AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Learning-guided MAPF search

Uses large neighborhood search and machine learning to solve lifelong and anytime multi-agent path finding.

73 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Human-AI and goal-conditioned, keyword cluster “mapf · agent path · path finding”

State History Tried Checklist Preprints Who Export

Where it stands

Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.

16 papers in 2023–24 and 28 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 5.4% to 7.3%.

How this is decided

A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.

How it got here

2016: 4162017: 2172018: 4182019: 7192020: 4202021: 4212022: 4222023: 9232024: 7242025: 13252026: 1526

Most cited paper of each year

Earliest papers here

What has been tried

56 papers here report 506 measured changes (a component added, removed or swapped, compared with a baseline): helped 64%, no clear effect 8%, hurt 23%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.

Ablations in recent papers

What was changedMeasured onResultEffectPaper
add the learnable policy for collision avoidancewarehouse · throughputHelpedLearn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · 36 citations)
Add online refinement and Focal Search bound 2 to guided PIBTSortation (600–1000 agents) · throughput6.5±0.6 -> 11.4±0.3; 5.2±0.5 -> 11.5±0.6; 3.5±0.3 -> 10.2±0.7HelpedTraffic Flow Optimisation for Lifelong Multi-Agent Path Finding (AAAI 2024 · 24 citations)
Add MCTS to CostTracerWarehouse-like map · throughputalmost twiceHelpedDecentralized Monte Carlo Tree Search for Partially Observable Multi-Agent Pathfinding (AAAI 2024 · 14 citations)
add goal information for all agentsRandom, Mazes, Warehouse, and Cities-tiles · success rate95.7% -> 97.6%; 71.6% -> 74.6%; 92.8% -> 94.1%; 88.4% -> 82.0%MixedMAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale (AAAI 2025 · 11 citations)
decrease the guidance update interval mwarehouse map, static and dynamic task distributions · throughputHelpedOnline Guidance Graph Optimization for Lifelong Multi-Agent Path Finding (AAAI 2025 · 11 citations)
set neighborhood size to 4 instead of 8Random-small, 60% agent density · success rate59% -> 55%HurtLNS2+RL: Combining Multi-agent Reinforcement Learning with Large Neighborhood Search in Multi-agent Path Finding (AAAI 2025 · 8 citations)
increase neighborhood-size options to E = 3random and city maps; 128-second time budget · sum of delaysHelpedAdaptive Anytime Multi-Agent Path Finding Using Bandit-Based Large Neighborhood Search (AAAI 2024 · 8 citations)
Add the swap operation to LSRPWarehouse · scalability and runtimeHelpedLoosely Synchronized Rule-Based Planning for Multi-Agent Path Finding with Asynchronous Actions (AAAI 2025 · 7 citations)
Add the previous uncommitted paths as guidance for LaCAM* replanningWarehouse, p = 0.01, 600 agents · average solution cost (×1000)4.733±0.376 -> 3.667±0.282HelpedConcurrent Planning and Execution in Lifelong Multi-Agent Path Finding with Delay Probabilities (AAAI 2025 · 3 citations)
Add partial expansion (PE)Single-shot grid maps · success rateHelpedMulti-Agent Motion Planning for Differential Drive Robots Through Stationary State Search (AAAI 2025 · 3 citations)
Increase high-level suboptimality above 1Standard benchmark maps · scalabilityMixedWindowed MAPF with Completeness Guarantees (AAAI 2025 · 3 citations)
Restrict seed selection to the top-K delayed agentsDen520d and City, 700 agents · sum of delaysHelpedAnytime Multi-Agent Path Finding with an Adaptive Delay-Based Heuristic (AAAI 2025 · 2 citations)

What reviewers will expect

Datasets and baselines named in the 14 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.

Baselines

  • LaCAM* · 4 of 14 papers
  • EECBS · 3 of 14 papers
  • RHCR · 3 of 14 papers
  • LNS2 · 2 of 14 papers
  • SCRIMP · 2 of 14 papers

Recent preprints

Loading recent arXiv preprints…

    Who publishes on it

    Institutions on the most papers of this approach since 2023 (of 44 papers; from OpenAlex and the papers' affiliations).

    Use it

    Download BibTeX of 38 key papersOpen the topic in the interactive map

    Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.

    Same keyword cluster

    Elsewhere in the topic

    Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).

    Get the next scime Dispatch and a monthly field update

    Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy