Supervised learning for sensorless wake sensing in formation flight

Today, the use of drones in civil and military applications will only increase; many more drones are therefore expected to be airborne in the next years. One main drawback of existing drones is their operational range. The energy saving benefits of formation flight have been known for some time, but could not yet be applied to civil aircrafts due to obvious safety requirements. Fixed-wing drones are, on the other hand, less safety-sensitive, and could be the first field in which formation flight becomes applied to extend their range. The main problem to positioning an aircraft in a wake is called wake sensing. The goal of this master's thesis is to use artificial intelligence to perform wake sensing without additional sensors. In order to accomplish this, an existing open-source simulator is first modified to add the physics of the wake generated by the leader aircraft. Then, a simplified simulation environment is used to develop a proof of concept that artificial intelligence algorithms have to potential to help sensing the wake of the leader.