Formulation of Tradeoffs in Planning Under Uncertainty

Abstract

Planning under uncertainty with multiple, competing objectives is impossible when goals are represented as predicates and the effects of actions are modeled as deterministic functions of situations. Decision-theoretic models, on the other hand, do not address the problem of constructing strategies from more primitive representations of actions. In this proposal, I describe a method for formulating plans from large knowledge bases that can accomodate uncertain and partial satisifaction of goals. At the core of the planner is a dominance prover that derives admissibility properties of plan classes. The representation for the effects of actions is based on a qualitative formalism for asserting influences among variables. The planner makes decisions "up to tradeoffs," an intuitive description that seems to characterize the power of a dominance prover based on the qualitative influence formalism.