Egidio D'Amato

Incremental Nonlinear Dynamic Inversion and Control Allocation for a Tilt Rotor UAV

A flight control system for a tilt rotor UAV has been synthesized based on an incremental nonlinear dynamic inversion technique. As known the effectiveness of dynamic inversion controllers depends on an accurate knowledge of the plant dynamics. Tilt rotor aircraft are flexible platforms capable of flying in a wide envelope, ranging from hover to forward flight conditions. Inherent dynamics of the vehicle are highly nonlinear in such wide envelope and thrust vectoring, required to control the aircraft in hover and transition phases, introduces dramatic cross coupling effects. A component based approach has been used to model nonlinear aircraft dynamics in the entire flight envelope and to compute local control derivatives at any flight condition. The resulting model was not affine in the control, and an incremental approach was chosen to solve the problem. Effector redundancy was managed developing a control allocation module for distributing control effort among the available actuators, based on their availability and effectiveness. A daisy chaining approach was chosen to manage redundancies and alleviate saturation problems. Control laws have been split in two loops: an outer loop controlling slower dynamics and outputting virtual controls for an inner loop controlling faster dynamics. Different piloting logics were identified for hover and forward flight conditions, although the largest possible degree of commonality was sought. A command blending strategy was devised to control the aircraft during transition phases. Three control logics have been implemented and tested, mixing first order and second order error dynamics in the inner loop, first order and second order reference models in the outer loop, kinematic relations and dynamic inversion algorithm in the outer loop. Results for all three controllers are presented, showing satisfactory tracking of reference inputs with moderate control effort. Differences between the three control logics are deemed marginal. Robustness and disturbance rejection characteristics have not been assessed in this paper. However those are known problems affecting dynamic inversion. Further research will be carried out in the future to improve robustness and noise sensitivity implementing robust outer loops based on H∞ synthesis and adding neural networks to capture modeling errors.

Fault tolerant low cost IMUS for UAVs

In this paper a Fault Tolerant system for the attitude estimation of an Unmanned Aerial Vehicle (UAV) using low cost magnetometers, accelerometers, and gyroscopes, implemented in an Inertial Measurement Unit (IMU) is proposed. An approach based on the Unscented Kalman Filter (UKF) for Detection, Isolation and Reconfiguration is investigated in the presence of a Hardware Duplex IMU mounted on board for flight control purposes. A state automaton, with comparative logics, detects if anomalies occurr on one of the two IMUs; then based on a comparison between the measured variables and their UKF estimations, the fault is isolated and identified. The proposed approach is an alternative to standard Hardware Triplex IMU architectures based on triple physical redundancy. At the expense of a higher computational cost and a small delay in isolating faults, the UKF based approach allows to limit the number of IMUs to two, or to guarantee a fail safe behavior in the presence of a double fault, even if contemporary, with Triplex IMUs. Experimental results on the implementation of a multiple IMU platform on a quadrotor flight control board are finally presented.

Attitude and velocity high-gain control of a tilt-trirotor UAV

The tilting trirotor presented in this paper is able to perform levelled flight at almost null attitude which is an important requirement for fixed wing tilt rotors in view of the lift control during transition from hovering to forward flight. The nonlinear dynamics, the large effect of wind due to the small scale, and the absence of aerodynamic data at low speed and Reynolds numbers, are crucial factors when designing controllers. A two nested loops, high-gain controller scheme is proposed to control attitude and both longitudinal and vertical speed. Controller design is based on the selection of high-gain PI control actions together with suitable sliding surfaces where the closed loop system state variables are forced to evolve. This approach guarantees robustness and fast tuning. To demonstrate the effectiveness of the proposed controller, also in the presence of actuator dynamics and saturations, numerical simulations have been conducted on a detailed simulator of the tri-rotor.

Bi-level flight path optimization for UAV formations

A two-stage optimization model for flight path planning of cooperative UAVs in formation flight in the presence of polygonal obstacles and no-fly zones is proposed. Adopting a Visibility Graph (VG) approach, the virtual formation leader plans its flight path composed of circular arcs and segments connecting obstacles vertices. Then groups of obstacles, being not permeable by the flight formation without UAVs separation or formation shape deformation, are clustered and the VG is revised with the addition of so called rendez-vous waypoints, forcing the formation to be recomposed at a given location beyond groups of obstacles. Such rendez-vous waypoints are optimized at a higher hierarchical level with respect to the flight path optimization leading to a Stackelberg game. The validity of the proposed approach and optimization model is shown by means of numerical simulations where flight paths are obtained calculating shortest path on the revised Visibility Graph and waypoints positions are optimized via a genetic algorithm. Finally, anti-collision among UAVs is achieved via a simple potential based method.