Managing Infinite Abstractions in Numeric Pattern Database Heuristics
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (43) · 36227-36235
DOI: 10.1609/aaai.v40i43.40941 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Markus Fritzsche und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Pattern Database (PDB) heuristics are an established approach in optimal classical planning that is used in state-of-the-art planning systems. PDBs are based on projections, which induce an abstraction of the original problem. Computing all cheapest plans in the abstraction yields an admissible heuristic. Despite their success, PDBs have only recently been adapted to numeric planning, which extends classical planning with numeric state variables. The difficulty in supporting numeric variables is that the induced abstractions, in contrast to classical planning, are generally infinite. Thus, they cannot be explored exhaustively to compute a heuristic. The foundational work that introduced numeric PDBs employed a simple approach that computes only a finite part of the abstraction. We analyze this framework and identify cases where it necessarily results in an uninformed heuristic. We propose several improvements over the basic variant of numeric PDBs that lead to enhanced heuristic accuracy.
Zitieren
APA
Fritzsche, M., Gnad, D., Gruntov, M., & Shleyfman, A. (2026). Managing Infinite Abstractions in Numeric Pattern Database Heuristics. Proceedings of the AAAI Conference on Artificial Intelligence, 40(43), 36227-36235. https://doi.org/10.1609/aaai.v40i43.40941
Vancouver
Fritzsche M, Gnad D, Gruntov M, Shleyfman A. Managing Infinite Abstractions in Numeric Pattern Database Heuristics. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(43):36227-36235. doi:10.1609/aaai.v40i43.40941
BibTeX
@article{Fritzsche2026,
author = {Fritzsche, Markus and Gnad, Daniel and Gruntov, Mikhail and Shleyfman, Alexander},
title = {Managing Infinite Abstractions in Numeric Pattern Database Heuristics},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {43},
pages = {36227--36235},
doi = {10.1609/aaai.v40i43.40941}
}
EUVIMED