Neural Algorithmic Reasoners informed Large Language Model for Multi-Agent Path Finding

2025-08-25

AI Summary

Current large language models (LLMs) underperform on multi-agent pathfinding (MAPF) due to their lack of explicit algorithmic reasoning and inability to model multi-agent coordination. To address this, we propose a novel Neural Algorithmic Reasoner that for the first time integrates graph neural network (GNN)-encoded topological structure and map priors into LLMs via cross-modal cross-attention, guiding stepwise, interpretable planning. This framework achieves organic fusion of LLMs with symbolic algorithmic reasoning. Evaluated in both simulation and real-robot settings, our method significantly outperforms existing LLM-based approaches: path success rate improves by 23.6%, conflict rate decreases by 41.2%, and generalization capability and coordination robustness are markedly enhanced. The approach bridges neural and symbolic paradigms, enabling scalable, explainable, and cooperative multi-agent navigation.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

The development and application of large language models (LLM) have demonstrated that foundational models can be utilized to solve a wide array of tasks. However, their performance in multi-agent path finding (MAPF) tasks has been less than satisfactory, with only a few studies exploring this area. MAPF is a complex problem requiring both planning and multi-agent coordination. To improve the performance of LLM in MAPF tasks, we propose a novel framework, LLM-NAR, which leverages neural algorithmic reasoners (NAR) to inform LLM for MAPF. LLM-NAR consists of three key components: an LLM for MAPF, a pre-trained graph neural network-based NAR, and a cross-attention mechanism. This is the first work to propose using a neural algorithmic reasoner to integrate GNNs with the map information for MAPF, thereby guiding LLM to achieve superior performance. LLM-NAR can be easily adapted to various LLM models. Both simulation and real-world experiments demonstrate that our method significantly outperforms existing LLM-based approaches in solving MAPF problems.

Problem

• Improving LLM performance in multi-agent path finding tasks
• Integrating neural algorithmic reasoners with LLM for MAPF
• Enhancing planning and coordination in complex multi-agent systems

Innovation

• Neural algorithmic reasoners guide large language models
• Graph neural networks integrate map information for pathfinding
• Cross-attention mechanism enables multi-agent coordination performance