Scalable Long-Horizon Planning with Staggered Updates for Lifelong MAPF

2026-08-06

AI Summary

This work addresses the challenge of large-scale multi-agent path planning, where existing approaches struggle to simultaneously achieve real-time performance, scalability, and long-horizon coordination—particularly on general maps. The paper introduces PUSH, a novel framework that uniquely integrates subset-based planning, staggered time windows, and priority inheritance. By updating multi-step paths for only a subset of agents at each step, PUSH enables efficient long-horizon reasoning while maintaining real-time responsiveness. The method synthesizes core ideas from PIBT, EPIBT, and TP, augmented with backtracking and anytime optimization strategies. Experimental results demonstrate that PUSH can plan for scenarios involving tens of thousands of agents within one second, significantly outperforming current baselines in throughput and proving suitable for real-world applications requiring sustained coordination.

Abstract

Lifelong Multi-Agent Path Finding (LMAPF) requires generating collision-free paths for large agent fleets under strict real-time constraints. Reactive frameworks such as PIBT and Enhanced PIBT (EPIBT) scale effortlessly to thousands of agents through rule-based, step-by-step coordination but suffer from severe temporal myopia, making them ineffective in scenarios where long-horizon reasoning is essential. RHCR plans windowed paths over multi-step horizons but incurs substantial planning overheads that hinder scalability. TP tackles both challenges by planning only subsets of agents at each timestep, yet its applicability is restricted to highly structured maps. To achieve long-horizon planning at scale across general maps, we propose Path Updates over Staggered Horizons (PUSH), a LMAPF planner capable of coordinating thousands of agents in under a second while planning over multi-step horizons. PUSH combines the key advantages of PIBT, RHCR, and TP. Like TP, PUSH reduces computational complexity by planning only a subset of agents at each timestep using staggered planning windows. Unlike TP, however, PUSH plans RHCR-style windowed paths in general maps without relying on restrictive map assumptions. To maintain high throughput in congested environments, PUSH further integrates EPIBT-inspired priority inheritance, backtracking, and anytime improvements into its windowed planning. Empirical evaluations across two realistic MAPF scenarios requiring long-horizon reasoning show that PUSH scales to the same massive agent loads as EPIBT (e.g., 10k agents) while achieving significantly higher system throughput than all baselines.

Problem

• Lifelong Multi-Agent Path Finding
• long-horizon planning
• scalability
• real-time constraints
• collision-free paths

Innovation

• Lifelong MAPF
• Staggered Planning
• Long-Horizon Reasoning
• Scalable Multi-Agent Planning
• Windowed Path Planning