Kismet: Fundamentals and Applications

By Fouad Sabry

What Is Kismet

Dr. Cynthia Breazeal of the Massachusetts Institute of Technology created the robot head known as Kismet in the 1990s as an experiment in emotional computing. Kismet is a machine that is capable of recognizing and simulating emotions. The name Kismet derives from a Turkish word meaning "fate" or occasionally "luck".

How You Will Benefit

(I) Insights, and validations about the following topics:

Chapter 1: Kismet (robot)

Chapter 2: Affective computing

Chapter 3: Facial expression

Chapter 4: Lip reading

Chapter 5: Paul Ekman

Chapter 6: Cynthia Breazeal

Chapter 7: Domo (robot)

Chapter 8: Prosody (linguistics)

Chapter 9: Social cue

Chapter 10: Emotion recognition

(II) Answering the public top questions about kismet.

(III) Real world examples for the usage of kismet in many fields.

(IV) 17 appendices to explain, briefly, 266 emerging technologies in each industry to have 360-degree full understanding of kismet' technologies.

Who This Book Is For

Professionals, undergraduate and graduate students, enthusiasts, hobbyists, and those who want to go beyond basic knowledge or information for any kind of kismet.

• Language: English
• Publisher: One Billion Knowledgeable
• Release date: Jul 7, 2023

Book preview

Chapter 1: Kismet (robot)

In the 1990s, Dr. Cynthia Breazeal conducted an experiment in affective computing at the Massachusetts Institute of Technology using a robot head she created called Kismet. Kismet is derived from a Turkish word that means fate or luck..

Kismet has sensors that allow it to see, hear, and feel its surroundings, allowing it to interact with humans more naturally. Kismet acts out its emotions with a range of vocalizations, facial expressions, and physical gestures. Emotions are conveyed through shifts in the position of the jaw, the chin, the lips, and the brows. Estimated raw material costs came to $25,000 USD.

Kismet's SNS, as its artificial intelligence software is called, was built with human models of cognizant behavior in mind. Here are its six constituent parts:.

Raw data from cameras and microphones is processed by this system. Kismet's vision system has the controversial capability of recognizing human skin tone in addition to eye and motion detection. Kismet temporarily turns off its motion detection system whenever its head is moved so that it doesn't get confused with its own motion. It uses its stereo cameras to make distance estimates, so it can detect threats, such as large, nearby objects with a lot of motion.

Dr. Breazeal compares her interactions with the robot to those between a caretaker and a child, with herself playing the role of the caretaker and the robot that of a child. In this summary, Dr. Breazeal serves as a learning framework for the human-robot relationship, providing a foundation upon which Kismet can grow. Dr. Brazeal provides a demonstration of Kismet's abilities, describing the robot's motivational state through a series of facial expressions: This one is anger (laugh) extreme anger, disgust, excitement, fear, happiness, interest, sadness, surprise, tiredness, and sleep.

Kismet uses a wide range of phonemes in his proto-language, not unlike a baby's babbling. Emotions are conveyed through modifications to the voice synthesizer DECtalk's pitch, rhythm, articulation, etc. The difference between a question and a statement is indicated by the speaker's use of intonation. The animators' mantra, simplicity is the secret to successful lip animation, informed the team's approach to lip synchronization, which contributed to the game's realism. The goal was to create a visual short hand that passes unchallenged by the viewer, rather than to replicate lip movements word for word.

{End Chapter 1}

Chapter 2: Affective computing

Systems and devices that can detect, interpret, process, and simulate human emotions are the focus of research and development in the field of affective computing. It combines elements from computer science, psychology, and the study of the mind. Giving machines emotional intelligence, such as the ability to simulate empathy, is one of the motivations for the research. A good emotional response from a machine would be one that can read its human users' emotions and change its own behavior accordingly.

Passive sensors that record information about the user's physical state or behavior without analyzing it are often the starting point for emotion detection. The information obtained is comparable to the clues that people use to identify the feelings of those around them. Video cameras can record nonverbal cues like expressions and body language, while audio recorders can pick up sounds like voices. Physiological data, such as skin temperature and galvanic resistance, can be directly measured by other sensors in order to deduce emotional cues.

Extracting useful patterns from the collected data is essential for emotional recognition. This is accomplished through the use of machine learning procedures for processing various modalities, such as voice recognition, NLP, and emotion analysis. Most of these methods aim to generate labels that are consistent with what a human perceiver would assign in a similar scenario. For instance, a computer vision system could be trained to identify a furrowed brow as a confused, concentrating, or slightly negative expression (as opposed to positive, which it might say if they were smiling in a happy-appearing way). These designations may or may not be accurate reflections of the individual's internal state.

Affective computing also includes the development of computational devices with the potential to demonstrate either innate emotional capabilities or convincingly simulate emotions. Current technological capabilities make the simulation of emotions in conversational agents a more feasible method for improving human-machine interaction.

Both continuous and categorical methods have been used in the fields of psychology, cognitive science, and neuroscience to describe and organize people's emotional experiences. Negative versus positive, relaxed versus agitated, and so on are common axes used in the continuous method.

Emotions are typically categorized into happy, sad, angry, fearful, surprised, and disgusted under the categorical approach. Machines can be trained to generate continuous or discrete labels using a variety of regression and classification models in machine learning. Models that allow combinations across the categories, such as a happy-surprised face or a fearful-surprised face, are also occasionally constructed.

Some of the many forms of input data utilized by the emotion recognition task are discussed below.

Affective technologies use this information to infer a person's emotional state based on subtle changes in their speech that occur