This model serves as the baseline for the Ocean Plastic Collection environment, trained and tested on task 3 using the Proximal Policy Optimization (PPO) algorithm.
Environment: Ocean Plastic Collection
Task: 3
Algorithm: PPO
Episode Length: 5000
Training max_steps: 3000000
Testing max_steps: 150000
Train & Test Scripts
Download the Environment
Evaluation results
- Cumulative Reward on hivex-ocean-plastic-collectionself-reported-2.0388235569000246 +/- 1.647487721687271
- Global Reward on hivex-ocean-plastic-collectionself-reported0.0 +/- 0.0
- Local Reward on hivex-ocean-plastic-collectionself-reported-2.854285717010498 +/- 1.1865717722656555