Learning Domain-Independent Heuristics for Grounded and Lifted Planning - LAAS-Robotique
Conference Papers Year : 2024

Learning Domain-Independent Heuristics for Grounded and Lifted Planning

Abstract

We present three novel graph representations of planning tasks suitable for learning domain-independent heuristics using Graph Neural Networks (GNNs) to guide search. In particular, to mitigate the issues caused by large grounded GNNs we present the first method for learning domain-independent heuristics with only the lifted representation of a planning task. We also provide a theoretical analysis of the expressiveness of our models, showing that some are more powerful than STRIPS-HGN, the only other existing model for learning domain-independent heuristics. Our experiments show that our heuristics generalise to much larger problems than those in the training set, vastly surpassing STRIPS-HGN heuristics.
Fichier principal
Vignette du fichier
aaai24.pdf (397.12 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04561847 , version 1 (14-07-2024)

Licence

Copyright

Identifiers

Cite

Dillon Z Chen, Sylvie Thiébaux, Felipe Trevizan. Learning Domain-Independent Heuristics for Grounded and Lifted Planning. 38th AAAI Conference on Artificial Intellifence (AAAI 2024), Association for the Advancement of Artificial Intelligence, Feb 2024, Vancouver, Canada. pp.20078-20086, ⟨10.1609/AAAI.V38I18.29986⟩. ⟨hal-04561847⟩
157 View
28 Download

Altmetric

Share

More