Using classical planners to solve conformant probabilistic planning problems

Motivated by the success of the translation-based approach for conformant planning, introduced by Palacios and Geffner, we present two variants of a new compilation scheme from conformant probabilistic planning problems (CPP) to variants of classical planning. In CPP, we are given a set of actions - which we assume to be deterministic in this paper, a distribution over initial states, a goal condition, and a value 0 < p ≤ 1. Our task is to find a plan π such that the goal probability following the execution of π in the initial state is at least p. Our first variant translates CPP into classical planning with resource constraints, in which the resource represents probabilities of failure. The second variant translates CPP into cost-optimal classical planning problems, in which costs represents probabilities. Empirically, these techniques show mixed results, performing very well on some domains, and poorly on others. This indicates that compilation-based technique are a feasible and promising direction for solving CPP problems and, possibly, more general probabilistic planning problems.