<p dir="ltr">Road safety analysis typically relies on the correlation between road surface conditions, lane marking status, or lane width and crash data. Traditionally, this data is surveyed in the field after road construction or car accidents, which is labor-intensive, time-consuming, and hazardous. With the development of mobile mapping systems (MMS) in recent years, the ability to collect lane marking retroreflectivity or lane width information has been greatly improved. By utilizing Light Detection and Ranging (LiDAR) point clouds and RGB images captured by MMS, it is possible to establish lane marking inventory that includes the conditions of pavement markers (such as lane marking retroreflectivity and lane width) for road safety analysis.</p><p dir="ltr">This dissertation aims to develop a comprehensive framework to extract lane markings and report their characteristics using MMS datasets for transportation safety. The proposed approaches include geometric/morphological and deep learning-based approaches based on the LiDAR point clouds acquired by MMS. A normalization strategy is developed to ensure consistent intensity values across laser beams/LiDAR units/MMS for the same objects, thereby enhancing the lane marking extraction. In addition, an image-aided LiDAR approach is proposed to improve the extraction process further. Following the extraction, lane marking classification and characterization, including intensity profile generation and lane width estimation, are conducted to establish comprehensive lane marking inventory.</p><p dir="ltr">To evaluate the proposed strategies, lane marking extraction with and without intensity normalization is also conducted. The results show that the proposed intensity normalization significantly improves the performance of lane marking extraction, regardless of the approach or data used. The geometric approach using normalized intensity achieves F1-scores higher than 90%, outperforming the learning-based models. Furthermore, the intensity derived from two different MMS is compared for performance evaluation, and the agreement of normalized intensity values is within a range of 3.1 to 3.8 (the used MMS provide intensity as an integer number within 0 to 255). Through the normalization, a positive linear relationship between LiDAR normalized intensity and retroreflectivity is found, with the strongest relationship providing an R² of 0.72 and a Pearson's correlation coefficient of 0.85. A comparison of the correlation between original/normalized intensity and retroreflectivity revealed a stronger correlation between original intensity and retroreflectivity. For image-aided LiDAR approach, the image information indeed enhanced the LiDAR-based lane marking extraction approach, as evidenced by the highest F1-score (92.5%) of the image-aided LiDAR approach, outperforming the LiDAR-based (90.3%) and image-based (77.8%) ones. Specifically, the recall increases by 4.0% – from 87.6% (LiDAR-based) to 91.6% (image-aided LiDAR) – surpasses the slight improvement in the precision of 0.2% – from 93.2% (LiDAR-based) to 93.4% (image-aided LiDAR).</p><p dir="ltr">Finally, a Potree-based web portal is developed to visualize the results derived through the proposed lane marking extraction/classification/characterization strategies. This portal includes a function that enables the projection between 2D images and 3D point clouds, as well as tools for displaying intensity profiles and lane width estimates. It is capable of rendering a large dataset of {approximately 4.2 billion LiDAR points} in around ten seconds and allows for the visualization of intensity profiles and lane width estimates. Users can select points of interest in an intensity profile/lane width plot. This selection will result in the corresponding point being showcased in the LiDAR data on the web portal. Furthermore, the LiDAR point can be projected onto the corresponding image.</p><p dir="ltr">The above proposed strategies facilitate the investigation of the relationship between LiDAR intensity and mobile retroreflectivity. To ensure quality control, lane markings derived from geometric and learning-based extraction approaches were compared. These strategies were evaluated using two MMS (equipped with multiple imaging and LiDAR sensors), covering extensive road segments exceeding 400 miles. Furthermore, a reporting mechanism based on multi-modal data from various MMS sensors was implemented to visualize the results derived from the proposed strategies and to serve as a quality control tool. Consequently, the proposed strategies are easily adaptable for different MMS or the regular updating of lane marking inventory.</p>

Les applications pour véhicules autonomes et les systèmes d’aide avancée à la conduite (Advanced Driving Assistance Systems - ADAS) mettent en oeuvre des processus permettant à des systèmes haut niveau de réaliser une prise de décision. Pour de tels systèmes, la connaissance du positionnement précis (ou localisation) du véhicule dans son environnement est un pré-requis nécessaire. Cette thèse s’intéresse à la détection de la structure de scène, au processus de localisation ainsi qu’à la modélisation d’erreurs. A partir d’un large spectre fonctionnel de systèmes de vision, de l’accessibilité d’un système de cartographie ouvert (Open Geographical Information Systems - GIS) et de la large diffusion des systèmes de positionnement dans les véhicules (Global Positioning System - GPS), cette thèse étudie la performance et la fiabilité d’une méthode de localisation utilisant ces différentes sources. La détection de marquage sur la route réalisée par caméra monoculaire est le point de départ permettant de connaître la structure de la scène. En utilisant, une détection multi-noyau avec pondération hiérarchique, la méthode paramétrique proposée effectue la détection et le suivi des marquages sur la voie du véhicule en temps réel. La confiance en cette source d’information a été quantifiée par un indicateur de vraisemblance. Nous proposons ensuite un système de localisation qui fusionne des informations de positionnement (GPS), la carte (GIS) et les marquages détectés précédemment dans un cadre probabiliste basé sur un filtre particulaire. Pour ce faire, nous proposons d’utiliser les marquages détectés non seulement dans l’étape de mise en correspondance des cartes mais aussi dans la modélisation de la trajectoire attendue du véhicule. La fiabilité du système de localisation, en présence d’erreurs inhabituelles dans les différentes sources d’information, est améliorée par la prise en compte de différents indicateurs de confiance. Ce mécanisme est par la suite utilisé pour identifier les sources d’erreur. Cette thèse se conclut par une validation expérimentale des méthodes proposées dans des situations réelles de conduite. Leurs performances ont été quantifiées en utilisant un véhicule expérimental et des données en libre accès sur internet. / Autonomous Vehicles (AV) applications and Advanced Driving Assistance Systems (ADAS) relay in scene understanding processes allowing high level systems to carry out decision marking. For such systems, the localization of a vehicle evolving in a structured dynamic environment constitutes a complex problem of crucial importance. Our research addresses scene structure detection, localization and error modeling. Taking into account the large functional spectrum of vision systems, the accessibility of Open Geographical Information Systems (GIS) and the widely presence of Global Positioning Systems (GPS) onboard vehicles, we study the performance and the reliability of a vehicle localization method combining such information sources. Monocular vision–based lane marking detection provides key information about the scene structure. Using an enhanced multi-kernel framework with hierarchical weights, the proposed parametric method performs, in real time, the detection and tracking of the ego-lane marking. A self-assessment indicator quantifies the confidence of this information source. We conduct our investigations in a localization system which tightly couples GPS, GIS and lane makings in the probabilistic framework of Particle Filter (PF). To this end, it is proposed the use of lane markings not only during the map-matching process but also to model the expected ego-vehicle motion. The reliability of the localization system, in presence of unusual errors from the different information sources, is enhanced by taking into account different confidence indicators. Such a mechanism is later employed to identify error sources. This research concludes with an experimental validation in real driving situations of the proposed methods. They were tested and its performance was quantified using an experimental vehicle and publicly available datasets.