If you’re gearing up to dive into the world of Drone Reinforcement Learning with PEDRA-2.0, you’ve landed at the right spot! This guide will walk you through everything you need to know — from installations to running your first simulation.

What is PEDRA?

PEDRA stands for Programmable Engine for Drone Reinforcement Learning Applications. Think of it as a highly customizable playground where drones learn to navigate autonomously through realistic virtual environments created with the Unreal Engine. PEDRA allows both single and multiple drones to train in complex scenarios, making it a versatile tool for researchers and developers.

Exciting Updates in Version 2.0

• Support for multi-drone environments.
• Outdoor environment compatibility.
• Improved code structure and organization.
• Comprehensive and detailed documentation.

It’s highly advisable to utilize version 2.0 for better stability in your projects. However, if you’re still interested in PEDRA 1.0, you can download it here.

Installation Steps

1. Clone the Repository

PEDRA comes in two distinct versions:

• PEDRA: Supports single drone operations.
• D-PEDRA: Supports distributed operations for multiple drones.

To download either version, run the following commands:

git clone --single-branch --branch PEDRA https://github.com/aqeelanwar/PEDRA.git
git clone --single-branch --branch D-PEDRA https://github.com/aqeelanwar/PEDRA.git

2. Install Required Packages

Ensure you’re using Python 3.6. Next, install the required packages depending on whether you have an NVIDIA GPU or not:

cd PEDRA
pip install -r requirements_gpu.txt # For systems with NVIDIA GPU
pip install -r requirements_cpu.txt  # For systems without NVIDIA GPU

3. Install Epic Unreal Engine

The environments available in PEDRA utilize Unreal Engine, so installing version 4.18.3 is essential. Follow guidelines provided in this instruction link.

Running PEDRA

1. Download Simulated Environments

You can create your own environments by following the [FAQ documentation](unreal_envsreadme.md) or download simulated environments from this link. Available environments include:

• Indoor: Long, Twist, VanLeer, Techno, Pyramid, FrogEyes, GT, Complex, UpDown, Cloud.
• Outdoor: Courtyard, Forest, OldTown.

2. Configure Your Environment

Edit the configuration files located in the configs folder to set simulation and algorithm parameters. Key parameters include:

• run_name: Name for your simulation.
• algorithm: The selected algorithm for your task.
• num_agents: Number of drones used.

3. Run the Simulation

After checking that your config files are set up, start the simulation using:

cd PEDRA
python main.py

Your simulation should initialize various components, including the environment and PyGame user interface.

Troubleshooting Common Issues

• If you receive errors regarding missing packages, ensure all dependencies have been correctly installed.
• Check that your configuration files are accurately set and compatible with the algorithms you plan to use.
• Running PEDRA on a non-supported environment can lead to crashes. Stick to the recommended setup.

For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

Viewing Learning Parameters

During your simulation, TensorBoard allows you to visualize the training processes effectively. Make sure you’ve logged the correct data paths to access desired metrics.

Conclusion

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.

Now you’re ready to embark on your journey with PEDRA-2.0. Happy coding and drone flying!