Human–robot interaction

Human–robot interaction studies how people and robots affect one another during shared activities. Its subject includes physical robots, remotely operated machines, autonomous systems, and interfaces through which people direct or supervise robots. Unlike conventional robotics, which often evaluates mechanical performance in isolation, HRI also examines human expectations, mental models, workload, trust, emotion, language, and social context.¹

The field grew from work in human–computer interaction, artificial intelligence, social psychology, and autonomous robotics. Researchers distinguish interaction from simple automation: an interaction involves reciprocal adaptation, such as a robot interpreting a person’s gesture, asking for clarification, or changing its plan after observing human behavior. Studies may be conducted in laboratories, homes, hospitals, factories, public spaces, or extreme environments.

HRI design depends on the roles assigned to people and robots, the communication channels available, and the consequences of error. Common modes include teleoperation, supervisory control, shared autonomy, physical collaboration, and social interaction.² A person may directly control a robot, delegate a goal while retaining oversight, or work beside a robot that recognizes actions and coordinates movement.

Communication can combine speech, gaze, gesture, touch, graphical displays, lights, sound, and robot motion. Effective interfaces make the robot’s intentions and limitations legible without overwhelming the user. In collaborative work, explainable artificial intelligence, adjustable autonomy, and timely alerts can help people predict what a robot will do and intervene when necessary. Physical interaction additionally requires force limits, collision avoidance, compliant mechanisms, and procedures for safe recovery.

Trust and safety are central HRI concerns because people may rely on robots whose capabilities are difficult to judge. Trust is influenced by reliability, transparency, appearance, prior experience, perceived competence, and the costs of a robot’s mistakes; excessive trust can produce misuse, while insufficient trust can lead to disuse.³

Evaluation therefore combines technical and human measures. Researchers may record task completion, errors, intervention frequency, response time, workload, perceived usability, trust, acceptance, and physiological or behavioral indicators. Experiments also test differences among users, including age, expertise, disability, culture, and familiarity with automation. Safety standards such as ISO 13482 address personal-care robots and related hazards, but safe HRI also depends on training, organizational procedures, privacy protections, and clear responsibility when autonomous decisions cause harm.⁴

HRI includes overlooked settings in which interaction is indirect, collective, or highly constrained. A robot may influence a crowd without speaking, coordinate with a team rather than one operator, or support a person with motor, sensory, or cognitive disabilities. Researchers also study robot-robot coordination as it affects human observers, children’s responses to machines, cross-cultural differences in expectations, and the social effects of robots that merely appear autonomous.

Some of the field’s hardest problems arise from ambiguity rather than movement: people use underspecified instructions, gestures vary by culture, and a robot’s silence may be interpreted as competence, failure, or consent. Long-term deployment raises further questions about adaptation, attachment, privacy, deskilling, and changing patterns of responsibility. These edge cases have made participatory design, human-subjects research, and deployment studies important complements to short laboratory trials.⁵

Shared autonomy

A control arrangement in which a human and an autonomous robot jointly determine actions, with authority distributed between them.

Adjustable autonomy

A system’s ability to vary how much decision-making authority is given to the human or the robot.

Mental model

A person’s internal understanding of how a robot works, what it can do, and how it will behave.

Social robot

A robot designed to communicate or behave in ways intended for social interaction with people.