Why Can’t You Do That HAL? Explaining Unsolvability of Planning Tasks

Authors: Sarath Sreedharan, Siddharth Srivastava, David Smith, Subbarao Kambhampati

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical analysis and user studies show the validity of our methods as well as their computational efficacy on a number of benchmark planning domains.
Researcher Affiliation Collaboration 1CIDSE, Arizona State University, Tempe, AZ 85281 USA 2PSresearch ssreedh3@asu.edu, siddharths@asu.edu, david.smith@psresearch.xyz, rao@asu.edu
Pseudocode No The paper describes methods and formulations but does not include a clearly labeled 'Pseudocode' or 'Algorithm' block with structured steps.
Open Source Code No The paper provides a link to supplementary material for study setup ('http://bit.ly/2HQ5sTv') but does not provide a link or explicit statement about the availability of the source code for its described methodology.
Open Datasets Yes The first five domains and their problem instances consisted of standard IPC domains and problem instances used in previous IPC competitions [International Planning Competition, 2011]. ... The next three domains were selected from the set used for the 2016 unsolvability competition [Unsolvability International Planning Competition, 2016].
Dataset Splits No The paper evaluates its methods on standard planning benchmarks but does not specify explicit training, validation, or test dataset splits for these problems.
Hardware Specification No The paper does not provide any specific hardware details such as GPU or CPU models, or cloud computing specifications used for running the experiments.
Software Dependencies No The paper mentions using 'the fast-downward implementation of [Keyder, Richter, and Helmert, 2010]' but does not provide a specific version number for this or any other software dependency.
Experiment Setup Yes All instances were run with a timeout of 100 minutes (all problems were solvable under this time limit) and all landmarks were generated using the fast-downward implementation of [Keyder, Richter, and Helmert, 2010] (where we set the subset sizes to one for the first five domains and to two for the rest).