Fixed-wing UAV tracking of evasive targets in 3-dimensional space

In this thesis, we explore the development of autonomous tracking and interception strategies for single and multiple fixed-wing Unmanned Aerial Vehicles (UAVs) pursuing single or multiple evasive targets in 3-dimensional (3D) space. We considered a scenario where we intend to protect high-value facilities from adversarial groups employing ground-based vehicles and quadrotor swarms and focused on solving the target tracking problem. Accordingly, we refined a min-max optimal control algorithm for fixed-wing
UAVs tracking ground-based targets, by introducing constraints on bank angles and turn rates to enhance actuator reliability when pursuing agile and evasive targets. An intelligent and persistent evasive control strategy for the target was also devised to ensure robust performance testing and optimisation.

These strategies were extended to 3D space, incorporating three altitude control algorithms to facilitate flexible UAV altitude control, leveraging various parameters such as desired UAV altitude and image size on the tracking camera lens. A novel evasive quadrotor algorithm was introduced, systematically testing UAV tracking efficacy against various evasive scenarios while implementing anti-collision measures to ensure UAV safety and adaptive optimisation improve the achieved performance. Using decentralised control strategies, cooperative tracking by multiple UAVs of single evasive quadrotor-type and dynamic target clusters was developed along with a new altitude control strategy and task assignment logic for efficient target interception. Lastly, a countermeasure strategy for tracking and neutralising non-cooperative adversarial targets within restricted airspace was implemented, using both Nonlinear Model Predictive Control (NMPC) and optimal controllers.

The major contributions of this thesis include optimal control strategies, evasive target control, 3D target tracking, altitude control, cooperative multi-UAV tracking, adaptive optimisation, high-precision projectile algorithms, and countermeasures. We envision practical applications of the findings from this research in surveillance, security, search and rescue, agriculture, environmental monitoring, drone defence, and autonomous delivery systems. Future efforts to extend this research could explore adaptive evasion, enhanced collaborative UAV swarms, machine learning integration, sensor technologies, and real-world testing.