Drone Reinforcement Learning Simulation

Overview

This repository contains the simulation environment and training scripts for a deep reinforcement learning agent to control a quadcopter/drone in Python. Using OpenAI Gym (or Gymnasium) and Stable-Baselines3, you can train policies (e.g., PPO) to navigate the drone from a start point to a target point in 3D space, with options for video recording and custom callbacks.

Table of Contents

• Features
• Prerequisites
• Installation
• Usage
  • Training
  • Recording Video
• Project Structure
• Contributing
• License
• Contact

Features

• Custom Gym environment (DroneGymEnv) with physics-based simulation
• Support for single- and multi-environment vectorized training
• Pre-built PPO training and evaluation scripts
• TensorBoard logging and custom callbacks
• Video recording of rollouts (GIF)

Prerequisites

• Python 3.8+

Installation

1. Clone this repository:

   git clone https://github.com/henryplas/drone_rl.git
   cd drone_rl

2. Create a virtual environment and activate it:

   conda env create -f environment.yaml
   conda activate drone-rl

Usage

Training

Run the main training script using PPO:

python train.py 

Recording Video

To record rollouts to an MP4:

python test.py

Project Structure

drone_rl/
├── drone.py            # Custom Gym environment implementation
├── train.py            # Training entry point
├── test.py             # Video recording utility
├── environment.yaml    # Python dependencies
├── tensorboard/        # TensorBoard and model outputs
└── README.md           # This file

Contributing

Contributions are welcome! Please open an issue or submit a pull request:

1. Fork the repo
2. Create a feature branch (git checkout -b feature-name)
3. Commit your changes (git commit -m "Add new feature")
4. Push to the branch (git push origin feature-name)
5. Open a pull request

Methodology

(TODO) Curriculum learning steps: add curriculum added pos-target to obs made the target easier used different shaping at first

Future Work

• Smoothing out flight controls.
• Trajectory planning
• Playing tag via Multi Agent RL

License

This project is licensed under the MIT License. See LICENSE for details.

Contact

For questions or feedback, contact Henry Plaskonos at henry.plaskonos@gmail.com. Happy flying!