Progression Heuristics for Planning with Probabilistic LTL Constraints

Authors: Ian Mallett, Sylvie Thiebaux, Felipe Trevizan11870-11879

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments show that they further widen the scalability gap between heuristic search and verification approaches to these planning problems. [...] Section 6 gives experimental results and Section 7 concludes with related and future work.
Researcher Affiliation Academia Ian Mallett, Sylvie Thi ebaux, Felipe Trevizan Research School of Computer Science, The Australian National University Ian.Mallett@anu.edu.au, Sylvie.Thiebaux@anu.edu.au, Felipe.Trevizan@anu.edu.au
Pseudocode No The paper describes mathematical formulations and processes using equations and text, but it does not provide any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No Both PPDDL and PRISM versions of the problems are available at https://gitlab.com/fwt/mo-pltl-ssps-benchmarks. This link provides access to the problem definitions/benchmarks, not the source code for the authors' implemented methodology.
Open Datasets Yes For evaluation, we use the Factory and Wall-e domains from (Baumgartner, Thi ebaux, and Trevizan 2018), and a new domain called Priority Search. [...] Both PPDDL and PRISM versions of the problems are available at https://gitlab.com/fwt/mo-pltl-ssps-benchmarks.
Dataset Splits No The paper does not describe dataset splits for training, validation, or testing in the typical machine learning sense. It evaluates planning problem solvers on defined problem instances.
Hardware Specification Yes The experiments were ran on an Intel i7-7700@3.6GHz using Gurobi 8.1.1 on a single thread and a 20mins and 4Gb cutoff.
Software Dependencies Yes The experiments were ran on an Intel i7-7700@3.6GHz using Gurobi 8.1.1 on a single thread and a 20mins and 4Gb cutoff.
Experiment Setup Yes The experiments were ran on an Intel i7-7700@3.6GHz using Gurobi 8.1.1 on a single thread and a 20mins and 4Gb cutoff. We used the default options of PRISM and the -lp flag to use their LP approach to MO-PLTL SSPs because, without this flag, PRISM was unable to solve any of our benchmarks.