This thesis proposes novel vision-based controllers for the guidance of Unmanned Aerial Vehicles (UAVs). It considers scenarios involving both single and multiple vehicles. For the case of a single-vehicle, novel Image-based visual servo control (IBVS) approaches are proposed for both fixed-wing and vertical take-off and landing (VTOL) UAVs operating in urban or congested environments. Navigation tasks in a complex environment with obstacle avoidance capabilities are considered. In particular, the landing of fixed-wing UAVs on an airstrip and the landing of VTOL-UAVs that includes an obstacle avoidance strategy are considered. The originality of the study lies in the direct exploitation of the centroid of the image of the observed pattern together with the optical flow, thereby eliminating the need to estimate the position and the velocity of the UAV. For multiple vehicles, novel bearing formation controllers are designed for formations under both directed and undirected interaction topologies. In order to relax the classical conditions required by bearing rigidity theory and to lift the scale ambiguity caused by bearings, persistence of excitation of the desired bearing reference is explored. The proposed methodology is supported by rigorous mathematical tools (This involves nonlinear dynamical systems and analysis using Lyapunov theory to formally prove the asymptotic (or exponential) stability of the system, guarantee robustness, and finally ensure good performance of the closed-loop system). Further support is provided by real experiments and/or simulation results.
Vision-based control of Unmanned Aerial Vehicles