Small autonomous unmanned aerial vehicles (UAVs) are increasingly seen as ideal platforms for many civilian applications such as pipeline inspection, border surveillance, traffic monitoring, natural disaster monitoring and vineyard missions. This thesis address two problems that occurs naturally in these applications involving single or multiple UAVs, namely, (1) Multiple depot resource allocation and (2) Infrastructure monitoring.

In multiple depot resource allocation, we primarily address routing problems that are generalizations of the single Travelling Salesman Problem (TSP). In particular, the routing problems we consider have the UAVs start from distinct locations. The feature that differentiates the routing problems involving UAVs from similar problems previously studied in the literature is that there are constraints on the motion of a UAV. This thesis addresses the constraint that captures the inability of a fixed wing UAV to turn at any arbitrary yaw rate. We provide both approximation algorithms and lower bounds with constant bounding factors for three generalizations of the single TSP.

In the infrastructure monitoring application, we provide vision based control algorithms that equips a fixed wing UAV to track curved and irregular structures such as roads, canals, borders, coastlines based on visual feedback. Most of the vision based tracking work in the literature mainly deals with ground vehicles. In this thesis, we present algorithms and experimental results whereby a fixed wing UAV travelling at 20 m/sec can follow locally linear structures with 10 meters of cross track error. Also, results for a fixed wing UAV searching and mapping the coordinates of a 2 mile stretch of a river with a cross track error of around 9 meters are presented.