Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Lifelong Multi-Agent Path Finding in Large-Scale Warehouses
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 |
|---|---|---|---|
| Use RHCR instead of holding endpoints | Lifelong MAPF with PBS · Fulfillment warehouse, 60–140 agents | throughput: 2.17 (-6.80%) -> 2.33; 3.33 (-6.33%) -> 3.56; 4.35 (-4.25%) -> 4.55 | Helped |
| Use RHCR instead of reserving dummy paths | Lifelong MAPF with PBS · Fulfillment warehouse, 60–140 agents | throughput: 2.19 (-6.00%) -> 2.33; 3.41 (-4.16%) -> 3.56; 4.50 (-1.06%) -> 4.55 | Helped |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with PBS · Sorting center, 400 agents | runtime: $1.83\pm0.01$ -> $0.61\pm0.00$ | Helped |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with PBS · Sorting center, 400 agents | throughput: 12.46 -> 12.27 (-1.56%) | No clear effect |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with PBS · Sorting center, 1,000 agents | throughput | No clear effect |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with ECBS · Sorting center, 100 agents | runtime: $0.38\pm0.00$ -> $0.07\pm0.00$ | Helped |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with ECBS · Sorting center, 100 agents | throughput: 3.16 -> 3.19 (+1.02%) | No clear effect |
| Use bounded horizon w=5 instead of w=∞ (ablation) | RHCR with CA* · Sorting center, 100 agents | throughput: 3.17 -> 3.19 (+0.53%) | No clear effect |
| Use dynamic horizon adjustment with p=60 rather than a fixed horizon (ablation) | RHCR with ECBS, 60 agents · Fulfillment warehouse | throughput: 1.72 -> 2.10; 2.02 -> 2.10 | Helped |
| Use dynamic horizon adjustment with p=60 rather than a fixed horizon (ablation) | RHCR with ECBS, 60 agents · Fulfillment warehouse | runtime: 0.07s -> 0.35s; 0.17s -> 0.35s | Hurt |
| Increase the horizon when progress is insufficient, using p=1 (ablation) | RHCR · Deadlock example | deadlock avoidance | Helped |
| Use a very small horizon that does not ensure agent progress (ablation) | RHCR · Deadlock example | goal completion | Hurt |
Closest papers by meaning
- 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 0.90)
- Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · similarity 0.90)
- Concurrent Planning and Execution in Lifelong Multi-Agent Path Finding with Delay Probabilities (AAAI 2025 · similarity 0.90)
- Loosely Synchronized Rule-Based Planning for Multi-Agent Path Finding with Asynchronous Actions (AAAI 2025 · similarity 0.89)
- Online Guidance Graph Optimization for Lifelong Multi-Agent Path Finding (AAAI 2025 · similarity 0.89)
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.