Development of Flight Rules and Distributed Traffic Management for Autonomous UAVs

Abstract

This thesis investigates how to ensure safe, efficient, and fair operations for autonomous unmanned aerial vehicles (UAVs) in the future civilian use case of point-to-point goods delivery. If UAV fleet sizes grow to predicted levels, then centralised traffic control, similar to traditional aviation, would likely be infeasible. As such we first develop a tactical-level conflict management method that uses velocity obstacles and a ‘right hand’ rule to produce accelerations that steer UAVs such that two UAVs can pass each other with some safe separation and ensures scalability. We then show in simulation that, while this method is effective for many initial conditions, when the angle of approach between two UAVs is small, one of the UAVs experiences a much longer delay than the other, despite applying the same manoeuvre. From this result we develop a hybrid avoidance method where the UAV can choose from a set of avoidance behaviours based on the relative position and velocity of its neighbour. We then explore how this tactical conflict management performs in a large-scale setup defined by a set of origins and destinations that form streams of UAV traffic. These streams form crossings, around which the UAVs will need to engage in avoidance manoeuvres and thus incur delay compared to a straight-line path. We use this delay to characterise the performance of these setups for various demand levels and show that, as the demand increases toward some maximum, the delay increases rapidly and non-linearly. From this we also show that, at these high demand levels, when we split a traffic stream into two parallel streams we can decrease the average delay, despite having increased the number of crossings. Finally we show how we can improve these large-scale traffic setups by designing three high-level traffic management methods. These methods are waypoint-defined routes, floor field zones, and platoon formation.

Date of Award	3 Oct 2023
Original language	English
Awarding Institution	University of Bristol
Sponsors	Thales Group UK
Supervisor	Jonathan Lawry (Supervisor) & R E Wilson (Supervisor)