This work investigates the research topics related to visual aerial navigation in loosely structured and cluttered environments. During the inspection of the desired infrastructure the robot is required to fly in an environment which is uncertain and only partially structured because, usually, no reliable layouts and drawings of the surroundings are available. To support these features, advanced cognitive capabilities are required, and in particular the role played by vision is of paramount importance. The use of vision and other onboard sensors such as IMU and GPS play a fundamental to provide high level degree of autonomy to flying vehicles. In detail, the outline of this thesis is organized as follows • Chapter 1 is a general introduction of the aerial robotic field, the quadrotor platform, the use of onboard sensors like cameras and IMU for autonomous navigation. A discussion about camera modeling, current state of art on vision based control, navigation, environment reconstruction and sensor fusion is presented. • Chapter 2 presents vision based control algorithms useful for reactive control like collision avoidance, perching and grasping tasks. Two main contributions are presented based on relative depth map and image based visual servoing respectively. • Chapter 3 discusses the use of vision algorithms for localization and mapping. Compared to the previous chapter, the vision algorithm is more complex involving vehicle’s poses estimation and environment reconstruction. An algorithm based on RGB-D sensors for localization, extendable to localization of multiple vehicles, is presented. Moreover, an environment representation for planning purposes, applied to industrial environments, is introduced. • Chapter 4 introduces the possibility to combine vision measurements and IMU to estimate the motion of the vehicle. A new contribution based on Pareto Optimization, which overcome classical Kalman filtering techniques, is presented. • Chapter 5 contains conclusion, remarks and proposals for possible developments.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    The Role of Vision Algorithms for Micro Aerial Vehicles


    Beteiligte:

    Erscheinungsdatum :

    2014-03-31


    Anmerkungen:

    Loianno, Giuseppe (2014) The Role of Vision Algorithms for Micro Aerial Vehicles. [Tesi di dottorato]


    Medientyp :

    Hochschulschrift


    Format :

    Elektronische Ressource


    Sprache :

    Italienisch , Englisch


    Klassifikation :

    DDC:    629



    Toward autonomous stereo-vision control of micro aerial vehicles

    Rawashdeh, Samir A. / Aladem, Mohamed | IEEE | 2016


    Monocular vision-based obstacle avoidance for Micro Aerial Vehicles

    Karlsson, Samuel | BASE | 2020

    Freier Zugriff

    Evaluation of Machine Vision Algorithms for Autonomous Aerial Refueling for Unmanned Aerial Vehicles

    Mario Fravolini / Giampiero Campa / Marcello Napolitano | AIAA | 2007


    Vision-Aided Autonomous Landing and Ingress of Micro Aerial Vehicles

    Brockers, Roland / Ma, Jeremy C. / Matthies, Larry H. et al. | NTRS | 2012


    Aerial vehicles with machine vision

    SCHWINDT STEFAN ALEXANDER | Europäisches Patentamt | 2022

    Freier Zugriff