🚀 immitation-learning - Safe Reinforcement Learning for Driving

🛠 Download Now

---

🌟 Overview

TSAC-DE is a project designed to create a safe reinforcement learning algorithm for autonomous driving. It works within the CARLA simulator, allowing for both experimentation and practical application. This project incorporates:

• A spatio-temporal transformer encoder that can process sequential driving data.
• Deep ensemble critics that provide reliable uncertainty estimates.
• A trust score that helps in making safe decisions.
• Control barrier functions (CBFs) to ensure all safety constraints are met.

---

📁 Project Structure

The repository is organized into the following main folders:

tsac-de/
├─ tsac_de/ # Core package (agents, models, safety, utils)
├─ configs/ # Experiment configs (YAML)
├─ carla/ # CARLA wrappers

Core Package

This folder contains all main functionalities including agents, models, safety mechanisms, and utilities to support the safe reinforcement learning project.

Experiment Configs

This folder holds configuration files in YAML format. These files define different experiment setups, making it easy to adjust parameters for testing.

CARLA Wrappers

This is where you will find wrappers for the CARLA simulator, enabling smooth interaction between our models and the simulation environment.

---

🛠 Getting Started

To start using TSAC-DE, follow these steps:

1. System Requirements

   • Windows, macOS, or Linux.
   • At least 8 GB of RAM.
   • An up-to-date web browser.
   • Optional: A GPU for better performance.

2. Download & Install

   • Visit the Releases Page to download the latest version.
   • After downloading, unzip the file and follow the installation instructions included.

3. Run the Application

   • Locate the unzipped folder on your system.
   • Find the executable file and double-click it to run the application.
   • Follow the on-screen instructions to proceed.

---

📋 Configuration

After installation, you may want to configure your settings for optimal performance. In the configs/ folder, you will find sample YAML configuration files.

Example Config

You can edit these files to change parameters like:

• Learning rate
• Number of training episodes
• Environment settings

Make sure to save your changes before running the application.

---

📊 Running Experiments

TSAC-DE allows you to run various experiments to test the algorithm's performance.

1. Select an Experiment

   • Navigate to the configs/ folder.
   • Choose a configuration file that fits your needs.

2. Start the Experiment

   • Once a configuration is chosen, open a terminal.
   • Run the command that matches your platform.

3. View Results

   • The application will generate logs and visuals based on the output.
   • Analyze the results to gauge the safety and performance of the algorithm.

---

⚙️ Getting Help

If you encounter issues while using TSAC-DE, please consider:

• Checking the Documentation: Each component has a dedicated section in the documentation folder.
• Seeking Community Support: You can reach out to the community members on GitHub for assistance.

---

🔗 Additional Resources

• CARLA Simulator: Learn more about the simulator used in this project.
• GitHub Issues Page: Report bugs or suggest features.

For further updates and details, keep the application updated regularly by checking the Releases Page.

---

📝 License

This project is licensed under the MIT License. You can freely use, modify, and distribute the code as per the license terms found in the repository.