Research map / Multi-agent systems and games / Human-AI and goal-conditioned
MAPF-LNS2: Fast Repairing for Multi-Agent Path Finding via Large Neighborhood Search
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| repair colliding paths with large-neighborhood replanning | PP initial plans on random-32-32-20 · random-32-32-20, 300 agents | success rate: 0.00 -> 1.00 | Helped |
| replace repeated random restarts with large-neighborhood repair | PPR on random-32-32-20 · random-32-32-20, 200 agents | runtime (s): 6.69 -> 0.14 | Helped |
| use SIPPS instead of space-time A* (ablation) | MAPF-LNS2 · random-32-32-20, 400 agents | success rate: 0.68 -> 0.88 | Helped |
| use SIPPS instead of space-time A* (ablation) | MAPF-LNS2 · random-32-32-20, 400 agents | runtime (s): >162 -> >78 | Helped |
| use adaptive neighborhood selection combining the three selection methods (ablation) | MAPF-LNS2 · random-32-32-20 | success rate: 0.48 -> 0.88; 0.60 -> 0.88; 0.76 -> 0.88 | Helped |
| use ALNS instead of random neighborhood selection (ablation) | MAPF-LNS2 · random-32-32-20, 400 agents | runtime (s): >192 -> >78 | Helped |
| use ALNS instead of failure-based neighborhood selection (ablation) | MAPF-LNS2 · random-32-32-20, 400 agents | runtime (s): >155 -> >78 | Helped |
| use ALNS instead of collision-based neighborhood selection (ablation) | MAPF-LNS2 · random-32-32-20, 400 agents | runtime (s): >128 -> >78 | Helped |
| use SIPPS instead of space-time A* (ablation) | EECBS · largest-agent random-scenario instances across MAPF benchmark maps | success rate: almost twice | Helped |
| use SIPPS instead of space-time A* (ablation) | EECBS · largest-agent random-scenario instances across MAPF benchmark maps | runtime: 5 times (or more) faster | Helped |
| extend the runtime limit to one hour | MAPF-LNS2 on warehouse-20-40-10-2-2 · warehouse-20-40-10-2-2 | agents solved: 8,000 | Helped |
Closest papers by meaning
- Anytime Multi-Agent Path Finding with an Adaptive Delay-Based Heuristic (AAAI 2025 · similarity 0.94)
- Traffic Flow Optimisation for Lifelong Multi-Agent Path Finding (AAAI 2024 · similarity 0.92)
- LNS2+RL: Combining Multi-agent Reinforcement Learning with Large Neighborhood Search in Multi-agent Path Finding (AAAI 2025 · similarity 0.92)
- Windowed MAPF with Completeness Guarantees (AAAI 2025 · similarity 0.91)
- Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · similarity 0.91)
- Adaptive Anytime Multi-Agent Path Finding Using Bandit-Based Large Neighborhood Search (AAAI 2024 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.