|
|
Zoom
Preface
Evolutionary robotics is a new technique for automatic creation of autonomous robots.
It is inspired by the darwinian principle of selective reproduction of the fittest. It is a
new approach which looks at robots as autonomous artificial organisms that develop their
own skills in close interaction with the environment without human intervention. Heavily
drawing from natural sciences like biology and ethology, evolutionary robotics makes
use of tools like neural networks, genetic algorithms, dynamic systems, and biomorphic
engineering.
The term evolutionary robotics has been introduced only quite recently (Cliff, Harvey
and Husband 1993), but the idea of representing the control system of a robot as an artificial
chromosome subject to the laws of genetics and of natural selection dates back to the end
of the 1980’s when the first simulated artificial organisms with a sensory motor system
began evolving on computer screens. At that time, however, real robots were still machines
that required accurate programming efforts and careful manipulation. Toward the end of
that period, a few engineers began questioning some of the basic principles of robot design
and came up with a new generation of robots that shared important characteristics with
simple biological systems: robustness, simplicity, small size, flexibility, modularity. Above
all, these robots were designed so that they could be programmed and controlled by people
with different backgrounds and levels of technical skills. In the years 1992 and 1993, the
first experiments on artificial evolution of autonomous robots were reported by our team
at the Swiss Federal Institute of Technology in Lausanne, by a team at the University of
Sussex at Brighton, and by a team at the University of Southern California. The success
and potentials of these researches triggered a whole new activity in evolutionary robotics
in labs across Europe, Japan, and the United States.
In the very last few years evolutionary robotics has gathered the interest of a large com-
munity of researchers with different research interests and backgrounds (ranging from AI
and robotics, to biology and cognitive science, to the study of social behavior). Continuous
investment, growth, and progress in evolutionary robotics has caused a substantial matura-
tion of the methodology and of the issues involved, and at the same time has generated a
diversification of the basic methodology. This book provides a comprehensive description
of what evolutionary robotics is, of what its scientific and technological milieu are, of the
various methods employed, of the results achieved so far, and of the future directions. The
book aims at clarity of explanation, avoiding as much as possible (or accurately explaining)
scientific jargon. The reader is gently introduced to the subject following a historical and
logical path. The book describes the most used techniques (genetic algorithms, neural net-
works, etc.), presents several experiments of increasing complexity together with related
issues as they arise, and shows the most promising future directions.
View Printable Page
Zoom
Preface
Evolutionary robotics is a new technique for automatic creation of autonomous robots.
It is inspired by the darwinian principle of selective reproduction of the fittest. It is a
new approach which looks at robots as autonomous artificial organisms that develop their
own skills in close interaction with the environment without human intervention. Heavily
drawing from natural sciences like biology and ethology, evolutionary robotics makes
use of tools like neural networks, genetic algorithms, dynamic systems, and biomorphic
engineering.
The term evolutionary robotics has been introduced only quite recently (Cliff, Harvey
and Husband 1993), but the idea of representing the control system of a robot as an artificial
chromosome subject to the laws of genetics and of natural selection dates back to the end
of the 1980’s when the first simulated artificial organisms with a sensory motor system
began evolving on computer screens. At that time, however, real robots were still machines
that required accurate programming efforts and careful manipulation. Toward the end of
that period, a few engineers began questioning some of the basic principles of robot design
and came up with a new generation of robots that shared important characteristics with
simple biological systems: robustness, simplicity, small size, flexibility, modularity. Above
all, these robots were designed so that they could be programmed and controlled by people
with different backgrounds and levels of technical skills. In the years 1992 and 1993, the
first experiments on artificial evolution of autonomous robots were reported by our team
at the Swiss Federal Institute of Technology in Lausanne, by a team at the University of
Sussex at Brighton, and by a team at the University of Southern California. The success
and potentials of these researches triggered a whole new activity in evolutionary robotics
in labs across Europe, Japan, and the United States.
In the very last few years evolutionary robotics has gathered the interest of a large com-
munity of researchers with different research interests and backgrounds (ranging from AI
and robotics, to biology and cognitive science, to the study of social behavior). Continuous
investment, growth, and progress in evolutionary robotics has caused a substantial matura-
tion of the methodology and of the issues involved, and at the same time has generated a
diversification of the basic methodology. This book provides a comprehensive description
of what evolutionary robotics is, of what its scientific and technological milieu are, of the
various methods employed, of the results achieved so far, and of the future directions. The
book aims at clarity of explanation, avoiding as much as possible (or accurately explaining)
scientific jargon. The reader is gently introduced to the subject following a historical and
logical path. The book describes the most used techniques (genetic algorithms, neural net-
works, etc.), presents several experiments of increasing complexity together with related
issues as they arise, and shows the most promising future directions.
View Printable Page
|
