Publicación: Lane Detection based on Multiple Frames Information
Fecha
2021-09-19
Autores
Editor/a
Director/a
Tutor/a
Coordinador/a
Prologuista
Revisor/a
Ilustrador/a
Derechos de acceso
info:eu-repo/semantics/openAccess
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Nacional de Educación a Distancia (España). Escuela Técnica Superior de Ingeniería Informática. Departamento de Inteligencia Artificial
Resumen
One of the fundamental challenges in the field of autonomous driving is the ability to detect dynamic objects, such as vehicles or pedestrians, and statics ones, such as lanes, in the surroundings of the vehicle. The accurate perception of the environment is crucial for a safe decision making and motion planning. In recent years, advanced driver-assistance systems (ADAS) are getting more important, incorporating new features that a few years ago were limited to luxury cars or even not technically possible. In particular, one of this features is the lane keeping assist system which keeps the car centered in the lane. This system is not just a relevant part of actual driving support features, but it is a crucial function of future fully autonomous vehicles (AD). A few years ago, many of these systems relied on traditional computer vision algorithms, based on computational expensive manually calibrated methods. Their lack of robustness under the long tale of driving scenarios makes them not very suitable for scalability. Moreover the limited computational resources of embedded system puts additional requirements on the design of real time capable algorithms. Nowadays, the state-of-the-art object and structure detectors for advanced driverassistance systems are based on machine learning and, in particular, deep learning approaches. However, such approaches still mainly function on single-frame basis and do not exploit the (high) temporal correlation of the signals representing the perceived environment. Single-frame detection networks might work well under circumstances where the lanes are perfectly visible, but show a lack of performance under certain situations, like occlusions, shadows, rain, snow, lane degradation, etc. To address the aforementioned problem, this thesis introduces temporal information for lane binary segmentation, applying convolutional long short-term memory (ConvLSTM) and convolutional neural networks (CNN) to improve substantially the performance of single-frame architecture under challenging and adverse situations.
Descripción
Categorías UNESCO
Palabras clave
lane detection, autonomous driving, convolutional neuronal networks, recurrent neuronal networks
Citación
Centro
Facultades y escuelas::E.T.S. de Ingeniería Informática
Departamento
Inteligencia Artificial