Physical, co-located, remote.

Temporal interaction dimensions

Timespan, duration, frequency.

Verbal communication: Generation

Robot produces content + delivery (tone, timing, gesture sync).

Verbal communication: Recognition

Robot uses semantic parsing, grounding, multimodal cues.

Kinesics

Body movements: pointing, nodding, regulating turn-taking.

Gaze functions

Shows attention, regulates turns, conveys attitudes.

Affective computing

Detecting + responding to human emotion.

Arousal-Valence Model

Emotions described by energy level + positivity/negativity.

Theory of Mind (ToM)

Understanding what others think, feel, or intend. Helps robots predict what humans will do next.

Legible motion

Robot moves in ways humans easily understand (clear intent).

Intention recognition

Robot predicts what humans will do next.

Handover collaboration

Robot safely and clearly exchanges objects with humans.

Socially Aware Navigation

Robots move respecting intentions, norms, comfort zones.

Socially Guided ML: Demonstrations

Robot learns by watching humans.

Socially Guided ML: Social cues

Robot uses gaze, feedback, gestures to improve learning.

Socially Guided ML: Scaffolding

Human teaches robot step-by-step.

Socially Guided ML: Transparency

Robot shows its learning state.

Multimodal integration

Robot fuses speech, gesture, gaze, emotion to act socially