STEREO VISION BASED SLAM IN DYNAMIC OUTDOOR ENVIRONMENTS USING DEEP LEARNING

"This thesis presents a Simultaneous Localization and Mapping (SLAM) system focused on dynamic environments using convolutional neural networks. The proposed system employs a stereo camera as the input of the SLAM for the acquisition of left and right images and depth map. The neural network is...

Descripción completa

Detalles Bibliográficos
Autor: Daniela Esparza
Tipo de recurso: tesis de maestría
Estado:Versión publicada
Fecha de publicación:2019
País:México
Institución:Centro de Investigaciones en Óptica
Repositorio:Repositorio Institucional CIO
Idioma:inglés
OAI Identifier:oai:cio.repositorioinstitucional.mx:1002/1099
Acceso en línea:http://cio.repositorioinstitucional.mx/jspui/handle/1002/1099
Access Level:acceso abierto
Palabra clave:info:eu-repo/classification/Autor/SLAM
info:eu-repo/classification/Autor/Dynamic Environment
info:eu-repo/classification/Autor/Outdoor Environment
info:eu-repo/classification/Autor/Semantic Segmentation
info:eu-repo/classification/Autor/Neural Networks
info:eu-repo/classification/Autor/Stereo Vision
info:eu-repo/classification/cti/7
info:eu-repo/classification/cti/33
info:eu-repo/classification/cti/3304
info:eu-repo/classification/cti/120304
Descripción
Sumario:"This thesis presents a Simultaneous Localization and Mapping (SLAM) system focused on dynamic environments using convolutional neural networks. The proposed system employs a stereo camera as the input of the SLAM for the acquisition of left and right images and depth map. The neural network is used for object detection and segmentation to avoid erroneous maps and wrong system location. The main job of the neural network is to find out objects within the scene, and to use its features for dynamic detection. Moreover, the processing time of the proposed system is fast and can run in real-time being able to run in outdoor and indoor environments."