PhD title:

Co-advisor: Isabelle FANTONI
Grant from Mexicain Government
Location: Heudiasyc
Date PhD finished: November 26th, 2012

Current position: IR at Tell-Environment

This thesis is focused in the design of original and robust control strategies to stabilize an Unmanned Aerial Vehicle (UAV) in presence of wind disturbances. The proposed control strategies have been tested in simulations and in real-time experiments in two different platforms. It introduces the mathematical model of a UAV in presence of wind. We obtained the dynamical model which takes into account the complementary forces induced by the wind for a Planar Vertical Take-Off and Landing (PVTOL) aircraft and for the quadrotor rotorcraft.

On the other hand, three different nonlinear control laws based on the Lyapunov analysis have been developed to stabilize the UAV in presence of wind. The first approach uses the Robust Control Lyapunov Functions (RCLFs). Given the complexity of the problem, we begun with a mini car which moves on its longitudinal axis. This result has been extended to the case of the PVTOL aircraft and to the quadrotor rotorcraft. Several simulations have been carried out to validate the proposed algorithms. To test its viability in a real application, we have realized experiments using a PVTOL prototype. The simulations and experimental results in real time showed the good performance of the control law in closed loop.

The second approach is based on the saturation functions. We have proposed a robust analysis with respect to unknown external disturbances and nonlinear uncertainties in the model. The proof takes the hypothesis that the wind is bounded. The algorithms have been tested in a quadrotor prototype and the results showed a good performance even in presence of wind disturbances.

The last approach considers the intrinsic properties of the quadrotor flying vehicle, specially the passivity. Thus, a sub-optimal control law has been developed. The analysis is based on the full energy of the system, the passivity, the Lyapunov theory and the use of dynamic programming. The simulation results have showed that this control law can be useful when the flying vehicle has to do more complex maneuvers than hover.

Finally, a control scheme using a state observer has been developed. This scheme uses the Extended Kalman Filter (EKF) to estimate the position in the (x,y) plain and the vertical velocity ˙z of a quadrotor rotorcraft. Using the measurements of an inertial measurement unit, an altitude sensor, a vision system and the control inputs the system state is estimated. The vision system is used to compute the translational velocities of the vehicle and it is composed by a camera and an optical flow algorithm. The estimator has been validated by experiments in real time and the results have been very conclusive.