Formation Control of Quadrotor Vehicles

The use of multiple quadrotors to perform cooperative tasks is an active research ?eld of crucial importance. One particular mission of interest involves a formation of quadrotors performing infrastructure inspection autonomously. This thesis describes a scenario where a Target quadrotor is performing a task in a GPS-denied environment, and two Sensor quadrotors with GPS capabilities estimate the Target’s position combining their positions with bearings-only measurements. The selection of the Sensors’ positions can greatly a?ect the estimation performance, motivating the optimization of several objective functions based on the Fisher Information Matrix (FIM). Two novel estimation algorithms are created and tested for both cases where the Target is stationary and moving. The results show geometric con?gurations that increase the estimation accuracy, illustrating the importance of maximizing the FIM for Target localization using quadrotors. Assuming a double integrator model for the target vehicle, an LQR controller is combined with the nonlinear estimator to achieve tracking of a desired trajectory. The results show the in?uence of the Sensors’ positioning not only on the estimation error, but also on the di?erence between the true and desired position of the Target. Finally, a network localization problem with multiple Targets and two Sensors with full knowledge of their position is considered. The Targets estimate their position using not only bearings measurements coming from one of the Sensors, but also coming from another Target. A construction method of network formations that guarantees the convergence of the Targets’ estimates to their actual positions is described and demonstrated.Double Degree Programme IST/TUDelftFlight Dynamics & Control and AvionicsAerospace Engineerin