fszewczyk / rocket-landing-rl

Custom OpenAI Gym for vertical rocket landing and Deep Q-Learning implementation.

Made using Ray Space

Call me Franco :)

I'm a Software Engineer from Poland 🇵🇱 based in The Netherlands 🇳🇱.

😆 Fun projects	🧑‍🔬 Yeah, science!
🔭 Ray Space Software ray tracing environment written in C++ that lets you create images from space.	🧮 Double Pendulum Hamiltonian Neural Network learns the chaotic dynamics of a double pendulum system.
🇵🇱 Husaria++ C++ translated to Polish. Literally.	🚀 Deep Rocket Landing Custom OpenAI gym and a Deep-Q Network implementation for measuring the impact of curriculum in a reinforcement learning scenario.
🎮 Shkyera Engine Simple game engine written in C++ with full Python interoperability.	🧠 Shkyera Grad Autograd engine based no micrograd with the implementation of various optimizers, layers, and loss functions.

And a lot more!

rocket-landing-rl's Issues

Visualization

Tasks:

• Design a matplotlib dashboard
• Pygame integration showing a flight
• Implement different models

Report

Tasks:

• Run multiple tests
• Run statistical tests on results
• Write report

Designing reward

Tasks:

• Time
• Location
• Orienation
• Linear Speed
• Angular Speed

Physics Engine

Tasks:

• Linear Dynamics
• Rotation around arbitrary axis
• Angular Dynamics

Integrating pytorch

Tasks:

• Learn PyTorch
• Design network
• Create test environment
• Test basic learning
• Design learning process
• Validate learning process