Mountain Car Agent

This is the project for the COMP4125 Designing Intelligent Agents in 2025.

Main repo of our group

from zjy2414's repo: https://github.com/zjy2414/Mountain-Car-Agent

Structure of the repo

.
├─humanoid_code # Codes for the humanoid problem.
|    ├─ standup_sb3_predict_a2c.py # Code for A2C.
|    ├─ standup_sb3_predict.py # Code for SAC, TD3 and PPO.
|    ├─ standup_sb3_train_a2c.py # Code for A2C.
|    └─ standup_sb3_train.py # Code for SAC, TD3 and PPO.
├─mountain_car_code # Codes for the mountain car problem without using Q-learning.
|    ├─ predict.py # Testing code.
|    └─ train.py # Training code.
├─mountain_car_q_learning # Codes for the mountain car problem using Q-learning.
|    ├─ exp.py # Version before applying the Grid Search method to find the best hyperparameter configurations and after integrating for the Q-learning algorithm on the Mountain Car environment.
|    ├─ grid_search.py # Version after applying the Grid Search method.
|    └─ q_learning.py # Version before integrating.
└─other_unused_code # Codes that are not used eventually.
    └─ humanoid.py

How to use codes?

1. To use the latest version of the Q-learning code, please use the grid_search.py in mountain_car_q_learning. This code is developed containing three evaluation methods, including descriptive statistics, visual analysis and inferential statistics.

python grid_search.py

2. To use the version before applying the Grid Search method of the Q-learning code, please use the exp.py in mountain_car_q_learning.

python exp.py

3. The code is developed in VSCode with the help of a virtual environment of Anaconda3.
4. Other codes can represent the process of this project.