During recent decades, Unmanned Aerial Vehicles (UAVs) has increased their success in several areas of applications thanks to their versatility. Furthermore, the growing availability of low-cost electronics has pushed their use in civil consumer applications. Attitude control is one of the needed tasks to effectively carry out missions, whose accuracy and adaptability to high payload imbalances or atmospheric disturbances are fundamental requirements. This paper focuses on the design of a flight control scheme based on Model Reference Adaptive Control with Neural Networks. The effectiveness of the proposed controller is assessed through experimental tests carried out on a Crazyflie 2.1 quadrotor.

In the rapidly evolving landscape of unmanned aerial vehicles (UAVs), the optimization of cooperative behaviors holds paramount importance. In the field of cooperative trajectory planning for formations of unmanned aerial vehicles (UAVs), this paper presents a novel approach that separates geometric path planning from collision avoidance. While the geometric path optimization considers a piecewise path made of straight segments, clothoid curves and circular arcs to avoid fixed obstacles, the collision avoidance between UAVs is solved through the definition of a mixed-integer quadratic programming optimal control problem.

This paper is focused on the design of a Model Predictive Control (MPC) algorithm for a micro-satellite with a mass of about 20 kg, equipped with an umbrella-like deployable front structure. This control device allows the vehicle to maneuver and track a prescribed trajectory during the re-entry phase by changing the aerobrake surface. The proposed MPC controller is aimed at minimizing the error between the desired target position at an altitude of about 30 km, after which the satellite follows an uncontrolled ballistic trajectory. A single control move is updated at a sampling rate of 0.1 Hz trying to balance performance with computational burden for a possible real time implementation. To prove that the proposed MPC strategy implies a limited loss of performance, a comparison with MPC controllers optimizing more than one control move has been carried out.