Vetrella, Amedeo Rodi (2017) Cooperation and Autonomy for UAV Swarms. [Tesi di dottorato]


Download (60MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Lingua: English
Title: Cooperation and Autonomy for UAV Swarms
Vetrella, Amedeo
Date: 10 December 2017
Number of Pages: 177
Institution: Università degli Studi di Napoli Federico II
Department: dep11
Dottorato: phd046
Ciclo di dottorato: 30
Coordinatore del Corso di dottorato:
Accardo, DomenicoUNSPECIFIED
Fasano, GiancarmineUNSPECIFIED
Date: 10 December 2017
Number of Pages: 177
Uncontrolled Keywords: Cooperative Navigation; Unmanned Aerial Vehicle (UAV); Data Fusion; Kalman Filtering
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-IND/05 - Impianti e sistemi aerospaziali
Date Deposited: 06 Jan 2018 14:08
Last Modified: 26 Mar 2019 10:05


In the last few years, the level of autonomy of mini- and micro-Unmanned Aerial Vehicles (UAVs) has increased thanks to the miniaturization of flight control systems and payloads, and the availability of computationally affordable algorithms for autonomous Guidance Navigation and Control (GNC). However, despite the technological evolution, operations conducted by a single micro-UAV still present limits in terms of performance, coverage and reliability. The scope of this thesis is to overcome single-UAV limits by developing new distributed GNC architectures and technologies where the cooperative nature of a UAV formation is exploited to obtain navigation information. Moreover, this thesis aims at increasing UAVs autonomy by developing a take-off and landing technique which permits to complete fully autonomous operations, also taking into account regulations and the required level of safety. Indeed, in addition to the typical performance limitations of micro-UAVs, this thesis takes into account also those applications where a multi-vehicle architecture can improve coverage and reliability, and allow real time data fusion. Furthermore, considering the low cost of micro-UAV systems with consumer grade avionics, having several UAVs can be more cost effective than equipping a single vehicle with high performance equipment. Among several research challenges to be addressed in order to design and operate a distributed system of vehicles working together for real time applications, this thesis focuses on the following topics regarding cooperation and autonomy: Improvement of UAV navigation performance: This research topic aims at improving the navigation performance of an UAV flying cooperatively with one or more UAVs, considering that the only integration of low cost inertial measurement units (IMUs), Global Navigation Satellite Systems (GNSS) and magnetometers allows real time stabilization and flight control but may not be suitable for applications requiring fine sensor pointing. The focus is set on outdoor environments and it is assumed that all vehicles of the formation are flying under nominal Global Positioning System (GPS) coverage, hence, the main navigation improvement is in terms of attitude estimation. In particular, the key concept is to exploit Differential GPS (DGPS) among vehicles and vision-based tracking to build a virtual additional navigation sensor whose information is then integrated within a sensor fusion algorithm based on an Extended Kalman Filter (EKF). Both numerical simulations and flight results show the potential of sub-degree angular accuracy. In particular, proper formation geometries, and even relatively small baselines, allow achieving a heading uncertainty that can approach 0.1°, which represents a very important result taking into account typical performance levels of IMUs onboard small UAVs. UAV navigation in GPS challenging environments: This research topic aims at developing algorithms for improving navigation performance of UAVs flying in GPS-challenging environments (e.g. natural or urban canyons, or mixed outdoor-indoor settings), where GPS measurements can be unavailable and/or unreliable. These algorithms exploit aiding measurements from one or more cooperative UAVs flying under nominal GPS coverage and are based on the concepts of relative sensing and information sharing. The developed sensor fusion architecture is based on a tightly coupled EKF that integrates measurements from onboard inertial sensors and magnetometers, the available GPS pseudoranges, position information from cooperative UAVs, and line-of-sight information derived by visual sensors. In addition, if available, measurements coming from a monocular pose estimation algorithm can be integrated within the developed EKF in order to counteract the position error drift. Results show that aiding measurements from a single cooperative UAV do not allow eliminating position error drift. However, combining this approach with a standalone visual-SLAM, integrating valid pseudoranges in the tightly coupled filtering structure, or exploiting ad hoc commanded motion of the cooperative vehicle under GPS coverage drastically reduces the position error drift keeping meter-level positioning accuracy also in absence of reliable GPS observables. Autonomous take-off and landing: This research activity, conducted during a 6 month Academic Guest period at ETH Zürich, focuses on increasing reliability, versatility and flight time of UAVs, by developing an autonomous take-off and landing technique. Often, the landing phase is the most critical as it involves performing delicate maneuvers; e.g., landing on a station for recharging or on a ground carrier for transportation. These procedures are subject to constraints on time and space and must be robust to changes in environmental conditions. These problems are addressed in this thesis, where a guidance approach, based on the intrinsic Tau guidance theory, is integrated within the end-to-end software developed at ETH Zürich. This method has been validated both in simulations and through real platform experiments by using rotary-wing UAVs to land on static platforms. Results show that this method achieves smooth landings within 10 cm accuracy, with easily adjustable trajectory parameters.


Downloads per month over past year

Actions (login required)

View Item View Item