The Computational Complexity of Propositional STRIPS Planning：（命题条规划的计算复杂度）.pdfVIP

The Computational Complexity of Prop ositional STRIPS Planning Tom Bylander Division of Mathematics Computer Science and Statistics University of Texas at San Antonio San Antonio Texas USA email bylanderringercsutsaedu phone March to app ear in Articial Intel ligenc e Abstract I present several computational complexity results for prop ositiona l STRIPS planning ie STRIPS planning restricted to ground formulas Dierent planning problems can b e dened by restricting the typ e of formulas placing limits on the numb er of pre and p ostconditions by restricting negation in pre and p ostconditions and by requiring optimal plans For these typ es of restrictions I show when planning is tractable p olynomial and intractable NP hard In general it is PSPA CEcomplete to determine if a given planning instance has any solutions Extremely severe restrictions on b oth the op erators and the formulas are required to guarantee p olynomial time or even NPcompleteness For example when only ground literals are p ermitted determining plan existence is PSPA CEcomplete even if op erators are limited to two preconditions and two p ostconditions When denite Horn ground formulas are p ermitted determining plan existence is PSPA CEcomplete even if op erators are limited to zero preconditions and one p ostcondition One of the interesting tractable problems is if each op erator is restricted to p ositive preconditions and one p ostcondition only ground literals The blo cksworld problem slightly mo died is a sub