CFLight Project

Overview

This repository contains the implementation of the CFLight project, a framework for traffic light optimization and related experiments. It includes various Python scripts for training, testing, and visualization.

Prerequisites

• Python 3.10
• Required packages listed in requirements.txt

Installation

1. Ensure Python 3.10 is installed on your system.
2. Install the required dependencies by running:

   pip install -r requirements.txt

   Note: This project also supports additional packages like SUMO, which should be installed separately if not included in requirements.txt.

Usage

To run the main CFLight script, execute:

python CFLight.py

File Structure

• CFLight_ablation.py: Ablation study script for CFLight.
• CFLight_causal_vision.py: Causal vision implementation for CFLight.
• CFLight_loss.py: Loss function definitions for CFLight.
• CFLight_q.py: Q-learning or related module.
• CFLight_r.py: Reward function or related module.
• SCM_Experiments.py: Experiments with Structural Causal Models.
• buffer.py: Buffer implementation (likely for reinforcement learning).
• draw.py: Visualization or drawing utilities.
• gan_cf.py: Generative Adversarial Network for counterfactuals.
• gan_cf_loss.py: Loss functions for GAN counterfactuals.
• model.py: Main model definition.
• model_loss.py: Loss functions for the model.
• plt.py: Plotting utilities.
• plt_ab.py: Ablation plotting script.
• readme.md: This file.
• requirements.txt: List of Python dependencies.
• safe_act.py: Safe action selection module.
• safe_loss.py: Safe loss function implementation.

Notes

• Ensure SUMO is properly configured if used in your experiments.
• Refer to individual script comments for detailed usage instructions.