LC-MAPF provides iterative message exchange among agents to enable<br>progressive refinement of predicted action distributions over<br>multiple communication rounds.</p>\n<p><a href=\"https://cdn-uploads.huggingface.co/production/uploads/65c0db0fbda79a18292dfbb7/vDuPmTvtpOdDdVwxA8Y27.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/65c0db0fbda79a18292dfbb7/vDuPmTvtpOdDdVwxA8Y27.png\" alt=\"image\"></a></p>\n","updatedAt":"2026-05-15T11:14:45.215Z","author":{"_id":"65c0db0fbda79a18292dfbb7","avatarUrl":"/avatars/1201b8282664c2d8c18beaba2396c03b.svg","fullname":"Alsu Sagirova","name":"alsu-sagirova","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.6383206844329834},"editors":["alsu-sagirova"],"editorAvatarUrls":["/avatars/1201b8282664c2d8c18beaba2396c03b.svg"],"reactions":[],"isReport":false}},{"id":"6a0700e58b2b577e300cc3b0","author":{"_id":"65c0db0fbda79a18292dfbb7","avatarUrl":"/avatars/1201b8282664c2d8c18beaba2396c03b.svg","fullname":"Alsu Sagirova","name":"alsu-sagirova","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false},"createdAt":"2026-05-15T11:17:57.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"Experimental evaluations show that LC-MAPF outperforms SOTA learnable MAPF approaches.\n\n\n\n","html":"<p>Experimental evaluations show that LC-MAPF outperforms SOTA learnable MAPF approaches.</p>\n<p><a href=\"https://cdn-uploads.huggingface.co/production/uploads/65c0db0fbda79a18292dfbb7/2Blje08ULnEx9-mZ5h3RG.png\" rel=\"nofollow\"><img src=\"https://cdn-uploads.huggingface.co/production/uploads/65c0db0fbda79a18292dfbb7/2Blje08ULnEx9-mZ5h3RG.png\" alt=\"image\"></a></p>\n","updatedAt":"2026-05-15T11:17:57.123Z","author":{"_id":"65c0db0fbda79a18292dfbb7","avatarUrl":"/avatars/1201b8282664c2d8c18beaba2396c03b.svg","fullname":"Alsu Sagirova","name":"alsu-sagirova","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.6991381049156189},"editors":["alsu-sagirova"],"editorAvatarUrls":["/avatars/1201b8282664c2d8c18beaba2396c03b.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.07637","authors":[{"_id":"6a06ff723192c37877924fbf","name":"Valeriy Vyaltsev","hidden":false},{"_id":"6a06ff723192c37877924fc0","name":"Alsu Sagirova","hidden":false},{"_id":"6a06ff723192c37877924fc1","name":"Anton Andreychuk","hidden":false},{"_id":"6a06ff723192c37877924fc2","name":"Oleg Bulichev","hidden":false},{"_id":"6a06ff723192c37877924fc3","name":"Yuri Kuratov","hidden":false},{"_id":"6a06ff723192c37877924fc4","name":"Konstantin Yakovlev","hidden":false},{"_id":"6a06ff723192c37877924fc5","name":"Aleksandr Panov","hidden":false},{"_id":"6a06ff723192c37877924fc6","name":"Alexey Skrynnik","hidden":false}],"publishedAt":"2026-05-12T00:00:00.000Z","submittedOnDailyAt":"2026-05-15T00:00:00.000Z","title":"Learning to Communicate Locally for Large-Scale Multi-Agent Pathfinding","submittedOnDailyBy":{"_id":"65c0db0fbda79a18292dfbb7","avatarUrl":"/avatars/1201b8282664c2d8c18beaba2396c03b.svg","isPro":false,"fullname":"Alsu Sagirova","user":"alsu-sagirova","type":"user","name":"alsu-sagirova"},"summary":"Multi-agent pathfinding (MAPF) is a widely used abstraction for multi-robot trajectory planning problems, where multiple homogeneous agents move simultaneously within a shared environment. Although solving MAPF optimally is NP-hard, scalable and efficient solvers are critical for real-world applications such as logistics and search-and-rescue. To this end, the research community has proposed various decentralized suboptimal MAPF solvers that leverage machine learning. Such methods frame MAPF (from a single agent perspective) as a Dec-POMDP where at each time step an agent has to decide an action based on the local observation and typically solve the problem via reinforcement learning or imitation learning. We follow the same approach but additionally introduce a learnable communication module tailored to enhance cooperation between agents via efficient feature sharing. We present the Local Communication for Multi-agent Pathfinding (LC-MAPF), a generalizable pre-trained model that applies multi-round communication between neighboring agents to exchange information and improve their coordination. Our experiments show that the introduced method outperforms the existing learning-based MAPF solvers, including IL and RL-based approaches, across diverse metrics in a diverse range of (unseen) test scenarios. Remarkably, the introduced communication mechanism does not compromise LC-MAPF's scalability, a common bottleneck for communication-based MAPF solvers.","upvotes":16,"discussionId":"6a06ff723192c37877924fc7","ai_summary":"Multi-agent pathfinding solver enhanced with learnable communication module improves coordination and performance while maintaining scalability.","ai_keywords":["multi-agent pathfinding","Dec-POMDP","reinforcement learning","imitation learning","multi-round communication","feature sharing","pre-trained model"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6633243f5ddb7702ad3ec216","avatarUrl":"/avatars/0abaf60f1a52e148c3f6fce57eb31eb4.svg","isPro":false,"fullname":"Aleksandr Panov","user":"grafft","type":"user"},{"_id":"652ced57756a15d750266362","avatarUrl":"/avatars/5ef33e4935ccc6e930ceb2475c270bb1.svg","isPro":false,"fullname":"Alexey Skrynnik","user":"tviskaron","type":"user"},{"_id":"6388ba34ec1f539adc092b56","avatarUrl":"/avatars/1d87a657dd81e6ca025cc020f3205525.svg","isPro":false,"fullname":"Konstantin Sobolev","user":"k-sobolev","type":"user"},{"_id":"65c0db0fbda79a18292dfbb7","avatarUrl":"/avatars/1201b8282664c2d8c18beaba2396c03b.svg","isPro":false,"fullname":"Alsu Sagirova","user":"alsu-sagirova","type":"user"},{"_id":"69bd06ffbf0127945f304564","avatarUrl":"/avatars/908a1c6b0c3b5be569a7359141d95f92.svg","isPro":false,"fullname":"Alexei Ossadtchi","user":"ossadtchi","type":"user"},{"_id":"660ee18e2dcd816ad14b3739","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/660ee18e2dcd816ad14b3739/2pPMurtSOHMA96eVk0k7w.jpeg","isPro":false,"fullname":"Maria Marina","user":"zlatamaria","type":"user"},{"_id":"64c8b321cb2f1bf0e7c0f54b","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64c8b321cb2f1bf0e7c0f54b/JflXxMVnG9I0IB5YNyhXF.jpeg","isPro":false,"fullname":"Aydar Bulatov","user":"booydar","type":"user"},{"_id":"6668687caee0993c95b0eb81","avatarUrl":"/avatars/301fe1f395e0a129b1c9785868fa9858.svg","isPro":false,"fullname":"Egor Cherepanov","user":"avanturist","type":"user"},{"_id":"64aac2ac9a803a657dffe53f","avatarUrl":"/avatars/db9968b4a309d7c480973ae28e4861c7.svg","isPro":false,"fullname":"Nikita","user":"eteron","type":"user"},{"_id":"634c084c4abe84057588fd63","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/634c084c4abe84057588fd63/-3doVHYwCqbkPQtwVX9aM.jpeg","isPro":false,"fullname":"Maxim Kurkin","user":"dondosss","type":"user"},{"_id":"6616356e1788281d8fe2cc0c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6616356e1788281d8fe2cc0c/AK4gZYOMBXBcP4dNjZiib.jpeg","isPro":false,"fullname":"Konstantin Yakovlev","user":"konstantin-yakovlev","type":"user"},{"_id":"6263c886c39850dc093aa710","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6263c886c39850dc093aa710/3Y1wopAJpDLMaVfH0HYJc.jpeg","isPro":false,"fullname":"Alla Chepurova","user":"screemix","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.07637.md"}">
Learning to Communicate Locally for Large-Scale Multi-Agent Pathfinding
Abstract
Multi-agent pathfinding solver enhanced with learnable communication module improves coordination and performance while maintaining scalability.
AI-generated summary
Multi-agent pathfinding (MAPF) is a widely used abstraction for multi-robot trajectory planning problems, where multiple homogeneous agents move simultaneously within a shared environment. Although solving MAPF optimally is NP-hard, scalable and efficient solvers are critical for real-world applications such as logistics and search-and-rescue. To this end, the research community has proposed various decentralized suboptimal MAPF solvers that leverage machine learning. Such methods frame MAPF (from a single agent perspective) as a Dec-POMDP where at each time step an agent has to decide an action based on the local observation and typically solve the problem via reinforcement learning or imitation learning. We follow the same approach but additionally introduce a learnable communication module tailored to enhance cooperation between agents via efficient feature sharing. We present the Local Communication for Multi-agent Pathfinding (LC-MAPF), a generalizable pre-trained model that applies multi-round communication between neighboring agents to exchange information and improve their coordination. Our experiments show that the introduced method outperforms the existing learning-based MAPF solvers, including IL and RL-based approaches, across diverse metrics in a diverse range of (unseen) test scenarios. Remarkably, the introduced communication mechanism does not compromise LC-MAPF's scalability, a common bottleneck for communication-based MAPF solvers.
Community
LC-MAPF provides iterative message exchange among agents to enable
progressive refinement of predicted action distributions over
multiple communication rounds.
Experimental evaluations show that LC-MAPF outperforms SOTA learnable MAPF approaches.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.07637 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.07637 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2605.07637 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.