Redundant Decompositions in PO HTN Domains: Goto Considered Harmful
Résumé
HTN planning is a widely used approach for solving planning problems by breaking them down into smaller sub-problems.
This approach is often motivated by the ability to add constraints between tasks, which can guide the search towards a solution and improve performance by reducing the search space. In this paper, we identify a common pattern in PO HTN planning that can lead to a pathological explosion of the search space, resulting in a significant decrease in computational performance. However, this is not a fatal issue. Alternative HTN models can be used to reduce the search space. We propose two models that maintain the expressiveness of the original problem while reducing the number of possible decompositions. Our results demonstrate improved computational performance on IPC benchmarks.
Domaines
Intelligence artificielle [cs.AI]Origine | Fichiers produits par l'(les) auteur(s) |
---|