Genetic Algorithms and Robotics

The following sections are included:

• FOREWORD

• PREFACE

• CONTENTS

The following sections are included:

• An Assemblage of Difficult Problems
  • Function Optimization and the Black Box Paradigm
  • The Travelling Salesperson Problem
  • Robot Trajectory Generation

• A Blind Search Strategy

• Illustrated Use of the Black Box

• Summary

The following sections are included:

• Evolutionary Mechanisms as Adaptive Search Procedures

• Why Use a Representation Instead of the Real Thing?

• The Schemata Processing Paradigm

• The Dynamics of Schemata Frequency

• Why They Are Called Genetic Algorithms

• When is the Use of Genetic Algorithms Advocated?

• The Epistasis² Dogma

• Summary

The following sections are included:

• The History of Genetic Algorithms

• Previous Work Relating to the Main Issues Raised in this Book

The following sections are included:

• What Makes Robots Tick

• The Human Interface

• Genetic Algorithms in Robotics

• Summary

The following sections are included:

• The Robot Simulation Package

• Trajectory Representation

• The Size of the Trajectory Space

• Multi-bit, Multi-allele parameters
  • Multi-allele Genes

• Searching for Hyperplanes in the Trajectory Space

• Over-specified Representations
  • An Example of a Redundant Trajectory Representation

• Summary

The following sections are included:

• Trajectory Reproduction
  • Analogous Crossover
  • Segregation Crossover

• The Mutation Operator

• Deletion and Addition Operators

• The Evaluation Mechanism

• The Selection Mechanism

• Summary

The following sections are included:

• Performance of the Trajectory Genetic Algorithm

• Between Random and Genetic Search

• Hill-climbing Versus Genetic Algorithms

• Summary

The following sections are included:

• Lamarckism

• Lamarckian Probability for Reproduction

• Results and Discussion

• Summary

The following sections are included:

• The Linear Assumption

• Basic Elements of Epistasis

• Calculating Epistasis for Two Illustrative Functions

• Partial Populations and Sampling Noise

• Summary and Conclusions

The following sections are included:

• The Anomaly

• Adaptation Anomaly in Nature
  • Pre-attentive Human Vision
  • Population Genetics
  • The Development of a Nervous System

• Entropy, Information and Adaptation

• The Evolution of Trajectories
  • ‘Unique’ Reasons for the Anomaly in the Trajectory-GA

• General Reasons for the Adaptation Anomaly

• Summary

This book grew out of a process control problem in robotics I found difficult to approach with classical optimization methods. Having no clear answer ‘how’ to solve the problem, I turned it around and asked ‘why’ such a problem is not amenable to traditional problem solving methods. A fellow student who noticed my anguish quoted a lecturer who said “…if you cannot answer the question, model the issue…”. In as much as my interest in GAs was originally aimed at solving a specific problem, GAs, with their immense diversity and richness, steered my interests towards the general issues of adaptation in natural and artificial systems. What struck me most – and still does – is that in contrast to the relatively simplistic mechanisms of a GA, its behaviour as a whole is complex. This facet of complex systems, the inherent difference between the functioning of the whole and the sum functioning of the parts, is a fundamental aspect which explains the sophistication of GA applications and, of course, of nature…

The following sections are included:

• APPENDIX A The Robot Dynamic Simulation Package ROSI

• BIBLIOGRAPHY

• INDEX