Online Multi-Agent Pathfinding

Authors: Jiří Švancara, Marek Vlk, Roni Stern, Dor Atzmon, Roman Barták7732-7739

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

Reproducibility Variable Result LLM Response
Research Type Experimental We implemented all the algorithms and evaluated their performance on a set of randomly generated problems.
Researcher Affiliation Academia 1Faculty of Mathematics and Physics, Charles University, Prague, The Czech Republic 2Czech Institute of Informatics, Robotics and Cybernetics, Czech Technical University in Prague, The Czech Republic 3Information Systems Engineering, Ben Gurion University, Be er Sheba, Israel
Pseudocode Yes Algorithm 1 Replan Single
Open Source Code No The paper does not provide an explicit statement or link for the open-sourcing of their implemented solvers.
Open Datasets No We created two datasets of online MAPF problems based on 4-connected grids designed to simulate intersections for autonomous vehicles.
Dataset Splits No The paper describes generating datasets but does not provide specific train/validation/test split percentages, sample counts, or references to predefined splits.
Hardware Specification Yes All experiments were conducted on Dell PC with an Intel R Core TM i7-2600K processor running at 3.40 GHz with 8 GB of RAM.
Software Dependencies Yes We created two such modified optimal offline MAPF solvers: one based on a reduction to Boolean satisfiability (SAT) via the Picat language and compiler (version 2.2#3) (Bart ak et al. 2017), and the other based on the Conflict-Based Search (CBS) algorithm (Sharon et al. 2015).
Experiment Setup Yes Whenever the runtime for a newly appeared agent exceeds the given time limit, which was set to 30 seconds, the output from RS is taken into results.