Mining-Gym

Mining-Gym is an open-source, configurable benchmarking environment for optimizing truck dispatch scheduling in open-pit mining using Reinforcement Learning (RL). The paper with all the details can be found here https://doi.org/10.48550/arXiv.2503.19195

Repository Files

• mgym_GymRun.py: Main script for training and playing RL-based scheduling agents.
• mgym_DefSchdRun.py: Script to run classical/rule-based schedulers.
• mgym_GymEnv.py: OpenAI Gym-compatible environment setting that wraps the DES Mining site simulator.
• mgym_DesEnv.py: Script for the DES-based Mining Site Simulator.
• scenario_loader.py: Utility module to load test scenarios.
• scheduler.py: Contains definitions for rule-based scheduling algorithms.
• config_extend_review.txt: Configuration file where you can customize simulation settings.
• T_scene_config.txt: Configuration file that contains fixed scenario parameters.
• environment.yml: Conda environment setup file to ensure reproducibility.

Setup Instructions

1. Create and activate the virtual environment:
   conda env create -f environment.yml

2. To train a new RL policy network, run:
   python mgym_GymRun.py train --num_episodes 10
   --num_episodes: Sets the number of training episodes.
   The RL algorithm is selected internally in the mgym_GymRun.py code.

3. To run a classical scheduler (e.g., random scheduling), use:
   python mgym_DefSchdRun.py --num_episodes 10 --algo_choice 1
   python mgym_DefSchdRun.py --num_episodes 10 --algo_choice 1 --scenario D
   --num_episodes: Number of episodes to simulate.
   --algo_choice: Selects the scheduler algorithm as defined in scheduler.py. --scenario: Selects fixed ( test-time scenario) configuration (A-F) Example: 1 stands for random scheduling.

4. To play using a pretrained model, run:
   python mgym_GymRun.py play --num_episodes 5 --model_path <path_to_saved_model.zip>
   Replace <path_to_saved_model.zip> with the actual path to your saved model.

5. To change configuration data:
   You can modify the simulation settings by editing the config_extend_review.txt file. This allows you to adjust parameters such as environment details, scheduler settings, and other simulation-related options.