Human-Robot Interaction

Human-robot interaction (HRI) is the interdisciplinary field of research that studies, designs and evaluates the ways in which people and robots communicate, collaborate and work together. It combines robotics, artificial intelligence and human-computer interaction with psychology, design and the social sciences. What distinguishes it from interacting with a computer is that the robot has autonomy, physical presence and decision-making capacity.

How an interaction is characterized

An interaction between a human and a robot is described along several axes. One is proximity: remote interaction, as in teleoperation, versus close interaction, when both share the same physical space. Another is the robot's degree of autonomy, on a continuum between teleoperated and autonomous. And another is the communication modalities: language, gestures, gaze or physical contact.

Key concepts

Several concepts are central. Physical safety is a priority when the robot shares space with people. Evaluation cannot be reduced to a slogan about trust: NIST's 2003 framework proposes assessing interaction through situational awareness and the information required by the human and the robot at different autonomy levels. And in very humanoid robots the “uncanny valley” appears (Masahiro Mori, 1970): the discomfort caused by a robot that looks almost human, but not quite.

Applications

Applications should be named with their scope. A 2018 NIST chapter examines HRI in industrial robotics, including collaborative-robot safety. And PaLM-E, published in 2023, combines textual, visual and continuous-state inputs for embodied tasks such as robotic manipulation planning. These are two documented lines of work; by themselves they do not prove general effectiveness in healthcare, education or autonomous driving.