I am very pleased to have this opportunity to write a foreword to
Automated Planning: Theory and Practice. With this wonderful text,
three established leaders in the field of automated planning have
met a long-standing need for a detailed, comprehensive book that
can be used both as a text for college classes—complete with
student exercises—and as a reference book for researchers and
practitioners. In recent years, comprehensive texts have been written
for several of the other major areas of Artificial Intelligence (AI),
including machine learning, naturallanguage processing, and constraint
satisfaction processing, but until now, the field of planning has
been devoid of such a resource, despite the considerable number of
advances in and the significant maturation of planning research in
the past decade. With Automated Planning: Theory and Practice, Dana
Nau, Malik Ghallab, and Paolo Traverso have filled that void and have
done so with a remarkably clear and well-written book.
The authors made an important decision about what to emphasize in
the text. Although, as they point out in the Preface, “the
bulk of research on automated planning focuses on…classical
planning,” they decided not to devote a proportional amount
of their text to this restrictive framework but instead to showcase
and emphasize techniques that move beyond the classical strictures.
They don’t ignore classical techniques: indeed, they provide
a comprehensive coverage of them. But they then go on to provide
ample coverage of such topics as temporal planning and resource scheduling,
planning under uncertainty, and the use of a wide range of modern
techniques for plan generation, including propositional satisfiability,
constraint satisfaction, and model checking.
Making good on its name, the book also includes a large section
on the practice of AI planning, including several case studies both
of application classes such as robot planning and of specific fielded
applications such as the Deep Space 1 Remote Agent and the Bridge
Baron game player. These case studies illustrate the process by which
the relatively generic methods of automated planning are transformed
into powerful domain-dependent tools and also show that, as a result
of this transformation, automated planning can be made useful—that
planning is good for much more than just stacking towers of blocks.
(Now if only I could transform myown bridge-playing strategies so
effectively!)
Automated Planning: Theory and Practice is a terrific contribution
to the AI literature and will be widely used not only by those of
us already active in the field of planning but also by our students
and colleagues who want to know more this important area of research.
Martha E. Pollack
University of Michigan
Copyright © 2004 by Elsevier Inc. All rights reserved. Used with
permission. |