Control arquitecture for the navigation system of robotic airship using incremental controllers

Abstract

This thesis is in the context of autonomous airships. Airships are advantageous platforms for long duration missions because the lifting gas sustain part of the vehicle weight, reducing energy consumption. Also, airships are platforms with low degree of intrusion in the environment, which is an important characteristic for environmental monitoring applications. This thesis is placed in the context of the project DRONI. This project aims to develop an autonomous airship for the enviromental monitoring in the Mamirauá reserve in Amazon rainforest. Airships are known to have a nonlinear dynamics with many uncertain parameters. The identification of such parameters require complex testing, such as wind tunnel tests. This thesis proposes Incremental Controllers (ICs) as a solution for the control loop in order to overcome the modeling lack. ICs reduce the sensitivity to model uncertainties by adding the dependency of derivative measuring, which increases the sensitivity to measurement noise and delay. Furthermore, actuator redundancy and saturation are issues when designing ICs for airships. Therefore, in this thesis, we discuss how to design ICs presenting solutions for the design issues mentioned before. Nevertheless, the control and guidance loops require access to linear and angular velocities. Meanwhile, the guidance must have access to positions, attitude, and incident wind in the airship body. These information can be retrieved by inertial sensors (such as gyroscope, accelerometer and magnetometer), GPS, and barometer. Thus, a comparison between different filtering strategies is presented in order to define which technique provides better feedback for the control and guidance loops. Also, the wind estimation problem is addressed. Traditional model-based estimators are presented, then a data-driven strategy is proposed. The results obtained led us to propose a novel estimation strategy that performs a fusion between both model-based and data-driven approaches resulting in a hybrid version of a wind estimator. Finally, the guidance problem is addressed. A modified version of the so-called Line-of-sight strategy is presented. This strategy receives adaptations for considering the kinematic restrictions of the airship and the varying behavior when transitioning between hovering and cruise flights. As a final result, we obtained a closed loop system addressing Guidance, Control and Estimation problems. As demonstrated in simulation results, this final control architecture is capable of covering all flight phases of a complete mission, namely: cruise, hovering, vertical take-off and landing.