Centre de documentation
scientifique

GNSS/INS Kalman filter integrity monitoring with uncertain time correlated error processes / Omar Garcia Crespillo (2022)  

Public Titre : GNSS/INS Kalman filter integrity monitoring with uncertain time correlated error processes Type de document : Thèse/HDR Auteurs : Omar Garcia Crespillo, Auteur ; Jan Skaloud, Directeur de thèse ; Michael Meurer, Directeur de thèse Editeur : Lausanne : Ecole Polytechnique Fédérale de Lausanne EPFL Année de publication : 2022 Importance : 180 p. Format : 21 x 30 cm Note générale : bibliographie Thèse présentée pour l'obtention du grade de Docteur ès Sciences Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Navigation et positionnement [Termes IGN] corrélation temporelle [Termes IGN] couplage GNSS-INS [Termes IGN] filtre de Kalman [Termes IGN] fréquence multiple [Termes IGN] modèle d'erreur [Termes IGN] modèle de Gauss-Markov [Termes IGN] navigation inertielle [Termes IGN] norme [Termes IGN] positionnement par GNSS [Termes IGN] Receiver Autonomous Integrity Monitoring [Termes IGN] système d'extension Résumé : (auteur) Safety-critical navigation applications require that estimation errors be reliably quantified and bounded. Over the last decade, significant effort has been put to guarantee a bounded position estimation by using Global Navigation Satellite Systems (GNSS) by means of satellite-based or ground-based augmentation systems (SBAS, GBAS) and Advanced Receiver Autonomous Integrity Monitoring (ARAIM) for aviation. This has been achieved by carefully designing models that overbound the different residual error components in range measurements (e.g., satellite clock and orbit, tropospheric and multipath among others). On the other hand, and as part of Aircraft based Augmentation Systems (ABAS), the use of Inertial Reference Systems (IRS) has been traditionally included as additional source of redundant navigation information. More recently, the use of Inertial Navigation Systems (INS) with a wider spectrum of possible inertial sensor qualities in tighter integration with single-frequency GNSS has seen its way in a new Minimum Operational Performance Standard (MOPS). New GNSS/INS systems and standards could still benefit from the methodologies and aspects developed for future dual-frequency/multiconstellation GNSS standards. However, safety-related GNSS systems like ARAIM are snapshot-based, that is, the position estimation is performed independently at every epoch, whereas GNSS/INS systems are typically based on Kalman filtering (KF). Therefore, the existing error overbounding models and methodologies are not enough to produce a robust KF position estimation since the impact of time-correlation in measurements must also be accounted for. Moreover, it has been observed that the time-correlation of different GNSS errors presents also some level of uncertain behavior, which makes very challenging for linear dynamic systems to produce a guaranteed solution. As proposed by GNSS Minimum Operational Performance Standards (MOPS), there are sources of time-correlated errors that can be well modelled using a first order Gauss-Markov process (GMP). Using this GMP parametric model, it is possible to capture the uncertain timecorrelated nature of error processes by allowing the variance and time correlation constant of the GMP model to be in a bounded range. Under this situation, the first part of this thesis studies the propagation of the uncertain models through the Kalman filter estimation and provides new theoretical tools in time and frequency domain to bound the KF error estimation covariance. As a result, tight stationary bounding models on the GMP uncertain processes are derived in both continuous and discrete time domain. This is extended to non-stationary models that provide tighter error bounding during an initial transient phase when measurements are first introduced (which will be relevant in scenarios with changing number of visible satellites). The new models can very easily be used during the KF implementation which might be very attractive by regulators and designers. In the second part of the thesis, the new overbounding GMP models are applied for a dual-frequency GPS-Galileo tightly-coupled GNSS/INS integration. The design of the filter and of error models is performed following compatibility with current aviation standards and ARAIM Working Group C results. The impact of the use of the new models is analysed in terms of conservativeness, integrity and continuity based on realistic operational simulations linked to airport runways. The benefit of an overbounded GNSS/INS solution is also compared with the current baseline ARAIM algorithm solution. This thesis supports the evolution of safe GNSS-based positioning systems from only snapshot based to filtered solutions. Ensuring integrity for Kalman filter in general and for GNSS/INS systems in particular is a game changer to achieve higher performance levels for future dualfrequency multi-constellation aviation services and is of vital importance for new ground and air applications like autonomous vehicles or urban air mobility. Note de contenu : Introduction 1- Preliminaries 2- Bounding Kalman Filter with uncertain error processes 3- Application to GNSS/INS integraty monitoring 4- Closing 5- Appendix Numéro de notice : 28688 Affiliation des auteurs : non IGN Thématique : POSITIONNEMENT Nature : Thèse étrangère Note de thèse : Thèse de Doctorat : Sciences : Lausanne : 2022 DOI : sans En ligne : https://infoscience.epfl.ch/record/292087?ln=fr Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100103

Learning to represent and reconstruct 3D deformable objects / Jan Bednarik (2022)  

Public Titre : Learning to represent and reconstruct 3D deformable objects Type de document : Thèse/HDR Auteurs : Jan Bednarik, Auteur ; Pascal Fua, Directeur de thèse ; M. Salzmann, Directeur de thèse Editeur : Lausanne : Ecole Polytechnique Fédérale de Lausanne EPFL Année de publication : 2022 Importance : 138 p. Format : 21 x 30 cm Note générale : bibliographie Thèse présentée pour l'obtention du grade de Docteur ès Sciences, Ecole Polytechnique Fédérale de Lausanne Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] appariement de formes [Termes IGN] apprentissage profond [Termes IGN] cohérence temporelle [Termes IGN] déformation de surface [Termes IGN] distorsion d'image [Termes IGN] géométrie de Riemann [Termes IGN] image 3D [Termes IGN] reconstruction d'objet [Termes IGN] semis de points [Termes IGN] vision par ordinateur Index. décimale : THESE Thèses et HDR Résumé : (auteur) Representing and reconstructing 3D deformable shapes are two tightly linked problems that have long been studied within the computer vision field. Deformable shapes are truly ubiquitous in the real world, whether be it specific object classes such as humans, garments and animals or more abstract ones such as generic materials deforming under stress caused by an external force. Truly practical computer vision algorithms must be able to understand the shapes of objects in the observed scenes to unlock the wide spectrum of much sought after applications ranging from virtual try-on to automated surgeries. Automatic shape reconstruction, however, is known to be an ill-posed problem, especially in the common scenario of a single image input. Therefore, the modern approaches rely on deep learning paradigm which has proven to be extremely effective even for the severely under-constrained computer vision problems. We, too, exploit the success of data-driven approaches, however, we also show that generic deep learning models can greatly benefit from being combined with explicit knowledge originating in traditional computational geometry. We analyze the use of various 3D shape representations for deformable object reconstruction and we distinctly focus on one of them, the atlas-based representation, which turns out to be especially suitable for modeling deformable shapes and which we further improve and extend to yield higher quality reconstructions. The atlas-based representation models the surfaces as an ensemble of continuous functions and thus allows for arbitrary resolution and analytical surface analysis. We identify major shortcomings of the base formulation, namely the infamous phenomena of patch collapse, patch overlap and arbitrarily strong mapping distortions, and we propose novel regularizers based on analytically computed properties of the reconstructed surfaces. Our approach counteracts the aforementioned drawbacks while yielding higher reconstruction accuracy in terms of surface normals on the tasks of single view-reconstruction, shape completion and point cloud auto-encoding. We dive into the problematics of atlas-based shape representation even deeper and focus on another pressing design flaw, the global inconsistency among the individual mappings. While the inconsistency is not reflected in the traditional reconstruction accuracy quantitative metrics, it is detrimental to the visual quality of the reconstructed surfaces. Specifically, we design loss functions encouraging intercommunication among the individual mappings which pushes the resulting surface towards a C1 smooth function. Our experiments on the tasks of single-view reconstruction and point cloud auto-encoding reveal that our method significantly improves the visual quality when compared to the baselines. Furthermore, we adapt the atlas-based representation and the related training procedure so that it could model a full sequence of a deforming object in a temporally-consistent way. In other words, the goal is to produce such reconstruction where each surface point always represents the same semantic point on the target ground-truth surface. To achieve such behavior, we note that if each surface point deforms close-to-isometrically, its semantic location likely remains unchanged. Practically, we make use of the Riemannian metric which is computed analytically on the surfaces, and force it to remain point-wise constant throughout the sequence. Our experimental results reveal that our method yields state-of-the-art results on the task of unsupervised dense shape correspondence estimation, while also improving the visual reconstruction quality. Finally, we look into a particular problem of monocular texture-less deformable shape reconstruction, an instance of the Shape-from-Shading problem. We propose a multi-task learning approach which takes an RGB image of an unknown object as the input and jointly produces a normal map, a depth map and a mesh corresponding to the observed part of the surface. We show that forcing the model to produce multiple different 3D representations of the same objects results in higher reconstruction quality. To train the network, we acquire a large real-world annotated dataset of texture-less deforming objects and we release it for public use. Finally, we prove through experiments that our approach outperforms a previous optimization based method on the single-view-reconstruction task. Note de contenu : 1- Introduction 2- Related work 3- Atlas-based representation for deformable shape reconstruction 4- Shape reconstruction by learning differentiable surface representations 5- Better patch stitching for parametric surface reconstruction 6- Temporally-consistent surface reconstruction using metrically-consistent atlases 7- Learning to reconstruct texture-less deformable surfaces from a single view 8- Conclusion Numéro de notice : 15761 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse étrangère Note de thèse : Thèse de Doctorat : Sciences : Lausanne, EPFL : 2022 DOI : 10.5075/epfl-thesis-7974 En ligne : https://doi.org/10.5075/epfl-thesis-7974 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100958

Reshaping perception for autonomous driving with semantic keypoints / Lorenzo Bertoni (2022)  

Public Titre : Reshaping perception for autonomous driving with semantic keypoints Type de document : Thèse/HDR Auteurs : Lorenzo Bertoni, Auteur Editeur : Lausanne : Ecole Polytechnique Fédérale de Lausanne EPFL Année de publication : 2022 Importance : 177 p. Format : 21 x 30 cm Note générale : bibliographie Thèse présentée pour l'obtention du grade de Docteur ès Sciences, Ecole Polytechnique Fédérale de Lausanne Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Intelligence artificielle [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] détection automatique [Termes IGN] détection d'objet [Termes IGN] détection de piéton [Termes IGN] estimation de pose [Termes IGN] navigation autonome [Termes IGN] système multi-agents [Termes IGN] vision par ordinateur Résumé : (auteur) The field of artificial intelligence is set to fuel the future of mobility by driving forward the transition from advanced driver-assist systems to fully autonomous vehicles (AV). Yet the current technology, backed by cutting-edge deep learning techniques, still leads to fatal accidents and does not convey trust. Current frameworks for 3D perception tasks, such as 3D object detection, are not adequate as they (i) do not generalize well to new scenarios, (ii) do not take into account measures of confidence in their predictions, and (iii) are not suitable for large-scale deployment as mainly based on costly LiDAR sensors. This doctoral thesis aims to study vision-based deep learning frameworks that can accurately perceive the world in 3D and generalize to new scenarios. We propose to escape the pixel domain using semantic keypoints, a sparse representation for every object in the scene containing meaningful information for 2D and 3D reasoning. The low-dimensionality enables downstream neural networks to focus on essential elements in the scene and improve their generalization capabilities. Furthermore, driven by the limitation of deep learning architectures outputting point estimates, we study how to estimate a confidence interval for each prediction. In particular, we emphasize vulnerable road users, such as pedestrians and cyclists, and explicitly address the long tail of 3D pedestrian detection to contribute to the safety of our roads. We further show the efficacy of our framework on multiple real-world domains by (a) integrating it in an existing AV pipeline, (b) detecting human-robot eye contact in real-world scenarios, and (c) helping verify the compliance of safety measures in the case of the COVID-19 outbreak. Finally, we publicly release the source code of all our projects and develop a unified library to contribute to an open science mission. Note de contenu : 1- Introduction 2- Semantic keypoints detection 3- Monocular 3D pedestrian localization and uncertainty estimation 4- Tackling the long tail of 3D pedestrian localization with stereo cameras 5- Autonomous driving applications of pedestrian 3D detection 6- Detecting pedestrians attention: Human-robot eye contact in the wild 7- Beyond autonomous driving: Social interactions and social distancing 8- Conclusion Numéro de notice : 24077 Affiliation des auteurs : non IGN Thématique : INFORMATIQUE Nature : Thèse étrangère Note de thèse : PhD Thesis : Sciences : EPFL : 2022 DOI : 10.5075/epfl-thesis-10072 En ligne : https://doi.org/10.5075/epfl-thesis-10072 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=102212

Context-aware image super-resolution using deep neural networks / Mohammad Saeed Rad (2021)  

Public Titre : Context-aware image super-resolution using deep neural networks Type de document : Thèse/HDR Auteurs : Mohammad Saeed Rad, Auteur ; Jean-Philippe Thiran, Directeur de thèse Editeur : Lausanne : Ecole Polytechnique Fédérale de Lausanne EPFL Année de publication : 2021 Importance : 148 p. Format : 21 x 30 cm Note générale : bibliographie Thèse présentée pour l'obtention du grade de Docteur ès Sciences Langues : Français (fre) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] image à basse résolution [Termes IGN] image à haute résolution [Termes IGN] pouvoir de résolution spectrale [Termes IGN] reconstruction d'image [Termes IGN] réseau antagoniste génératif [Termes IGN] segmentation sémantique [Termes IGN] vision par ordinateur Index. décimale : THESE Thèses et HDR Résumé : (auteur) Image super-resolution is a classic ill-posed computer vision and image processing problem, addressing the question of how to reconstruct a high-resolution image from its low-resolution counterpart. Current state-of-the-art methods have improved the performance of the single image super-resolution task significantly by benefiting from machine learning and AI-powered algorithms, and more specifically, with the advent of Deep Learning-based approaches. Although these advances allow a machine to learn and have better exploitation of an image and its content, recent methods are still unable to constrain the plausible solution space based on the available contextual information within an image. This limitation mostly results in poor reconstructions, even for well-known types of objects and textures easily recognizable for humans. In this thesis, we aim at proving that the categorical prior, which characterizes the semantic class of a region in an image (e.g., sky, building, plant), is crucial in super-resolution task for reaching a higher reconstruction quality. In particular, we propose several approaches to improve the perceived image quality and generalization capability of deep learning-based methods by exploiting the context and semantic meaning of images. To prove the effectiveness of this categorical information, we first propose a convolutional neural network-based framework that is able to extract and use semantic information to super-resolve a given image by using multitask learning, simultaneously for learning image super-resolution and semantic segmentation. The proposed decoder is forced to explore categorical information during training, as this setting employs only one shared deep network for both semantic segmentation and super-resolution tasks. We further investigate the possibility of using semantic information by a novel objective function to introduce additional spatial control over the training process. We propose penalizing images at different semantic levels using appropriate loss terms by benefiting from our new OBB (Object, Background, and Boundary) labels generated from segmentation labels. Then, we introduce a new test time adaptation-based technique to leverage high-resolution images with perceptually similar context to a given test image to improve the reconstruction quality. We further validate this approach's effectiveness by using a novel numerical experiment analyzing the correlation between filters learned by our network and what we define as `ideal' filters. Finally, we present a generic solution to enable adapting all our previous contributions in this thesis, as well as other recent super-resolution works trained on synthetic datasets, to real-world super-resolution problem. Real-world super-resolution refers to super-resolving images with real degradations caused by physical imaging systems, instead of low-resolution images from simulated datasets assuming a simple and uniform degradation model (i.e., bicubic downsampling). We study and develop an image-to-image translator to map the distribution of real low-resolution images to the well-understood distribution of bicubically downsampled images. This translator is used as a plug-in to integrate real inputs into any super-resolution framework trained on simulated datasets. We carry out extensive qualitative and quantitative experiments for each mentioned contribution, including user studies, to compare our proposed approaches to state-of-the-art method. Note de contenu : 1- Introduction 2- Brief image super-resolution review 3- Extracting image context by multi-task learning 4- Spatial control over image genertion process 5- Test-time adaptation based on perceptual similarity 6- Integrating into real-world SR 7- Conclusion Numéro de notice : 28652 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Thèse étrangère Note de thèse : Thèse de Doctorat : Sciences : EPFL, Lausanne : 2021 DOI : sans En ligne : https://infoscience.epfl.ch/record/286804?ln=fr Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99790

Spatiotemporal patterns of urbanization during the last four decades in Switzerland and their impacts on urban heat islands / Marti Bosch Padros (2021)  

Public Titre : Spatiotemporal patterns of urbanization during the last four decades in Switzerland and their impacts on urban heat islands Type de document : Thèse/HDR Auteurs : Marti Bosch Padros, Auteur ; Jérôme Chenal, Directeur de thèse ; Stéphane Joost, Directeur de thèse Editeur : Lausanne : Ecole Polytechnique Fédérale de Lausanne EPFL Année de publication : 2021 Importance : 145 p. Format : 21 x 30 cm Note générale : bibliographie Thèse présentée pour l'obtention du grade de Docteur ès Sciences Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Analyse spatiale [Termes IGN] aménagement du territoire [Termes IGN] analyse diachronique [Termes IGN] analyse spatio-temporelle [Termes IGN] changement climatique [Termes IGN] croissance urbaine [Termes IGN] densité de population [Termes IGN] espace vert [Termes IGN] étalement urbain [Termes IGN] ilot thermique urbain [Termes IGN] modèle de simulation [Termes IGN] modèle dynamique [Termes IGN] occupation du sol [Termes IGN] paysage urbain [Termes IGN] service écosystémique [Termes IGN] Suisse [Termes IGN] urbanisation Index. décimale : THESE Thèses et HDR Résumé : (auteur) Urbanization is nowadays a global phenomenon which is increasingly concentrating the world’s population in cities. In Switzerland, recent decades have seen an unprecedented loss of arable land due to urbanization, which has triggered amendments in the spatial planning laws with the aim of promoting urban densification. Nevertheless, despite remarkable efforts, the environmental impacts of distinctive urban patterns such as compact cities and urban sprawl remain poorly understood. One of the most remarkable environmental impacts of urbanization is the urban heat island effect, a phenomenon by which urban temperatures are warmer than in its rural surroundings. Central Europe, and therefore Switzerland, is among the regions in the world where temperatures are rising faster and the urban heat island effect is most prominent, which represents a central challenge for spatial planning. Most studies suggest that the urban heat island effect can be aggravated in compact cities, especially when considering the larger share of urban dwellers that are exposed to the highest temperatures. At the same time, the literature on the subject has seen a growing development of mitigation strategies, which suggest that the urban heat island effect can be significantly alleviated by an adequate planning of the building materials and urban green spaces. This doctoral dissertation intends to address the issues expressed above by performing a quantitative evaluation of the spatiotemporal patterns of urbanization in Switzerland and their impact on the urban heat island effect. To that end, the thesis adopts a landscape ecology perspective to quantify urban patterns and to spatially simulate the biophysical processes that underpin the urban heat island effect. The first article presents PyLandStats, an opensource library to compute landscape metrics in a repeatable and reproducible manner. In the second article, such a library is used to evaluate the spatiotemporal patterns of urbanization observed in the urban agglomerations of Bern, Lausanne and Zurich from 1980 to 2016. The results reveal that the outer zones of Bern and Lausanne are still undergoing diffusive urban expansion, whereas infill development is the dominant growth mode in both the inner and outer zones of Zurich. The thesis follows with the development of a spatially-explicit method to simulate urban heat mitigation using a recent model of urban cooling based on three biophysical mechanisms, namely tree shade, evapotranspiration and albedo. The study introduces an automated procedure to calibrate the parameters of the model, and shows that the proposed approach can outperform regression models based on remote sensing features. Then, in the fourth article, such an approach is applied to Lausanne in order to evaluate heat mitigation in a variety of urban greening scenarios which modify both the abundance and spatial configuration of the tree canopy cover. The simulations suggest a potential alleviation of the maximum nighttime temperatures of 2°C, which represents a major reduction of the human exposure to the urban heat island effect. Finally, a concluding chapter summarizes the main contributions of the dissertation and reviews key implications for urban planning in Switzerland. Overall, rather than prescribing urban densification as the customary strategy for spatial development, land use regulations and local plans should incorporate spatially-explicit evaluations of the ecosystem services provided by urban green spaces. Future research should extend the proposed approach to include further ecosystem services and explore trade-offs and spatially design solutions. Note de contenu : 1Urban greening scenarios for urban heat mitigation- Introduction 2- Quantifying spatial patterns of landscapes 3- Spatiotemporal patterns of urbanization in three Swiss urban agglomerations 4- Spatially-explicit simulation of urban heat islands 6- Synthesis and outlook Numéro de notice : 28666 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE/URBANISME Nature : Thèse étrangère Note de thèse : Thèse de Doctorat : Sciences : EPFL : 2021 DOI : sans En ligne : https://infoscience.epfl.ch/record/285477?ln=fr Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99863

Learning stereo reconstruction with deep neural networks / Stepan Tulyakov (2020)  

Permalink

On the adjustment, calibration and orientation of drone photogrammetry and laser-scanning / Emmanuel Clédat (2020)  

Permalink

Machine learning and geographic information systems for large-scale mapping of renewable energy potential / Dan Assouline (2019)  

Permalink

Parameter estimation with GNSS-reflectometry and GNSS synthetic aperture techniques / Miguel Angel Ribot Sanfelix (2018)  

Permalink

Toward a systematic integration of optical remote sensing for inland waters studies / Vincent Maurice Nouchi (2018)  

Permalink

Vehicle dynamic model based navigation for small UAVs / Mehran Khaghani (2018)  

Permalink

Fusion of multi-temporal Sentinel-2 image series and very-high spatial resolution images for detection of urban areas / Cyril Wendl (2017)

Permalink

Vision-based detection of aircrafts and UAVs / Artem Rozantsev (2017)  

Permalink

Saisie de la couverture neigeuse de sites avalancheux par des systèmes aéroportés / Julien Vallet (2002)  

Permalink

Nouveaux cadres de référence et transformations de coordonnées en géomatique / François Golay (2001)

Permalink