philippds's picture
Upload README.md
45d80b4 verified
metadata
library_name: hivex
original_train_name: DroneBasedReforestation_difficulty_5_task_3_run_id_2_train
tags:
  - hivex
  - hivex-drone-based-reforestation
  - reinforcement-learning
  - multi-agent-reinforcement-learning
model-index:
  - name: hivex-DBR-PPO-baseline-task-3-difficulty-5
    results:
      - task:
          type: sub-task
          name: drop_seed
          task-id: 3
          difficulty-id: 5
        dataset:
          name: hivex-drone-based-reforestation
          type: hivex-drone-based-reforestation
        metrics:
          - type: cumulative_distance_reward
            value: 1.317925215959549 +/- 0.28260177110908363
            name: Cumulative Distance Reward
            verified: true
          - type: cumulative_distance_until_tree_drop
            value: 48.28620391845703 +/- 7.283860263327832
            name: Cumulative Distance Until Tree Drop
            verified: true
          - type: cumulative_distance_to_existing_trees
            value: 64.57429847717285 +/- 5.444324231140867
            name: Cumulative Distance to Existing Trees
            verified: true
          - type: cumulative_normalized_distance_until_tree_drop
            value: 0.13179252222180365 +/- 0.02826017752675318
            name: Cumulative Normalized Distance Until Tree Drop
            verified: true
          - type: cumulative_tree_drop_reward
            value: 4.009531931877136 +/- 0.661158168654566
            name: Cumulative Tree Drop Reward
            verified: true
          - type: out_of_energy_count
            value: 0.03808173710480332 +/- 0.021781055433560147
            name: Out of Energy Count
            verified: true
          - type: recharge_energy_count
            value: 10.746735401153565 +/- 0.6862137746749559
            name: Recharge Energy Count
            verified: true
          - type: tree_drop_count
            value: 0.9473729825019837 +/- 0.03268839810742225
            name: Tree Drop Count
            verified: true
          - type: cumulative_reward
            value: 101.15483459472657 +/- 3.824644657818079
            name: Cumulative Reward
            verified: true

This model serves as the baseline for the Drone-Based Reforestation environment, trained and tested on task 3 with difficulty 5 using the Proximal Policy Optimization (PPO) algorithm.

Environment: Drone-Based Reforestation
Task: 3
Difficulty: 5
Algorithm: PPO
Episode Length: 2000
Training max_steps: 1200000
Testing max_steps: 300000

Train & Test Scripts
Download the Environment