Uso de Descriptores basados en Apariencia Global para la Localización y Creación de Mapas
Yerai Berenguer, Luis Payá, Adrián Peidró y Oscar Reinoso
Jornadas Nacionales de Robótica 2018 (Valladolid, 14-15 de junio del 2018. )
Ed. CEA-IFAC ISBN:978-84-09-02877-1
This work presents the most relevant contributions made within the thesis ’Uso de descriptores holísticos para la localización y creación de mapas: una aproximación al Graph-SLAM mediante apariencia visual’, which will be defended on April 27, 2018 at the University Miguel Hernández de Elche. Nowadays, the field of mobile robotics is increasingly present in dierent fields and applications such as autonomous driving. Due to this growing presence, the scientific community tries to develop new algorithms to provide mobile robots with greater
autonomy to make better decisions in their tasks. With this objective, the robot must develop dierent essential skills when it comes to navigating unknown environments. One of these skills is to create a map of the environment from information collected with the sensors equipped on it. Another of these skills is to be able to calculate its location within the previously created map. In the case that both tasks are executed at the same time it is known as SLAM. In the investigations of the thesis dealt with in this work, dierent algorithms for mapping and
localization are presented using information obtained through omnidirectional visual sensors. Also, for the description of the images, they use techniques based on the global appearance of the images. All algorithms are tested with dierent sets of images, both virtually generated images and images captured under real working conditions and taking into account common phenomena such as illumination changes and occlusions.