Introduction

Electric UAVs, especially VTOLs, will play a major role in the future of aviation. With improved battery technology, they will play an integral in both military and civilian applications. The flight dynamics of such aircraft are very dynamic and complex, making it hard to design a control system and a perfect fit for machine learning.

Project Asiris aims to create a unified framework for the development of such systems. It will provide a digital twin of the aircraft, on which the reinforcement learning algorithm can be trained.

The goals are:

  • to create a digital twin that is indistinguishable for the flight system, so the same software can be used for both the simulation and the real aircraft.
  • in this process, create a realistic, real-time simulation of the aircraft, that can be used for training and testing of the control system.
  • eventually create such a UAV, that can be used for evaluation the accuracy of the simulation and the control system.
  • hopefully learn a lot

Disclaimer

This is a one-student project. It will probably contain errors and bad design choices. I am by no means an expert in any of the fields, i am just an 18 year old student trying to learn something new. This should not be seen as a reference for anything.