Windowed MAPF with Completeness Guarantees

Authors: Rishi Veerapaneni, Muhammad Suhail Saleem, Jiaoyang Li, Maxim Likhachev

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments demonstrate the empirical performance of our theoretically complete SS-CBS algorithm. We first evaluate SS-CBS on standard benchmark maps (Stern et al. 2019) and observe that SS-CBS is indeed able to outperform the windowed baselines. We then evaluate SS-CBS on high-congestion small maps and showcase SS-CBS’s superiority in this regime. Our appendix contains additional analysis and results.
Researcher Affiliation Academia Carnegie Mellon University EMAIL
Pseudocode Yes Algorithm 1: Windowed Complete MAPF Framework
Open Source Code No The paper mentions "All algorithms were implemented in C++" but does not provide any explicit statement about open-source code availability, a repository link, or reference to supplementary materials containing code for the described methodology.
Open Datasets Yes We first evaluate SS-CBS on standard benchmark maps (Stern et al. 2019) and observe that SS-CBS is indeed able to outperform the windowed baselines. We then evaluate SS-CBS on high-congestion small maps and thus evaluated it on small maps with high congestion from Okumura (2022).
Dataset Splits No The paper describes running experiments on
Hardware Specification Yes All algorithms were implemented in C++ and run on a PC with a 2.30 GHz Intel i7-11800K CPU.
Software Dependencies No The paper states, "All algorithms were implemented in C++" but does not specify any version numbers for C++ compilers, libraries, or other software components used.
Experiment Setup Yes We evaluate on 4 benchmark maps (each column: random-32-32-20, ht chantry, den520d, warehouse-10-20-10-2-1) with a 1 minute timeout across 25 scenarios. We additionally end w CBS in failure when it repeats previously visited configurations 100 times (i.e. deadlock or livelock). ... We compare against CBS with window W = {1, 2, 4, 8, 16} (w CBS). ... Table 1 shows the average success from 20 seeds with a timeout of 1 minute...