POGEMA: A Benchmark Platform for Cooperative Multi-Agent Pathfinding
Alexey Skrynnik, Anton Andreychuk, Anatolii Borzilov, Alexander Chernyavskiy, Konstantin S. Yakovlev, Aleksandr Panov
Abstract
Multi-agent reinforcement learning (MARL) has recently excelled in solving challenging cooperative and competitive multi-agent problems in various environments, typically involving a small number of agents and full observability. Moreover, a range of crucial robotics-related tasks, such as multi-robot pathfinding, which have traditionally been approached with classical non-learnable methods (e.g., heuristic search), are now being suggested for solution using learning-based or hybrid methods. However, in this domain, it remains difficult, if not impossible, to conduct a fair comparison between classical, learning-based, and hybrid approaches due to the lack of a unified framework that supports both learning and evaluation. To address this, we introduce POGEMA, a comprehensive set of tools that includes a fast environment for learning, a problem instance generator, a collection of predefined problem instances, a visualization toolkit, and a benchmarking tool for automated evaluation. We also introduce and define an evaluation protocol that specifies a range of domain-related metrics, computed based on primary evaluation indicators (such as success rate and path length), enabling a fair multi-fold comparison. The results of this comparison, which involves a variety of state-of-the-art MARL, search-based, and hybrid methods, are presented.
(a) task
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
5 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- CAMAR: Continuous Actions Multi-Agent Routing cites MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale
- PettingZoo: Gym for Multi-Agent Reinforcement Learning cites Leveraging Procedural Generation to Benchmark Reinforcement Learning
- Pairwise is Not Enough: Hypergraph Neural Networks for Multi-Agent Pathfinding cites MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale
- Graph Attention-Guided Search for Dense Multi-Agent Pathfinding cites MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale
Cited by top-tier papers5
- MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at ScaleAnton Andreychuk, Konstantin S. Yakovlev, Aleksandr Panov, Alexey SkrynnikAAAI 2025 · 19 citations
- Pairwise is Not Enough: Hypergraph Neural Networks for Multi-Agent PathfindingRishabh Jain, Keisuke Okumura, Michael Amir, Pietro Lio et al.ICLR 2026 · 6 citations
- Graph Attention-Guided Search for Dense Multi-Agent PathfindingRishabh Jain, Keisuke Okumura, Michael Amir, Amanda ProrokAAAI 2026 · 4 citations
- CAMAR: Continuous Actions Multi-Agent RoutingArtem Pshenitsyn, Aleksandr Panov, Alexey SkrynnikAAAI 2026 · 2 citations
- SoMe: A Realistic Benchmark for LLM-based Social Media AgentsDizhan Xue, Jing Cui, Shengsheng Qian, Chuanrui Hu et al.AAAI 2026 · 1 citation
Builds on16
- Emergent Tool Use From Multi-Agent AutocurriculaBowen Baker, Ingmar Kanitscheider, Todor M. Markov, Yi Wu et al.ICLR 2020 · 751 citations
- Leveraging Procedural Generation to Benchmark Reinforcement LearningKarl Cobbe, Christopher Hesse, Jacob Hilton, John SchulmanICML 2020 · 685 citations
- QPLEX: Duplex Dueling Multi-Agent Q-LearningJianhao Wang, Zhizhou Ren, Terry Liu, Yang Yu et al.ICLR 2021 · 595 citations
- Google Research Football: A Novel Reinforcement Learning EnvironmentKarol Kurach, Anton Raichuk, Piotr Stanczyk, Michal Zajac et al.AAAI 2020 · 496 citations
- PettingZoo: Gym for Multi-Agent Reinforcement LearningJ. K. Terry, Benjamin Black, Nathaniel Grammel, Mario Jayakumar et al.NeurIPS 2021 · 478 citations
Related papers
- TaskForce: Cooperative Multi-agent Reinforcement Learning for Multi-task OptimizationWonhyeok Choi, Kyumin Hwang, Jihun Park, Kyoungmin Lee et al.CVPR 2026
- Memory, Benchmark & Robots: A Benchmark for Solving Complex Tasks with Reinforcement LearningEgor Cherepanov, Nikita Kachaev, Alexey K. Kovalev, Aleksandr I. PanovICLR 2026 · 43 citations
- BALROG: Benchmarking Agentic LLM and VLM Reasoning On GamesDavide Paglieri, Bartlomiej Cupial, Samuel Coward, Ulyana Piterbarg et al.ICLR 2025
- Prediction-Guided Multi-Objective Reinforcement Learning for Continuous Robot ControlJie Xu, Yunsheng Tian, Pingchuan Ma, Daniela Rus et al.ICML 2020 · 210 citations
- MultiAgentBench : Evaluating the Collaboration and Competition of LLM agentsKunlun Zhu, Hongyi Du, Zhaochen Hong, Xiaocheng Yang et al.ACL 2025 · 97 citations
