Front-to-Front Heuristic Search for Satisficing Classical Planning

Authors: Ryo Kuroiwa, Alex Fukunaga

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally evaluated the following search algorithms, all of which were implemented by modifying the Fast Downward planning system.
Researcher Affiliation Academia Ryo Kuroiwa and Alex Fukunaga Graduate School of Arts and Sciences, The University of Tokyo mhgeoe@gmail.com, fukunaga@idea.c.u-tokyo.ac.jp
Pseudocode Yes Algorithm 1 Top-to-Top Bidirectional Search (TTBS)
Open Source Code No All code will be made available on a public repository.
Open Datasets Yes We used 1229 solvable instances in 47 planning domains from classical satisficing tracks in International Planning Competition (IPC) 98-2018, whose SAS+ representations do not contain axioms and conditional effects.
Dataset Splits No The paper evaluates performance on benchmark instances but does not describe a typical train/validation/test split for a dataset.
Hardware Specification Yes All runs were given a 5 min. time limit and a 4GB memory limit on Intel(R) Xeon(R) CPU E5-2670 v3 @ 2.30GHz processors.
Software Dependencies No The paper mentions software like 'Fast Downward planning system (FD)' and 'FDr', but does not provide specific version numbers for these or other ancillary software components, which are required for a reproducible description.
Experiment Setup Yes Differently from Alc azar et al. (2014), we used eager GBFS instead of lazy GBFS and did not use preferred operators. All methods used the unit-cost version of hff [Hoffmann and Nebel, 2001], which is straightforwardly applicable in regression and bidirectional search [Alc azar et al., 2013; Alc azar et al., 2014], using eager evaluation, and the FIFO tie-breaking strategy. All runs were given a 5 min. time limit and a 4GB memory limit.