# Group-Aware Coordination Graph for Multi-Agent Reinforcement Learning

Source: [Group-Aware Coordination Graph for Multi-Agent Reinforcement Learning](<https://www.sota2.com/research/paper/68fa73271e209131cea5ab8a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:04.234Z

Authors: Wei Duan, Jie Lu, Junyu Xuan

Published: 2024-04-17T01:17:10.000Z

## Abstract

Cooperative Multi-Agent Reinforcement Learning (MARL) necessitates seamless collaboration among agents, often represented by an underlying relation graph. Existing methods for learning this graph primarily focus on agent-pair relations, neglecting higher-order relationships. While several approaches attempt to extend cooperation modelling to encompass behaviour similarities within groups, they commonly fall short in concurrently learning the latent graph, thereby constraining the information exchange among partially observed agents. To overcome these limitations, we present a novel approach to infer the Group-Aware Coordination Graph (GACG), which is designed to capture both the cooperation between agent pairs based on current observations and group-level dependencies from behaviour patterns observed across trajectories. This graph is further used in graph convolution for information exchange between agents during decision-making. To further ensure behavioural consistency among agents within the same group, we introduce a group distance loss, which promotes group cohesion and encourages specialization between groups. Our evaluations, conducted on StarCraft II micromanagement tasks, demonstrate GACG's superior performance. An ablation study further provides experimental evidence of the effectiveness of each component of our method.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Multi-agent Battle on MAgent Battle n=64 agents, 2M training steps (test)](<https://www.sota2.com/research/sota/multi-agent-battle-on-magent-battle-n-64-agents-2m-training-steps-test.md>): Multi-agent Battle; MAgent Battle n=64 agents, 2M training steps (test); Final Return: 0.46.

- [Multi-Agent Reinforcement Learning on SMAC 8m\_vs\_9m v1 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-8m_vs_9m-v1-test.md>): Multi-Agent Reinforcement Learning; SMAC 8m\_vs\_9m v1 (test); AUC: 32.73.

- [Multi-agent Reinforcement Learning on SMAC 25m v1 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-25m-v1-test.md>): Multi-Agent Reinforcement Learning; SMAC 25m v1 (test); Test Win Rate: 96.

- [Multi-Agent Reinforcement Learning on SMAC protoss\_8v8 v2 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-protoss_8v8-v2-test.md>): Multi-Agent Reinforcement Learning; SMAC protoss\_8v8 v2 (test); Win Rate: 35.8.

- [Multi-agent Battle on MAgent Battle n=36 agents (test)](<https://www.sota2.com/research/sota/multi-agent-battle-on-magent-battle-n-36-agents-test.md>): Multi-agent Battle; MAgent Battle n=36 agents 3M training steps (test); Final Return: 1.63.

- [Multi-agent Reinforcement Learning on SMAC 1c3s5z v1 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-1c3s5z-v1-test.md>): Multi-Agent Reinforcement Learning; SMAC 1c3s5z v1 (test); Test Win Rate: 89.7.

- [Multi-agent Reinforcement Learning on SMAC MMM2 v1 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-mmm2-v1-test.md>): Multi-Agent Reinforcement Learning; SMAC MMM2 v1 (test); Win Rate: 48.8.

- [Multi-Agent Reinforcement Learning on SMAC 3s5z v1 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-3s5z-v1-test.md>): Multi-Agent Reinforcement Learning; SMAC 3s5z v1 (test); Win Rate: 89.7.

- [Multi-agent Battle on MAgent Battle n=100 agents, 2M training steps (test)](<https://www.sota2.com/research/sota/multi-agent-battle-on-magent-battle-n-100-agents-2m-training-steps-test.md>): Multi-agent Battle; MAgent Battle n=100 agents, 2M training steps (test); Final Return: 0.08.

- [Multi-Agent Reinforcement Learning on SMAC terran\_10v10 v2 (test)](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-on-smac-terran_10v10-v2-test.md>): Multi-Agent Reinforcement Learning; SMAC terran\_10v10 v2 (test); AUC: 19.6.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2404.10976>)

[PDF](<https://arxiv.org/pdf/2404.10976.pdf>)
