Designing Sociable Robots

Cynthia Breazeal
Robotic Life Group
Media Lab
Massachusetts Institute of Technology
Building E15, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA

This is the abstract of a talk prepared for the International interdisciplinary seminar on new robotics, evolution and embodied cognition (IISREEC).12th to 15th November 2002, Lisbon, Portugal

Abstract: This talk motivates a new paradigm of interaction between humans and robots---robot as a sociable partner rather than just a tool. This runs counter to traditional uses of autonomous robots (e.g., sweeping minefields, inspecting oil wells, exploring other planets, etc.) that require very little (if any) interaction with humans. Other applications such as robot-assisted surgery, delivering hospital meals, vacuuming floors, or search and rescue bring autonomous robots into environments shared with people. However, our relationship with our robots has fundamentally not changed. As robots become more capable, they are still largely viewed as tools that extend the abilities of the scientist, surgeon, fire fighter, or otherwise highly trained specialist.

However, numerous experimental findings in the field of human computer interaction (HCI) and human robot interaction (HRI) are fueling a new movement to develop sociable robots---autonomous robots that are specifically designed to interact with, communicate with, and learn from people in familiar human terms. Over 70 HCI studies during the past ten years that have shown that people do not respond to interactive software on a desktop computer as a mere tool. Instead, individuals bring to bear a wide range of social rules and learned behaviors that guide their interactions with, and attitudes toward, interactive systems. These social and emotional responses occur even when: a) users know that they should not and believe that they do not exhibit these responses, b) designers do not attempt to elicit these responses, and c) the interfaces are as simple as plain text on a screen, with no reference to any social or emotional aspects (e.g. , no use of the word "I") .Furthermore, users exhibit social-emotional responses without explicit training, and regardless of their level of computer experience.

Similar findings are supported, if not strengthened, by more recent HRI studies. These findings also suggest that robots that participate in rich human-style social exchange with people offer a number of advantages. First, people would find working with them more enjoyable and would thus feel more competent. Second, communicating with them would not require any additional training since humans are already experts in social interaction. Third, if the robot could engage in various forms of social learning (imitation, emulation, tutelage, etc.), it would be easier for the human to teach the robot new tasks. Early instantiations of such robots are already motivating new domestic, entertainment, education, and health applications for ordinary people as part of their daily lives.

We argue that this not a trivial or accidental phenomenon. Rather, the social and emotional reactions that users have towards such systems are important keys to building more useful, successful, and productive technologies. In this talk, we outline our case in favor of sociable robots, review our own body of work in designing and evaluating socially intelliegent robots, and suggest future directions.

Related Links

For more information contact Luis Rocha at
Last Modified: November 6, 2002