| | |
Summary: PLANNING VIA HEURISTIC SEARCH
IN A
PETRI NkT FRAMEWORK.
K.M.Passim and PJ. A n W s
Depaitmentof Electricaland computerEnginewing
University of Notre Dame. None DameIN46556
Abstract
~n Artifiiial Intelligenceplanningsystrm's maincomponent6consistof a P~IUUX
and a problem domain. The problem Ppmain is Uy environment that the planner
reasons about and takes actions on. 'a"utis papd. a special type of Extended
Input/OutputPetrinet isddmsdandthenusad a8theproblemrepmentatirmfora wide
class of problem domains. A plarming strategy is developed using results fromthe
theory of heuristic seach. Inp.rticulpr.usmg thedevelopedPetri net frmnewmk and
metric spaces, a class of heuristicfunctionsis 6pSfikd that are both &anisible and
consistentforthe A* algorithm as
an illustrationoftheremlbtwo Simpieplsnningproblunsanmodelledmd solved.
plarmingsystemarchitectweis discussed
1.0 Introduction
According to the viewpoint presented in [13], Artificial Intelligence (AI)
planning systems consist of a planner snd a problem domain, their
|
