Final deliverable

This page is a self-contained showcase: report, slides, code, and media files. If something is broken contact me at the email below, but also try opening the individual files directly.

Name: Thodoris Evangelakos - 2020030187

Contact: tevangelakos@tuc.gr

Short description: The aim of this project was to train a UAV on synthetic data to navigate autonomously in a forest environment.
In order to achieve this, a reinforcement learning (SAC) approach was used with a custom environment and reward function.
This project highly benefited from the implementation of a custom fast simulator for rapid training and testing.