Real-time intelligent agent technology offers companies and
individuals unprecedented power in planning and reacting to
market forces, competitors, and trends. This book explains
the technology in-depth, and shows how it can be employed
for maximum impact. Two leading artificial intelligence
experts cover the key issues associated with real-time and
reactive planning in intelligent agent systems. Start by
understanding the domains -- from toy to real-world -- that
have been used to develop and test planning systems. Walk
through a detailed case study: a sample real-world,
real-time domain covering flight planning. Compare systems
that make decisions at planning and execution times, and
identify a framework for deciding when to plan and when to
react. Understand leading approaches such as Brooks'
subsumption architecture, Rosenschein/Kaelbling's situated
automata; Schoppers' universal planning, and more. Finally,
review real-time planning frameworks and new extensions
that overcome many of their limitations, including a new
language framework for describing reactive behaviors.For
all software engineers building real-world, real-time
systems using artificial intelligence.
Features
- Authored by two of the world's leading artificial
intelligence practitioners.
- Includes a comprehensive real-world case study.
- Covers the latest extensions to real-time planning
frameworks.
Contents
1. Evolution of Planning.
A Brief History of
Planning. Types of Domains Used In AI. Conventional
Planners. Pure (Hard-Wired) Reaction. Classical Real-time
Planners. Book Synopsis.
2. Subsuming Planning to Reaction. Minton's Utility Function.
Insect Intelligence and Robotics: The Work of Rodney
Brooks. Procedural Reasoning Systems. Situated Automata and
Situated Agents. Disadvantages and Next Step.
3. Integration of Planning and Reaction. The Planning to Reaction
Spectrum. The Contingency Space. The Need for Deciding
Between Planning and Reaction.
4. Deciding at Planning Time. Alternative Approaches.
Empirical Approach. Analytical Approach. The Need for Good
Real-time Components.
5. Execution Time: Reacting, (Re)Planning and
Deciding. Anytime Algorithms.
Action-based Decision Trees.
6. Real-time Planning. Guaranteeing Temporal
Adequacy. Building Guaranteed And Unguaranteed Plans.
7. Applications. A Language Framework for
Describing Reactive Behaviors. Formalizing Reaction
Decisions in Critical Domains for Pedagogical
Purposes.
8. Conclusion. Integration of the Ideas.
Language Development. Integration with Existing
Technologies. Development of Hardware.
index.