edoc

Reformulating Oversubscription Planning Tasks

Katz, Michael and Mirkis, Vitaly and Pommerening, Florian and Winterer, Dominik. (2018) Reformulating Oversubscription Planning Tasks. In: Proceedings of the 10 th Workshop on Heuristics and Search for Domain-independent Planning (HSDIP). Delft, pp. 81-88.

[img] PDF - Published Version
415Kb

Official URL: https://edoc.unibas.ch/65634/

Downloads: Statistics Overview

Abstract

Most modern heuristics for classical planning are specified in terms of minimizing the summed operator costs. Heuristics for oversubscription planning (OSP), on the other hand, maximize the utility on states. In this work we aim to provide the grounds for the adaptation of existing heuristics for classical planning to the OSP setting. To this end, we reformulate the OSP task to a classical planning task extended with an additional operator costs function, reflecting the utility information fully. We exemplify how existing heuristics from classical planning can be adapted to such a setting with a merge-and-shrink heuristic and empirically validate the feasibility of our approach.
Faculties and Departments:05 Faculty of Science > Departement Mathematik und Informatik > Informatik > Artificial Intelligence (Helmert)
UniBasel Contributors:Pommerening, Florian
Item Type:Conference or Workshop Item, refereed
Conference or workshop item Subtype:Conference Paper
Publisher:ICAPS
Note:Publication type according to Uni Basel Research Database: Conference paper
Language:English
Related URLs:
Last Modified:29 Jan 2019 13:27
Deposited On:15 Oct 2018 09:45

Repository Staff Only: item control page