Abstraction can be an effective technique for dealing with the complexity of planning tasks. This paper is aimed at assessing and identifying in which cases abstraction can actually speed-up the overall search. In fact, it is well known that the impact of abstraction on the time spent to search for a solution of a planning problem can be positive or negative, depending on several factors -including the number of objects defined in the domain, the branching factor, and the plan length. Experimental results highlight the role of such aspects on the overall performance of an algorithm that performs the search at the ground-level only, and compares them with the ones obtained by enforcing abstraction.
A Critical Look at the Abstraction Based on Macro-Operators
ARMANO, GIULIANO;VARGIU, ELOISA
2004-01-01
Abstract
Abstraction can be an effective technique for dealing with the complexity of planning tasks. This paper is aimed at assessing and identifying in which cases abstraction can actually speed-up the overall search. In fact, it is well known that the impact of abstraction on the time spent to search for a solution of a planning problem can be positive or negative, depending on several factors -including the number of objects defined in the domain, the branching factor, and the plan length. Experimental results highlight the role of such aspects on the overall performance of an algorithm that performs the search at the ground-level only, and compares them with the ones obtained by enforcing abstraction.File | Dimensione | Formato | |
---|---|---|---|
AIxIA-2004-critical look.pdf
accesso aperto
Tipologia:
versione post-print (AAM)
Dimensione
182.73 kB
Formato
Adobe PDF
|
182.73 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.