Decoupled Abstraction Heuristics
Title: Decoupled Abstraction Heuristics
SNIC Project: SNIC 2022/22-1038
Project Type: SNIC Small Compute
Principal Investigator: Daniel Gnad <daniel.gnad@liu.se>
Affiliation: Linköpings universitet
Duration: 2022-10-26 – 2023-11-01
Classification: 10201
Keywords:

Abstract

This work is in the context of classical planning as heuristic search, where a state space is systematically explored to find a sequence of actions that lead from the initial state to a state that satisfies the goal condition. More concretely, in this project we investigate abstraction heuristics for decoupled state-space search. The goal is to derive new methods for abstraction heuristics for decoupled search and evaluate them empirically, comparing to existing approaches.