Colloquium: Yamac Birol (C&O)

The increasing use of drones for both civilian and military applications has raised substantial safety and security concerns, including risks of misuse in restricted airspaces and on the battlefield. Loitering kamikaze drones, in particular, represent a significant threat, necessitating effective countermeasures. This paper proposes an onboard interception guidance algorithm designed to detect, track, and intercept such drones, relying on data from a camera and simulated radar input. The system is initiated using the target's initial location provided by ground-based sensors, while the onboard system processes camera data and simulated radar-derived distance values. The key challenge is ensuring successful interception within the limited operational range of these sensors. Several detection algorithms are implemented and compared, and the chosen detection method is integrated with the OpenCV CSRT tracker. A linear Kalman filter is employed to estimate the target’s position and velocity, effectively handling intermittent missed detections. The proposed solution is evaluated by intercepting a simulated target UAV with a real interceptor drone in TU Delft Cyberzoo, feeding targets position and velocity to the interceptor, alongside validating the detection and tracking algorithms.

Supervisor: Ewoud Jan Jacob Smeur