Rolling in Classical Planning with Conditional Effects and Constraints

Authors: Matteo Cardellini, Enrico Giunchiglia

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We prove that our approach is correct and complete, and run an experimental analysis on novel domains where a plan must have repetitions of an action. The analysis confirms our theoretical results and, by comparing to the case where rolling is disabled, confirms the benefits of rolling.
Researcher Affiliation Academia Matteo Cardellini , Enrico Giunchiglia DIBRIS, University of Genova, Genova, Italy EMAIL
Pseudocode Yes Algorithm 1 Computation of the rolling of a needed to reach s from s. 1: global T 0 a , . . . , T p a , R0 a, . . . , Rp a //previously computed 2: function REPETITIONS(a, s, s , p) 3: if T 0 a (s, s ) then 4: return 1 5: for j [1, p] do 6: if T j a(s, s ) then 7: s s : Rj 1 a (s, s ) 8: return 2j 1 + REPETITIONS(a, s , s , j 1) 9: return -1
Open Source Code No The paper mentions using 'PATTY solver3' and provides a URL 'https://pattyplan.com', which appears to be a project page, and 'pyeda tool4' with 'https://pyeda.readthedocs.io/'. However, it does not explicitly state that the source code for the methodology described in *this specific paper* is being released or is available at these links. The PATTY solver is mentioned as a tool they implemented *on*, and pyeda is a third-party tool they *employed*. There is no direct statement from the authors providing their own code for this paper.
Open Datasets No The paper introduces and defines three novel domains ('Counter', 'Cube', 'Loops') for experimental analysis. These are problem definitions and not pre-existing publicly available datasets in the traditional sense, and no specific access information (link, DOI, citation to a data repository) is provided for specific data instances or files from these domains.
Dataset Splits No The paper defines new problem domains (Counter, Cube, Loops) for classical planning, which involve generating problems based on parameters (e.g., number of bits B, cube side L, number of loops K). It does not mention using pre-existing datasets with defined training/testing/validation splits, nor does it specify how any generated problem instances are partitioned for these purposes.
Hardware Specification Yes Each problem and encoding has a time limit of 10 minutes on an Intel Xeon Platinum 8000 3.1GHz with 8GB of RAM.
Software Dependencies Yes We implemented the approach on the PATTY solver3 [Cardellini et al., 2024] which makes use of Z3 v4.12.2 [de Moura and Bjørner, 2008] to compute the model (if any) satisfying Π+ n and Π0 n.
Experiment Setup Yes Each problem and encoding has a time limit of 10 minutes on an Intel Xeon Platinum 8000 3.1GHz with 8GB of RAM. ... We limit the computation of all actions TCs to 3 minutes, equally divided among all actions, and for each action we return the last computed transition relation before the time limit.