Book Review from Kybernetes 32, no. 9/10, pp. 1559-61, 2003
Designing Sociable Robots
by Cynthia L. Breazeal
MIT Press, Cambridge, Massachusetts, 2002, xviii + 263 pp., ISBN 0-262-02510-8, hardback, with CD-ROM, £34.50
This is a comprehensive account of the development of a social robot called Kismet, designed to explore interactions between robot and human. It has been developed by the author, with help from a number of collaborators, in the Media Laboratory of MIT where she now heads a Robotic Presence research group. The Media Laboratory is also the birthplace of the Cog robot due to Rodney Brooks, also used to study interactions with humans. The motivation is explained in text on the CD-ROM as:
Sociable humanoid robots pose a dramatic and intriguing shift in the way one thinks about control of autonomous robots. Traditionally, autonomous robots are designed to operate as independently and remotely as possible from humans, often performing tasks in hazardous and hostile environments. However, a new range of application domains (domestic, entertainment, health care, etc.) are driving the development of robots that can interact and cooperate with people, and play a part in their daily lives.
There are references to humanoid robots of science fiction, a topical one being Lt-Commander Data of Star Trek: The Next Generation. Since facial expressions and head movements are important in communication, Kismet simulates only a head and the means of moving it within a limited range. He does not use speech as such, but can produce utterances whose tone indicates emotions, and can analyse received speech for its emotional content. This is termed proto-dialogue. The interaction of Kismet with a person is meant to imitate that between an infant and caretaker. The word "kismet" (where, strictly, the "i" should not have a dot) means fate or destiny in Turkish and the author's fascination with humanoid robots is such that she feels this to be applicable to her.
Kismet has a considerable amount of visual equipment since there are four video cameras altogether, two incorporated in the eyes and altering direction with them, and two that are fixed. The directing of the eyes toward an object of interest in the visual field gives a realistic indication of the robot's "interest" to an interacting human. Objects that can arouse interest are flesh-coloured oval shapes that are likely to be faces, and areas of bright colour that are likely to be toys. When a face has been detected, it is further analysed to locate the eyes and to focus on a position midway between them.
The robot's means of expressing emotions and reactions facially are quite elaborate and are based on studies of infant behaviour. A large number of servos allow directing of the eyes, and movement of the head as a whole as well as operation of eyelids, eyebrows, lips, and rather large brightly coloured ears. The lips are shaped to express disgust, pleasure and some other emotions, and also move suitably when vocalisation is produced. The overall appearance is undoubtedly a representation of a face, but in a caricature-like way and not flesh-coloured. It is mentioned that there was a conscious decision not to try to imitate a face in detail since people have an instinctive unease when faced with something that seems human but is not fully so. The cover of the book shows the robot with its inventor before a mirror, but clearly not offering the intriguing possibility of the robot accepting its own image as a face.
Some responses depend on low-level processing, so that a fear reaction comes from perceived movement that can be described as something looming over the robot, or by a sudden noise. Anything approaching close causes withdrawal within the limited amount of movement permitted. An interesting point is that, although the amount of movement is small it serves as an indication to a human that the robot's space has been invaded, so that the human also withdraws and in effect the robot's withdrawal is amplified.
Responses are also determined by higher-level processing expressed in terms of internal drives. Three drives play a part, described as social, stimulation and fatigue drives. The social drive is enhanced when a face is detected, and similarly the stimulation drive when a toy is in view. If the detected object is distant, appropriate "engagement" strategies, both vocal and facial, are used to entice it nearer. Social interaction can lead to alternation of utterances by the robot and human. The mechanism determining drives is quite complicated, and the sensitivity of the sensory apparatus to particular inputs is modified according to the levels of internal drives.
The project has proved successful, since the robot has engaged in lifelike and sustained interactions with humans, including those who were naive about its working and purpose. Evidence is given in video clips on the CD-ROM, where there are also sections of text and discourse by the author that give a good overview of the project. Before the video clips can be viewed it is necessary to install the item of software Quick View of version 4 or higher and a link is given to an Apple website where a free copy can be downloaded (of version 6 at the time of writing this review).
The work has immediate practical application in showing how domestic and personal robots, and robotic toys and pets, can most effectively interact with humans. The choices made in designing Kismet were determined by reference to findings in psychology, especially developmental psychology, and now the robot provides a basis for experiments in these fields. The author visualises further developments in which a sociable robot would be able to learn from experience. This is discussed in some detail, with the assumption that learning in the first place would be aimed at improving the effectiveness of the means of satisfying the internal drives, and that this would be a basis for acquiring language and other skills. A facility for being able to recall its own history, as "autobiographic memory" is suggested as an important aspect allowing reflection and perhaps dreams. This is an entertaining book, but at the same time it has practical value and opens up lines of investigation that can shed new light on human psychology and development.
Alex M. Andrew
Back to Reviews