Discrete Diffusion for Complex and Congested Multi-Agent Path Finding with Sparse Social Attention

2026-05-13

AI Summary

This work addresses the challenge of frequent conflicts and difficult repair in multi-agent path finding (MAPF) under complex, crowded environments, often caused by poor-quality initial solutions. The authors propose DiffLNS, a novel framework that integrates discrete denoising diffusion probabilistic models (D3PMs) into the warm-start phase of Large Neighborhood Search (LNS)-based solvers for the first time. Leveraging a sparse social attention mechanism, DiffLNS learns spatiotemporal priors from expert trajectories and directly models in the discrete action space to generate diverse, high-quality initial joint paths. The approach supports multimodal path generation and exhibits strong generalization capabilities, achieving an average success rate of 95.8% across challenging scenarios with up to 312 agents—surpassing the strongest baseline by 9.6 percentage points—and attaining either optimal or tied-for-optimal performance in all 20 test scenarios.

Abstract

Multi-Agent Path Finding (MAPF) is a coordination problem that requires computing globally consistent, collision-free trajectories from individual start positions to assigned goal positions under combinatorial planning complexity. In dense environments, suboptimal initial plans induce compound conflicts that hinder feasible repair. For repair-based solvers like LNS2, initial plan quality critically affects downstream repair, yet this factor remains underexplored. We propose DiffLNS, a hybrid framework that integrates a discrete denoising diffusion probabilistic model (D3PM) with LNS2. The D3PM serves as an initializer with sparse social attention that learns a spatiotemporal prior over coordinated multi-agent action trajectories from expert demonstrations and samples multiple joint plans. Operating directly on the categorical action space, our discrete diffusion preserves the MAPF action structure and samples from a multimodal joint-plan distribution to produce diverse drafts well suited for neighborhood repair. These drafts act as warm starts for downstream repair, which completes unfinished trajectories and resolves remaining conflicts under hard MAPF constraints. Experimental results show that despite being trained only on instances with at most 96 agents, the initializer generalizes to scenarios with up to 312 agents at inference time. Across 20 complex and congested settings, DiffLNS achieves an average success rate of 95.8%, outperforming the strongest tested baseline by 9.6 percentage points and matching or exceeding all baselines in all 20 settings. To the best of our knowledge, this is the first work to leverage discrete diffusion for warm-starting an LNS-based MAPF solver.

Problem

• Multi-Agent Path Finding
• Discrete Diffusion
• Initial Plan Quality
• Collision-Free Trajectories
• Combinatorial Planning Complexity

Innovation

• Discrete Diffusion
• Multi-Agent Path Finding
• Sparse Social Attention
• Warm Start
• LNS2