Conference Program

08:45	Opening Remarks
09:00	Poster Presentation Session MAD-PINN: A Decentralized Physics-Informed Machine Learning Framework for Safe and Optimal Multi-Agent Control Manan Tayal, Aditya Singh, Shishir Kolathaya, Somil Bansal Interpretable Multi-Agent Path Finding via Decision Tree Extraction from Neural Policies Thibault Lahire Acquisition of Cooperative Behavior in Multi-Agent Path Finding Haruto Sugawara, Hiroyuki Toda Reinforcement Learning for Multi-Agent Planning in the League of Robot Runners Alexandra Barancikova, Reuth Mirsky Parallelism Meets Adaptiveness: Scalable Documents Understanding in Multi-Agent LLM Systems Chengxuan Xia, Qianye Wu, Sixuan Tian, Yilun Hao Practical Consideration for Optimal MAPFr Solvers Andy Li, Zhe Chen, Daniel Harabor
09:30	Paper Presentation Session Pairwise is Not Enough: Hypergraph Neural Networks for Multi-Agent Pathfinding Rishabh Jain, Keisuke Okumura, Michael Amir, Pietro Lio, Amanda Prorok Anytime Single-Step MAPF Planning with Anytime PIBT Nayesha Gandotra, Rishi Veerapaneni, Muhammad Suhail Saleem, Daniel Harabor, Jiaoyang Li, Maxim Likhachev Blue-Sky: Multi-Agent Path Finding with Unassigned Agents (MAPFUA) Ariel Felner, Roni Stern
10:00	Tea Break
10:15	Keynote Talk Context-aware planning and allocation: coordinating robots at Amazon scale Alex Barbosa
11:00	Short Break
11:15	Poster Presentation Session On the Fundamental Limitations of Decentralized Learnable Reward Shaping in Cooperative Multi-Agent Reinforcement Learning Aditya Akella Modelling Multi-Agent Pathfinding Problems by Integrating Connectivity and No-Collision Constraints Jiri Svancara, Roman Barták, Ian Miguel, Joan Espasa Multi-Agent Path Finding for Large Agents Is Intractable Artem Agafonov, Konstantin Yakovlev Two-Oracle Path Planning Consolidated by Heuristic-Rich Information Xi Chen, Yue Zhang, Zhang Yonggang Collaborative Quadruped Transportation in 3D Terrain with Constrained Diffusion Williard Joshua Jose, Li Chen, Hao Zhang Minimizing Makespan with Conflict-Based Search for Optimal Multi-Agent Path Finding Amir Maliah, Dor Atzmon, Ariel Felner
11:35	Paper Presentation Session db-LaCAM: Fast and Scalable Multi-Robot Kinodynamic Motion Planning with Discontinuity-Bounded Search and Lightweight MAPF Akmaral Moldagalieva, Keisuke Okumura, Amanda Prorok, Wolfgang Hönig When Distance Matters: Wasserstein Trust Regions for Multi-Agent Coordination Chirayu Salgarkar, Ali Baheri Flow-Based Guidance Framework with Flex Distribution for w-Optimal Multi-Agent Path Finding Shao-Hung Chan, Thomy Phan, Sven Koenig Extending PiBT to support Heterogeneous Robot Fleets Arjo Chakravarty, Michael X Grey, M. A. Viraj J. Muthugala, Mohan Rajesh Elara
12:15	Lunch
13:00	Poster Session
14:00	Keynote Talk Multi-agent Path Finding: Heuristic Search Meets Machine Learning Konstantin Yakovlev
14:45	Paper Presentation Session MARL-GPT: Foundation Model for Multi-Agent Reinforcement Learning Maria Nesterova, Mikhail Kolosov, Anton Andreychuk, Egor Cherepanov, Oleg Bulichev, Alexey Kovalev, Konstantin Yakovlev, Aleksandr Panov, Alexey Skrynnik VAMP-MR: Vector-Accelerated Motion Planning and Execution for Multi-Robot-Arms Philip Huang, Chenrui Gao, Jiaoyang Li Fault Tolerance in Multi Agent Systems Savitha Suresh, Akshay Narayan
15:15	Tea Break and Poster Session
15:30	Paper Presentation Session Minimizing Makespan with LaCAM* for Optimal Multi-Agent Path Finding Omer Idgar, Dor Atzmon, Ariel Felner Dual-CBS: A Hierarchical Approach via Conflict-Based Search and Sampling for Multi-Agent Motion Planning Joonyeol Sim, Shao-Hung Chan, Sven Koenig CAMAR: Continuous Actions Multi-Agent Routing Artem Pshenitsyn, Aleksandr Panov, Alexey Skrynnik CoRL-MPPI: Enhancing MPPI With Learnable Behaviours For Efficent And Provably-Safe Multi-Robot Collision Avoidance Stepan Dergachev, Artem Pshenitsyn, Aleksandr Panov, Alexey Skrynnik, Konstantin Yakovlev Cube-Based Automated Storage and Retrieval Systems: A Multi-Agent Path Finding Approach Yi Zheng, Yimin Tang, T. K. Satish Kumar, Sven Koenig Optimization of Cable Routing During Construction Orion Sehn, Nathan R. Sturtevant, Brian Gue Learning to Communicate Locally for Large-Scale Multi-Agent Pathfinding Valeriy Vyaltsev, Alsu Sagirova, Anton Andreychuk, Yuri Kuratov, Konstantin Yakovlev, Aleksandr Panov, Alexey Skrynnik
16:40	Short Break
17:05	Paper Presentation Session Collaborative Multi-Robot Non-Prehensile Manipulation via Flow-Matching Co-Generation Yorai Shaoul, Zhe Chen, Mohamed Naveed Gul Mohamed, Federico Pecora, Maxim Likhachev, Jiaoyang Li Balancing Robustness and Efficiency in Multi-Agent Combinatorial Path Finding with Sum of Service Time Yehonatan Kidushim, Meir Kalech, Roni Stern, Dor Atzmon, Avraham Natan
17:15	Panel Discussion Aligning Research Direction Between Academia and Industry Alex Barbosa, Mohan Rajesh Elara, Keisuke Okumura, Konstantin Yakovlev
18:15	Closing Remarks QR code for post-workshop survey will be provided

Invited Speakers

Context-aware planning and allocation: coordinating robots at Amazon scale, by Alex Barbosa

Bio: Alex Barbosa is a Senior Applied Scientist at Amazon Robotics, where he researches algorithms for coordinating large-scale fleets of mobile robots. He earned his PhD from the University of Illinois at Urbana-Champaign, working with Professor Naira Hovakimyan on large-scale optimization under uncertainty, with applications to autonomous operation in crop fields. Prior to his PhD, Alex co-founded a company-ultimately acquired-that developed mobile inspection robots for the nuclear and oil & gas industries. His hands-on expertise is further shaped by extensive participation in robotics competitions such as BattleBots and RoboGames, where he earned multiple awards while having fun along the way.

Abstract: Amazon operates the world's largest fleet of mobile robots, with over a million systems deployed across hundreds of facilities worldwide. Coordinating fleets at this scale introduces challenging—and often unique—problems in path planning and task allocation. This talk explores methods for learning global state representations that inform local control policies and guide search-based algorithms for large-scale coordination. I will also introduce the Block Rearrangement Problem (BRaP) and discuss how Multi-Agent Path Finding (MAPF) provides a promising framework for addressing it.

---

Multi-agent Path Finding: Heuristic Search Meets Machine Learning, by Konstantin Yakovlev

Bio: Konstantin Yakovlev is a PhD in computer science and is currently a principal investigator at the Federal Research Center for Computer Science and Control of Russian Academy of Sciences and holds the same position at AIRI. Meanwhile, he is an associate professor at St.Petersburg University where he also leads the newly established Markov Lab (Andrey Markov is the famous Russian mathematician, known for his seminal contributions to computer science such as Markov chains (which laid the foundations of Markov decision processes - the foundational formalism for modern reinforcement learning). He was a professor at St.Petersburg University from 1880 to 1922. ). He has been contributing to multi-agent pathfinding for almost a decade and is known to the community as a co-author of the several MAPF planners that are tailored to solve MAPF under non-trivial assumptions like any-angle asynchronous moves, agents with volumes etc. Recently he has been involved in exploring how machine learning can be used to create robust and versatile MAPF algorithms that advance state of the art.

Abstract: Coordinating multiple mobile agents, such as warehouse robots or autonomous vehicles, in a shared environment is a core challenge in robotics and artificial intelligence. The Multi-Agent Path Finding (MAPF) formalism provides a fundamental abstraction for this problem, enabling the development of algorithms with strong theoretical guarantees. Heuristic search has been highly successful in this domain, producing optimal or bounded-suboptimal solutions with proven guarantees. Rule-based solvers and optimization techniques (like Large Neighbourhood Search) have also proven to be effective tools to tackle MAPF by providing high-quality solutions within a much lower computational budget. Still most of the modern search-based and rule-based MAPF solvers intrinsically assume that the multi-agent system is controlled in a centralized fashion and it is not trivial to adapt them to real-world constraints like partial observability and limited communication. This is where ML-based approaches may come on stage and overcome the limitations of conventional planning-based methods. This talk explores how heuristic search may be integrated with machine learning (i.e. deep reinforcement learning, imitation learning, transformers, large-scale learning) to create new generation of (decentralized) MAPF solvers. Specifically, I will talk about the methods like Switcher, Follower, MATS-LP, that are centred around the idea of creating a hybrid solver which relies on both search and learning while solving the problem at hand, and more recent approaches like MAPF-GPT (and its derivatives) which are based on imitation learning at scale and (almost) do not rely on search to decide which actions the agents should take at each configuration/time step.

Panel Discussion

The panel discussion, titled "Aligning Research Problems Between Academia and Industry" will feature experts in the field of Multi-Agent Path Finding (MAPF) who will discuss current challenges, future directions, and answer questions from the audience. The panelists include the two invited speakers, Alex Barbosa and Konstantin Yakovlev, along with Mohan Rajesh Elara and Keisuke Okumura.

Mohan Rajesh Elara

Bio: Dr. Mohan is currently an Associate Professor with the Engineering Product Development Pillar at the Singapore University of Technology and Design. He received his Ph.D. and M.Sc degrees from the Nanyang Technological University. His research interests are in robotics with an emphasis on reconfigurable platforms as well as research problems related to embodied-AI and autonomous systems. He has published more than 300 papers in leading journals, books, and conferences. Dr. Mohan is currently serving as an Associate Editor of the IEEE Robotics & Automation Letters and IEEE Nanotechnology Magazine. He is the recipient of the Tatler Asia's Gen.T Leaders of Tomorrow, SG Mark Design, ASEE Best of Design in Engineering Award, Tan Kah Kee Young Inventors' Award and A' Design award. He is the co-founder of Lionsbot, a Singapore-headquartered robotics company which designs and manufactures autonomous cleaning robots for global deployments. Also, he is the co-founder of the Wefaa Robotics, an edutech startup that offers transformative educational robotic products for the global markets. Dr. Mohan has served in various positions of organizing and technical committees of several international competitions and conferences.

Keisuke Okumura

Bio: Keisuke Okumura is a Researcher at the National Institute of Advanced Industrial Science and Technology (AIST), Japan, and a Visiting Scholar at the University of Cambridge, UK. He received his Ph.D. in Computer Science from Tokyo Institute of Technology in 2023. His research focuses on intelligent collective behaviour of swarm agents, particularly the development of multi-agent planning methods tailored to large-scale automation. His honours include the ICAPS 2022 Best Student Paper Award and Tokyo Tech's Ph.D. Dissertation Award (2024).

Advisory Committee