# MAPF-World: Action World Model for Multi-Agent Path Finding

arXiv: https://arxiv.org/abs/2508.12087 · HTML: https://mlsanity.com/paper/2508.12087/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in World models for navigation, one of 11 directions in World models.
- Builds on 23 of its 45 references that are in the map.
- First posted on arXiv 16 Aug 2025.

## Placed in

- World models / [World models for navigation](https://mlsanity.com/wm/world-models-for-navigation/)
- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 23 of 23)

- [On the Opportunities and Risks of Foundation Models](https://mlsanity.com/paper/2108.07258/) (2021)
- [Dream to Control: Learning Behaviors by Latent Imagination](https://mlsanity.com/paper/1912.01603/) (2019)
- [World Models](https://mlsanity.com/paper/1803.10122/) (2018)
- [Mastering Atari with Discrete World Models](https://mlsanity.com/paper/2010.02193/) (2020)
- [Recurrent World Models Facilitate Policy Evolution](https://mlsanity.com/paper/1809.01999/) (2018)
- [A Generalization of Transformer Networks to Graphs](https://mlsanity.com/paper/2012.09699/) (2020)
- [WorldVLA: Towards Autoregressive Action World Model](https://mlsanity.com/paper/2506.21539/) (2025)
- [Multi-Agent Pathfinding: Definitions, Variants, and Benchmarks](https://mlsanity.com/paper/1906.08291/) (2019)
- [Understanding World or Predicting Future? A Comprehensive Survey of World Models](https://mlsanity.com/paper/2411.14499/) (2024)
- [Priority Inheritance with Backtracking for Iterative Multi-agent Path Finding](https://mlsanity.com/paper/1901.11282/) (2019)
- [SCRIMP: Scalable Communication for Reinforcement- and Imitation-Learning-Based Multi-Agent Pathfinding](https://mlsanity.com/paper/2303.00605/) (2023)
- [Multi-Agent Coordination across Diverse Applications: A Survey](https://mlsanity.com/paper/2502.14743/) (2025)
- [Fast-in-Slow: A Dual-System Foundation Model Unifying Fast Manipulation within Slow Reasoning](https://mlsanity.com/paper/2506.01953/) (2025)
- [POGEMA: A Benchmark Platform for Cooperative Multi-Agent Pathfinding](https://mlsanity.com/paper/2407.14931/) (2024)
- [Improving LaCAM for Scalable Eventually Optimal Multi-Agent Pathfinding](https://mlsanity.com/paper/2305.03632/) (2023)
- [Multi-Robot Coordination and Layout Design for Automated Warehousing](https://mlsanity.com/paper/2305.06436/) (2023)
- [MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale](https://mlsanity.com/paper/2409.00134/) (2024)
- [Intractability of Optimal Multi-Robot Path Planning on Planar Graphs](https://mlsanity.com/paper/1504.02072/) (2015)
- [Decentralized Monte Carlo Tree Search for Partially Observable Multi-agent Pathfinding](https://mlsanity.com/paper/2312.15908/) (2023)
- [Arbitrarily Scalable Environment Generators via Neural Cellular Automata](https://mlsanity.com/paper/2310.18622/) (2023)
- [LNS2+RL: Combining Multi-Agent Reinforcement Learning with Large Neighborhood Search in Multi-Agent Path Finding](https://mlsanity.com/paper/2405.17794/) (2024)
- [Deploying Ten Thousand Robots: Scalable Imitation Learning for Lifelong Multi-Agent Path Finding](https://mlsanity.com/paper/2410.21415/) (2024)
- [Advancing Learnable Multi-Agent Pathfinding Solvers with Active Fine-Tuning](https://mlsanity.com/paper/2506.23793/) (2025)

## Abstract (arXiv)

Multi-agent path finding (MAPF) studies the problem of planning conflict-free paths for multiple agents from given start locations to designated goals, with applications in robot-assisted logistics and social navigation. Recent decentralized learned solvers have shown promise for large-scale MAPF, particularly when leveraging foundation models and large datasets. However, most existing methods rely on reactive policies, often resulting in congestion, deadlocks, and degraded generalization in high agent-density environments. To address these limitations, we propose MAPF-World, an autoregressive action world model for MAPF that unifies short-horizon local future prediction and action generation, enabling decision-making beyond immediate local observations. MAPF-World models short-horizon local dynamics by predicting the next local observation and neighboring agents' action intentions, capturing both spatial structures and temporal interaction patterns. We further introduce a spatio-agent positional encoding that integrates spatial awareness with agent-level semantics in Transformer-based architectures, facilitating more coordinated multi-agent behaviors. In addition, we augment existing MAPF benchmarks by introducing an automated map generator grounded in real-world urban layouts, aiming to narrow the gap between synthetic simulation and practical deployment scenarios. Extensive experiments across diverse map types and interaction settings demonstrate that MAPF-World achieves strong performance compared with existing learned solvers. Notably, it exhibits robust zero-shot generalization and maintains a high success rate even as agent density increases.