Browse Theses

As the limitations of traditional AI plan representations have become apparent, researchers have turned to temporal logic as one means of providing rich languages for structuring planning knowledge. But these languages lack the ability to represent uncertainty, as well as lacking any kind of link to a theory of rational behavior. Decision Theory provides a normative model of choice under uncertainty such that recommended choices can be said to be rational under a well-defined notion of rationality. But traditional decision-theoretic representations are limited in their ability to structure knowledge other than beliefs and preferences.This thesis integrates AI and decision-theoretic approaches to the representation of planning problems by developing a first-order logic of time, chance, and action for representing and reasoning about plans. The semantics of the logic incorporates intuitive properties of time, chance, and action central to the planning problem. The logical language integrates both modal and probabilistic constructs and allows quantification over time points, probability values, and domain individuals. Probability is treated as a sentential operator in the language, so it can be arbitrarily nested and combined with other logical operators. The language can represent the chance that facts hold and events occur at various times. It can represent the chance that actions and other events affect the future. The model of action distinguishes between action feasibility, executability, and effects. Using this distinction, a notion of expected utility for acts that may not be feasible is defined. This notion is used to reason about the chance that trying a plan will achieve a given goal. An algorithm for the problem of building construction planning is developed and the logic is used to prove the algorithm correct.