This model serves as the baseline for the Aerial Wildfire Suppression environment, trained and tested on task 4 with difficulty 2 using the Proximal Policy Optimization (PPO) algorithm.

Environment: Aerial Wildfire Suppression
Task: 4
Difficulty: 2
Algorithm: PPO
Episode Length: 3000
Training max_steps: 1800000
Testing max_steps: 180000

Train & Test Scripts
Download the Environment

Downloads last month: -; Downloads are not tracked for this model. How to track

Video Preview

Reinforcement Learning

Collection including hivex-research/hivex-AWS-PPO-baseline-task-4-difficulty-2

Aerial Wildfire Suppression

Collection

90 items • Updated Mar 20, 2025

Paper for hivex-research/hivex-AWS-PPO-baseline-task-4-difficulty-2

HIVEX: A High-Impact Environment Suite for Multi-Agent Research (extended version)

Paper • 2501.04180 • Published Jan 7, 2025

Evaluation results

Crash Count on hivex-aerial-wildfire-suppression
self-reported

0.9954545468091964 +/- 0.020327884646360022
Extinguishing Trees on hivex-aerial-wildfire-suppression
self-reported

0.28887686133384705 +/- 0.6787165224682201
Extinguishing Trees Reward on hivex-aerial-wildfire-suppression
self-reported

1.4443842768669128 +/- 3.393582550370347
Fire too Close to City on hivex-aerial-wildfire-suppression
self-reported

0.023333333805203436 +/- 0.052815469164145035
Preparing Trees on hivex-aerial-wildfire-suppression
self-reported

173.5697063446045 +/- 22.599638913107242
Preparing Trees Reward on hivex-aerial-wildfire-suppression
self-reported

173.5697063446045 +/- 22.599638913107242
Water Drop on hivex-aerial-wildfire-suppression
self-reported

1.2260812371969223 +/- 0.2007275933789739
Water Pickup on hivex-aerial-wildfire-suppression
self-reported

1.2260812371969223 +/- 0.2007275933789739
Cumulative Reward on hivex-aerial-wildfire-suppression
self-reported

65.80658979415894 +/- 25.50451599293347