Research map / Multi-agent systems and games
Human-AI and goal-conditioned: research map
1,320 papers on Human-AI and goal-conditioned in Multi-agent systems and games, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 4 clusters and 15 approaches. The busiest year so far is 2025.
What it is
Agents that work with people or with each other in shared environments: coordinating with humans, reasoning about others' beliefs (theory of mind), navigating and predicting the movement of other agents, and planning collision-free paths for many robots.
Terms you will see on this page:
- Theory of mind
- modelling what other agents believe and want.
- Multi-agent path finding (MAPF)
- planning paths for many agents without collisions.
- Embodied agent
- an agent acting in a physical or simulated 3D world.
Start reading with the most cited papers here:
- Curiosity-driven Exploration by Self-supervised Prediction (ICML 2017 · 1,773 citations)
- SoPhie: An Attentive GAN for Predicting Paths Compliant to Social and Physical Constraints (CVPR 2019 · 1,115 citations)
Within Multi-agent systems and games, its share shrank from 47.1% in 2023–24 to 20.2% in 2025–26 (296 → 386 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).
Explore Human-AI and goal-conditioned in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
What changed recently
Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):
- Embodied agent learning rising: 8% → 12% of the topic (24 → 47 papers)
- Collaborative multi-agent perception rising: 8% → 12% of the topic (24 → 45 papers)
- World-model-based agents rising: 7% → 10% of the topic (22 → 39 papers)
- Boundedly rational agent modeling falling: 9% → 6% of the topic (28 → 24 papers)
- Interactive multi-agent trajectory prediction falling: 7% → 4% of the topic (20 → 17 papers)
Papers from 2025–2026 that are already cited most:
- End-to-End Autonomous Driving Through V2X Cooperation (AAAI 2025 · 42 citations)
- Navigation World Models (CVPR 2025 · 34 citations)
- MuMA-ToM: Multi-modal Multi-Agent Theory of Mind (AAAI 2025 · 7 citations)
- Orpheus: Engineering Multiagent Systems via Communicating Agents (AAAI 2025 · 7 citations)
Approaches and key papers
Human-AI collaboration · 451 papers
Keywords: communication, ai, collaboration
Approaches in this cluster:
- Human-AI coordination agents (129 papers)
Trains agents that adapt to and collaborate with human partners, including in games like Hanabi. - Collaborative multi-agent perception (109 papers)
Shares features over communication graphs among agents for efficient collaborative perception and ad hoc teamwork. - Boundedly rational agent modeling (87 papers)
Models human behavior with bias, bounded inference, beliefs, and epistemic planning to assist agents. - Theory-of-mind social reasoning (74 papers)
Models mental states and social intelligence in agents for human interaction and grounded language. - Emergent communication in referential games (52 papers)
Studies languages that emerge between agents trained on referential and negotiation games.
Most cited and most cited since 2024:
- Learning Multiagent Communication with Backpropagation (NeurIPS 2016 · 276 citations)
- When2com: Multi-Agent Perception via Communication Graph Grouping (CVPR 2020 · 273 citations)
- ProAgent: Building Proactive Cooperative Agents with Large Language Models (AAAI 2024 · 63 citations)
- Communication-Efficient Collaborative Perception via Information Filling with Codebook (CVPR 2024 · 52 citations)
Exploration and world models · 400 papers
Keywords: environments, learn, exploration
Approaches in this cluster:
- Language-grounded reinforcement agents (97 papers)
Uses language, text games, and LLM guidance to train agents that follow goal specifications. - World-model-based agents (110 papers)
Builds learned or code-based world models for planning and generalist agent behavior. - Imitation and skill learning (105 papers)
Learns long-horizon skills and representations from demonstrations, instructions, and goal-reaching via world models. - Intrinsically motivated exploration (88 papers)
Drives agents with curiosity and information-based intrinsic rewards to learn skills without external reward.
Most cited and most cited since 2024:
- Curiosity-driven Exploration by Self-supervised Prediction (ICML 2017 · 1,773 citations)
- Noisy Networks For Exploration (ICLR 2018 · 571 citations)
- SubT-MRS Dataset: Pushing SLAM Towards All-weather Environments (CVPR 2024 · 60 citations)
- Equity-Transformer: Solving NP-Hard Min-Max Routing Problems as Sequential Generation with Equity Context (AAAI 2024 · 18 citations)
Embodied agents and navigation · 387 papers
Keywords: navigation, prediction, embodied
Approaches in this cluster:
- Embodied agent learning (113 papers)
Builds embodied robotic agents using vision-language distillation, affordance exploration, and generalist policies. - Interactive multi-agent trajectory prediction (118 papers)
Predicts joint future trajectories with relational reasoning, lane awareness, and factorized interaction models. - Vision-language navigation agents (99 papers)
Develops and analyzes navigation agents following instructions with structured decision spaces and robustness. - Data-driven traffic simulation (57 papers)
Simulates realistic multi-agent driving behavior with learned models and large-scale scenario platforms.
Most cited and most cited since 2024:
- SoPhie: An Attentive GAN for Predicting Paths Compliant to Social and Physical Constraints (CVPR 2019 · 1,115 citations)
- DESIRE: Distant Future Prediction in Dynamic Scenes With Interacting Agents (CVPR 2017 · 1,086 citations)
- HPNet: Dynamic Trajectory Forecasting with Historical Prediction Attention (CVPR 2024 · 88 citations)
- Towards Learning a Generalist Model for Embodied Navigation (CVPR 2024 · 63 citations)
Multi-agent path finding · 82 papers
Keywords: mapf, agent path, path finding
Approaches in this cluster:
- Learning-guided MAPF search (74 papers)
Uses large neighborhood search and machine learning to solve lifelong and anytime multi-agent path finding. - Conflict-based search for MAPF (8 papers)
Improves conflict-based search with better heuristics and prioritization for bounded-suboptimal path finding.
Most cited and most cited since 2024:
- Lifelong Multi-Agent Path Finding in Large-Scale Warehouses (AAAI 2021 · 250 citations)
- EECBS: A Bounded-Suboptimal Search for Multi-Agent Path Finding (AAAI 2021 · 212 citations)
- Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning (AAAI 2024 · 36 citations)
- Traffic Flow Optimisation for Lifelong Multi-Agent Path Finding (AAAI 2024 · 24 citations)
Related topics in Multi-agent systems and games
- LLM agents (1,294)
- Cooperative multi-agent RL (635)
- Games and equilibria (537)
