# Multi-agent Reinforcement Learning Combat on DroneCombat 5v5

Source: [Multi-agent Reinforcement Learning Combat on DroneCombat 5v5](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-combat-on-dronecombat-5v5>)

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

Data updated: 2026-03-18T08:13:13.660Z

Evaluation of multi-agent reinforcement learning combat performance in the DroneCombat 5v5 environment, measured by win rate and average episode steps.

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.

| Scope | Recorded value |
| --- | --- |
| Task | Multi-agent Reinforcement Learning Combat |
| Dataset (including any stated split) | DroneCombat 5v5 |
| Metric | Win Rate |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Win Rate](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-combat-on-dronecombat-5v5.md?metric=Win%20Rate>): Higher is better; unit: PERCENT.

- [Average Episode Steps](<https://www.sota2.com/research/sota/multi-agent-reinforcement-learning-combat-on-dronecombat-5v5.md?metric=Average%20Episode%20Steps>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Win Rate | Result date | Report paper |
| --- | --- | --- | --- |
| Dual-attention CTDE | 62 | Not recorded | 69ba31ac0636e5bf4afc09cc |
| No-communication ablation | 49 | Not recorded | 69ba31ac0636e5bf4afc09cc |
| Independent agents | 42 | Not recorded | 69ba31ac0636e5bf4afc09cc |

### Dual-attention CTDE

Win Rate: 62. Result ID: 69ba5e9847f34ce924a4223e.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31ac0636e5bf4afc09cc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba31ac0636e5bf4afc09cc.md>)

Recorded evaluation configuration: {"training\_framework":"CTDE","attention\_mechanism":"dual-attention"}. This record does not assert independent reproduction.

### No-communication ablation

Win Rate: 49. Result ID: 69ba5e9847f34ce924a4223f.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31ac0636e5bf4afc09cc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba31ac0636e5bf4afc09cc.md>)

Recorded evaluation configuration: {"communication":"disabled","training\_framework":"CTDE"}. This record does not assert independent reproduction.

### Independent agents

Win Rate: 42. Result ID: 69ba5e9847f34ce924a42240.

[Reporting paper](<https://www.sota2.com/research/paper/69ba31ac0636e5bf4afc09cc.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa5ec41e209131ce9e50bc.md>)

Recorded evaluation configuration: {"training\_framework":"no CTDE"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
