Traffic Signal Optimization: Comparison in various Reinforcement Learning Methods -- code

This folder contains Q-learning, DQN, DDQN, DDDQN and SAC models.

Core definitions and functions for all models could be found in the sub-folder "agent".

• Models
  • QlearningAgent.py
  • DQNAgent.py
  • SACAgent.py

Detailed explanations of these models could be found in the comment within our code files.

Hyperparameters and environment settings for each model are defined in base_settings.ini simulation_settings.ini test_setting.ini training_settings.ini.

Training and Testing

The training and testing code are in "trainer.py" and "tester.py", separately.

To compare the performance difference between models

Run "visualization.py"