Decision Theoretic Planning: Structural Assumptions and Computational Leverage

Craig Boutilier
Department of Computer Science
University of British Columbia
Vancouver, BC, CANADA, V6T 1Z4

Thomas Dean
Department of Computer Science
Brown University
Providence, RI 02912, U.S.A.

Steve Hanks
Harlequin Inc.
1201 Third Avenue, Suite 2380
Seattle, WA 98101 U.S.A.

Planning under uncertainty is a central problem in the study of automated sequential decision making, and has been addressed by researchers in many different fields, including AI planning, decision analysis, operations research, control theory and economics. While the assumptions and perspectives adopted in these fields often differ in substantial ways, many planning problems of interest to researchers in these fields can be modeled as Markov decision processes (MDPs) and analyzed using the techniques of decision theory.

This paper presents an overview and synthesis of MDP-related methods showing how they provide a unifying framework for modeling many classes of planning problems studied in AI. It also describes structural properties of MDPs that, when exhibited by particular classes of problems, can be exploited in the construction of optimal or approximately optimal policies or plans. Planning problems commonly possess structure in the reward and value functions used to describe performance criteria, in the functions used to describe state transitions and observations, and in the relationships among features used to describe states, actions, rewards, and observations.

Specialized representations, and algorithms employing these representations, can achieve computational leverage by exploiting these various forms of structure. Certain AI techniques---in particular those based on the use of structured, intensional representations---can be viewed in this way. This paper surveys several types of representations for both classical and decision theoretic planning problems, and planning algorithms that exploit these representations in a number of different ways to ease the computational burden of constructing policies or plans. It focuses primarily on abstraction, aggregation and decomposition techniques based on AI-style representations.

Journal of Arificial Intelligence Research, Volume 11, 1999

Return to List of Papers