by Michael Wooldridge.
One goal of modern computer science is to engineer computer programs that can act as autonomous, rational agents; software that can independently make good decisions about what actions to perform on our behalf and execute those actions. Applications of such agents range from small programs that intelligently search the Web buying and selling goods, to autonomous space probes. This book provides an advanced introduction to the theory and practice of such agents. It focuses on the belief-desire-intention (BDI) model of agency.
The book begins by introducing the BDI model in detail. Presupposing no prior knowledge of either agents or the BDI model, the book introduces a series of progressively more refined BDI agent designs, discussing the extent to which these designs capture our intuitions about rational behaviour. The book then introduces a BDI logic called LORA (which stands for Logic of Rational Agents). LORA makes it possible to represent the beliefs, desires, and intentions of agents, and how these agents and their environments change over time. After an informal, tutorial-style introduction to LORA, the book shows how the logic can be used to capture many components of a theory of rational agency, including such notions as communication and cooperation. The book concludes with a discussion on the role of logic in agent theory.
Every chapter concludes with a "notes and further reading" section, which provides a discussion of hot topics and extensive pointers to further reading. In addition, two appendices provide a comprehensive survey of the formal foundations for the book.
Reasoning about Rational Agents will be essential reading for students and researchers who wish to understand both the the software design issues and the logical theory of rational agency.
A review from amazon.com:
``An excellent treatment of the formal modelling of what Mike calls Rational Agents and others have (more provocatively) usually termed Intelligent Agents. What sets this apart from other books on the subject are Mike's refreshingly direct, no-nonsense writing, and his recognition that his audience may not have all of the formal tools at their fingertips. The Appendix ``Formal Foundations'' would be a good choice for anyone who wanted to understand modal logics, especially as they apply to time and intention. Even software engineers who would rather hack code than grapple with formalism will find this accesible, stimulating and useful.''
SummaryReasoning about Rational Agents by Michael Wooldridge. Published in June 2000 by The MIT Press (Cambridge, Massachussetts/London, England) as part of the Intelligent robotics and autonomous agents series.
240pp; includes bibliographical references and index.
Buy the Book