Autonomous UAV Swarm Formation Utilizing Gradient-Driven Contour Mapping for Radiation Source Localization

This thesis presents a drone swarm for radiation mapping to aid source localization. The Department of Energy advocates employing UAVs for this task, but existing approaches remain inefficient and impractical in real-world scenarios. Three custom drones are built and flight-tested. A control algorithm to follow a contour, a constant-intensity path, is designed using a gradient fit. By knowing the source’s direction, the drone swarm can fly in the optimal trajectory at every step, leaving nothing to assumption. A program is created that implements formation flight and autonomous navigation. It is tested via a software-in-the-loop simulation utilizing radiation sources and detectors modeled on their physical counterparts. Configurable missions are created to demonstrate contours for varying numbers of sources, relative locations, and strengths. The swarm, flying within a physics engine and using drones running production firmware, achieves contour mapping accuracy, efficiency, and reliability that surpass even purely theoretical results. Its detailed modular design simplifies physical deployment and enables extension to nonradiative signals.