Deciding Unsolvability in Temporal Planning under Action Non-Self-Overlapping

Authors: Stefan Panjkovic, Andrea Micheli, Alessandro Cimatti9886-9893

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We implemented the approaches, and carried out an experimental evaluation against other stateof-the-art TP tools. ... We performed a thorough experimental evaluation on both solvable and unsolvable benchmarks, comparing our approaches with the state-of-the-art tools.
Researcher Affiliation Academia Stefan Panjkovic, Andrea Micheli, Alessandro Cimatti Fondazione Bruno Kessler, Trento, Italy {spanjkovic, amicheli, cimatti}@fbk.eu
Pseudocode Yes Algorithm 1: The pseudo-code of TAMER-CTP
Open Source Code Yes full encoding details, all implementations and benchmarks are available in (Panjkovic, Micheli, and Cimatti 2022). ... Panjkovic, S.; Micheli, A.; and Cimatti, A. 2022. Deciding Unsolvability in Temporal Planning under Action Non-Self Overlapping: Additional Material. https://es.fbk.eu/people/amicheli/resources/aaai22. Accessed: 2022-03-22.
Open Datasets Yes We sourced the solvable benchmarks from (Valentini, Micheli, and Cimatti 2020): the set includes standard temporal IPC instances (Vallati et al. 2015)... Moreover, we considered the MATCHCELLAR IPC domain (Vallati et al. 2015).
Dataset Splits No The paper uses a set of solvable and unsolvable 'benchmarks' but does not specify explicit training, validation, or test dataset splits (e.g., percentages or counts) or cross-validation schemes.
Hardware Specification Yes All the experiments were performed on a Core i9-9900KS with 1800s/20GB of time and memory limit.
Software Dependencies No The paper mentions software tools like NUXMV and UPPAAL and that implementation was in C++ but does not provide specific version numbers for any software, libraries, or compilers used.
Experiment Setup Yes All the experiments were performed on a Core i9-9900KS with 1800s/20GB of time and memory limit. ... We employ an exploration algorithm based on A* (Hart, Nilsson, and Raphael 1968). Every state s has two values: G(s) that is the length of the path from the initial state to s, and HADD(s) that is the heuristic value of s computed using the standard hadd heuristic (Bonet and Geffner 2001).