La motivation qui a initié le projet de recherche ANR « Démonstrateur Gun Launched Micro Air Vehicle » est le besoin d’avoir un engin portatif qui permettrait d’obtenir rapidement des images d’une zone d’intérêt située à quelques centaines de mètres, avec la possibilité de pouvoir observer l’intérieur des bâtiments à travers leurs fenêtres ou en allant les explorer directement. Pour répondre à ce besoin, l’Institut franco-allemand de recherche de St Louis a eu l’idée de lancer un minidrone hélicoptère avec un canon. Le GLMAV, sous la forme d’un projectile, est lancé à partir d‘un tube portable à une distance de 500 m et une altitude de 100 m, où il pourra commencer à transmettre des images de la zone à observer. L’utilisation d’un système hybride projectile/minidrone a deux principaux avantages : cela permet d’augmenter l’autonomie du drone, et les premières images de la zone d’intérêt sont obtenues très rapidement. Au cours de cette thèse, nous nous sommes intéressés à la phase de transition, passer d’un projectile à un mini hélicoptère. Un modèle aérodynamique détaillé du GLMAV a été obtenu sur toute son enveloppe de vol. En prenant en compte les difficultés rencontrées lors de la phase de transition (perturbations des capteurs dues à l’accélération de 2500g au lancement, conditions initiales variables), nous avons développé une stratégie de commande, et une loi de commande en vitesse basée sur la technique du backstepping. Cette stratégie de commande a été validée en simulation. La loi de commande en orientation a été validée sur le prototype du GLMAV. Des travaux sur le flux optique, pour obtenir les vitesses latérales, ont été commencés. / The motivation that initiated the ANR research project "Démonstrateur Gun Launched Micro Air Vehicle" is the need to have a portable system which would permi tto quickly obtain images of an zone of interest placed at some hundred of meters, with the possibility to observe inside buildings either by their windows or by going inside them.To answer this need, the French-German Research Institute of St Louis got the idea o fusing a gun launched rotorcraft-MAV. The GLMAV, in its projectile form, is launched from a portable launching tube to a distance of 500m and a height of 100m, where it will collect and transmit visual information from the scene. The use of a projectile/rotorcraft-MAV hybrid system has two main advantages : it allows extending the MAV range,and the first images of the interest zone are obtained very quickly. During this PhD, we studied the transition phase, the passage from a projectile to a rotorcraft-MAV. A detailed aerodynamic model of the GLMAV has been obtained over his whole flight envelope. Taking into account the difficulties encountered during the transition phase (perturbation of the sensors caused by the 2500g acceleration at the launch, varying initial conditions),we developed a control strategy, and a velocity control law based on the backstepping methodology. This control strategy has been validated in simulation. The attitude control law has been validated on the GLMAV prototype. Studies on optical flow, to obtain the lateral velocities of the GLMAV, have been started.

The use of unmanned aerial vehicles (UAV), so-called drones, has been growingrapidly in the last decade. Today, they are used for, among other things, monitoring missions and inspections of places that are difficult for people to access. Toefficiently and robustly execute these types of missions, a swarm of drones maybe used, i.e., a collection of drones that coordinate together. However, this introduces new requirements on what solutions are used for control and navigation. Two important aspects of autonomous navigation of drone swarms are formationcontrol and collision avoidance. To manage these problems, we propose four different solution algorithms. Two of them use leader-follower control to keep formation, Artificial PotentialField (APF) for path planning and Control Barrier Function (CBF)/ExponentialControl Barrier Function (ECBF) to guarantee that the control signal is safe i.e.the drones keep the desired safety distance. The other two solutions use an optimal control problem formulation of a motion planning problem to either generate open-loop or closed-loop trajectories with a linear quadratic regulator (LQR)controller for trajectory following. The trajectories are optimized in terms of timeand formation keeping. Two different controllers are used in the solutions. Oneof which uses cascade PID control, and the other uses a combination of cascadePID control and LQR control. As a way to test our solutions, a scenario is created that can show the utilityof the presented algorithms. The scenario consists of two drone swarms that willtake on different missions executed in the same environment, where the droneswarms will be on a direct collision course with each other. The implementedsolutions should keep the desired formation while smoothly avoiding collisionsand deadlocks. The tests are conducted on real UAVs, using the open sourceflying development platform Crazyflie 2.1 from Bitcraze AB. The resulting trajectories are evaluated in terms of time, path length, formation error, smoothnessand safety.  The obtained results show that generating trajectories from an optimal control problem is superior compared to using APF+leader-follower+CBF/ECBF. However, one major advantage of the last-mentioned algorithms is that decision making is done at every time step making these solutions more robust to disturbancesand changes in the environment.

Research in Unmanned Air Vehicles (UAV's) continues to push the limitations of size and weight. As technical advances have made UAV's smaller and less expensive, they have become more flexible and extensive in their roles. To continue using smaller and less expensive components while retaining and even enhancing performance requires more sophisticated processing of sensor data in order for the UAV to accurately determine its state and thereby allow the use of feedback in controlling the aircraft automatically. This work presents a three-stage state-estimation scheme for the class of UAV's know as Miniature Air Vehicles (MAV's). The first stage estimates pitch and roll, the second stage estimates heading, and the third stage produces a position estimate and an estimate of wind speed and direction. All three stages make use of the extended Kalman filter, a framework for using a system dynamic model to predict future states and to update the predictions using weighted sensor measurements as they become available, where the weighting is based on the relative uncertainty of the dynamic model and the sensors. Using the three-stage state esti-mation scheme, significant improvements in the estimation of pitch, roll and heading have been achieved in simulation and flight testing. Performance of the navigation (position and wind) stage is comparable to an existing baseline algorithms for position and wind, and shows additional promise for use in dead reckoning when GPS updates become unavailable.

Using autonomous miniature air vehicles (MAVs) is a cost-effective, simple method for collecting data about the size, shape, and location characteristics of a forest fire. However, noise in measurements used to compute pose (location and attitude) of the on-board camera leads to significant errors in the processing of collected video data. Typical methods using MAVs to track fires attempt to find single geolocation estimates and filter that estimate with subsequent observations. While this is an effective method of resolving the noise to achieve a better geolocation estimate, it reduces a fire to a single point or small set of points. A georeferenced mosaic is a more effective method for presenting information about a fire to fire fighters. It provides a means of presenting size, shape, and geolocation information simultaneously. We describe a novel technique to account for uncertainty in pose estimation of the camera by converting it to the image domain. We also introduce a new concept, a Georeferenced Uncertainty Mosaic (GUM), in which we utilize a Sequential Monte Carlo method (a particle filter) to resolve that uncertainty and construct a georeferenced mosaic that simultaneously shows size, shape, geolocation, and uncertainty information about the fire.

Flapping flight shows promise for micro air vehicle design because flapping wings provide superior aerodynamic performance than that of fixed wings and rotors at low Reynolds numbers. In these flight regimes, unsteady effects become increasingly important. This thesis explores some of the unsteady effects that provide additional lift to flapping wings through an experiment-based optimization of the kinematics of a flapping wing mechanism in a water tunnel. The mechanism wings and flow environment were scaled to simulate the flight of the hawkmoth (Manduca sexta) at hovering or near-hovering speeds. The optimization was repeated using rigid and flexible wings to evaluate the impact that wing flexibility has on aerodynamic performance of flapping wings. The trajectories that produced the highest lift were compared using particle image velocimetry to characterize the flow features produced during the periods of peak lift. A leading edge vortex was observed with all of the flapping trajectories and both wing types, the strength of which corresponded to the measured amount of lift of the wing. This research furthers our understanding of the lift-generating mechanisms used in nature and can be applied to improve the design of micro air vehicles.

Optimal control for multicopters is difficult in part due to the low processing power available, and the instability inherent to multicopters. Deep imitation learning is a method for approximating an expert control policy with a neural network, and has the potential of improving control for multicopters. We investigate the performance and reliability of deep imitation learning with trajectory optimization as the expert policy by first defining a dynamics model for multicopters and applying a trajectory optimization algorithm to it. Our investigation shows that network architecture plays an important role in the characteristics of both the learning process and the resulting control policy, and that in particular trajectory optimization can be leveraged to improve convergence times for imitation learning. Finally, we identify some limitations and future areas of study and development for the technology. / Optimal kontroll för multikoptrar är ett svårt problem delvis på grund av den vanligtvis låga processorkraft som styrdatorn har, samt att multikoptrar är synnerligen instabila system. Djup imitationsinlärning är en metod där en beräkningstung expert approximeras med ett neuralt nätverk, och gör det därigenom möjligt att köra dessa tunga experter som realtidskontroll för multikoptrar. I detta arbete undersöks prestandan och pålitligheten hos djup imitationsinlärning med banoptimering som expert genom att först definiera en dynamisk modell för multikoptrar, sedan applicera en välkänd banoptimeringsmetod på denna modell, och till sist approximera denna expert med imitationsinlärning. Vår undersökning visar att nätverksarkitekturen spelar en avgörande roll för karakteristiken hos både inlärningsprocessens konvergenstid, såväl som den resulterande kontrollpolicyn, och att särskilt banoptimering kan nyttjas för att förbättra konvergenstiden hos imitationsinlärningen. Till sist påpekar vi några begränsningar hos metoden och identifierar särskilt intressanta områden för framtida studier.

No description available.

Ducted fan vehicles possess a superior ability to maximize payload capacity while minimizing vehicle size. Their ability to both hover and fly at high speed is a key advantage for information-gathering missions, particularly when close proximity to a target is essential. However, the ducted fan's aerodynamic characteristics pose difficulties for stable vehicle flight and therefore require complex control algorithms. In particular, they exhibit a large nose-up pitching moment during wind gusts and when transitioning from hover to forward flight. Understanding ducted fan aerodynamic behavior and how it can be altered through flow control techniques are the two prime objectives of this work. This dissertation provides a new paradigm for modeling the ducted fan's nonlinear behavior and new methods for changing the duct aerodynamics using active flow control. Steady and piezoelectric synthetic jet blowing are employed in the flow control concepts and are compared. The new aerodynamic model captures the nonlinear characteristics of the force, moment, and power data for a ducted fan, while representing these terms in a set of simple equations. The model attains excellent agreement with current and legacy experimental data using twelve non-dimensional constants. Synthetic jet actuators (SJA) have potential for use in flow control applications in UAVs with limited size, weight, and power budgets. Piezoelectric SJAs for a ducted fan vehicle were developed through two rounds of experimental designs. The final SJA design attained peak jet velocities in the range of 225 ft/sec (69 m/s) for a 0.03â x 0.80â rectangular slot. To reduce the magnitude of the nose-up pitching moment in cross-winds, two flow control concepts were explored: flow separation control at the duct lip, and flow turning at the duct trailing edge using a CoandÄ surface. Both concepts were experimentally proven to be successful. Synthetic jets and steady jets were capable of modifying the ducted fan flow to reduce pitching moment, but some cases required high values of steady blowing to create significant responses. Triggering leading edge separation on the duct lip was one application where synthetic jets showed comparable performance to steady jets operating at a blowing coefficient an order of magnitude higher. / Ph. D.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Unmanned Aerial Vehicles are often used for reconnaissance, search and rescue, damage assessment, exploration, and other tasks that are dangerous or prohibitively difficult for humans to perform. Often, these tasks include traversing indoor environments where radio links are unreliable, hindering the use of remote pilot links or ground-based control, and effectively eliminating Global Positioning System (GPS) signals as a potential localization method. As a result, any vehicle capable of indoor flight must be able to stabilize itself and perform all guidance, navigation, and control tasks without dependence on a radio link, which may be available only intermittently. Since the availability of GPS signals in unknown environments is not assured, other sensors must be used to provide position information relative to the environment. This research covers a description of different ranging sensors and methods for incorporating them into the overall guidance, navigation, and control system of a flying vehicle. Various sensors are analyzed to determine their performance characteristics and suitability for indoor navigation, including sonar, infrared range sensors, and a scanning laser rangefinder. Each type of range sensor tested has its own unique characteristics and contributes in a slightly different way to effectively eliminate the dependence on GPS. The use of low-cost range sensors on an inexpensive passively stabilized coaxial helicopter for drift-tolerant indoor navigation is demonstrated through simulation and flight test. In addition, a higher fidelity scanning laser rangefinder is simulated with an Inertial Measurement Unit (IMU) onboard a quadrotor helicopter to enable active stabilization and position control. Two different navigation algorithms that utilize a scanning laser and techniques borrowed from Simultaneous Localization and Mapping (SLAM) are evaluated for use with an IMU-stabilized flying vehicle. Simulation and experimental results are presented for each of the navigation systems.