Determining Action Reversibility in STRIPS Using Answer Set Programming with Quantifiers

Abstract

AbstractIn the field of automated planning, an action is called reversible when other actions can be applied in order to revert the effects of this action and return to the original state. In recent years, there has been renewed interest in this topic, which led to novel results in the widely known STRIPS formalism and the PDDL planning language.In this paper, we aim to solve the computational problem of deciding action reversibility in a practical setting, applying recent advances in the field of logic programming. In particular, a quantified extension of Answer Set Programming (ASP) named ASP with Quantifiers (ASP(Q)) has been proposed by Amendola, Ricca, and Truszczynski, which allows for stacking logic programs by quantifying over answer sets of the previous layer. This language is well-suited to express encodings for the action reversibility problem, since this problem naturally contains a quantifier alternation. In addition, a prototype solver for ASP(Q) is currently developed. We make use of the ASP(Q) language to offer an encoding for action reversibility, and then report on preliminary benchmark results on how well this encoding performs compared to classical ASP.KeywordsAutomated planningAnswer set programmingReasoning about action and change