Centre dedocumentation
scientifique

Convolutional neural network for traffic signal inference based on GPS traces / Yann Méneroux (2018) 

Public contenu dans Spatial big data and machine learning in GIScience, Workshop at GIScience 2018, Melbourne, Australia, 28 August 2018 / Martin Raubal (2018)     Titre : Convolutional neural network for traffic signal inference based on GPS traces Type de document : Article/Communication Auteurs : Yann Méneroux , Auteur ; V. Dizier, Auteur ; Mathieu Margollé, Auteur ; Marie-Dominique Van Damme , Auteur ; Hiroshi Kanasugi, Auteur ; Arnaud Le Guilcher , Auteur ; Guillaume Saint Pierre, Auteur ; Yugo Kato, Auteur Congrès : Workshop Spatial big data and machine learning (28 aout 2018; Melbourne, Australie), Commanditaire Editeur : Zurich : Eidgenossische Technische Hochschule ETH - Ecole Polytechnique Fédérale de Zurich EPFZ Année de publication : 2018 Projets : 1-Pas de projet / Importance : pp 9 - 12 Format : 21 x 30 cm Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Géomatique [Termes descripteurs IGN] apprentissage automatique [Termes descripteurs IGN] apprentissage profond [Termes descripteurs IGN] données maillées [Termes descripteurs IGN] réseau neuronal convolutif [Termes descripteurs IGN] trace GPS [Termes descripteurs IGN] trafic routier [Termes descripteurs IGN] traitement de données localisées Résumé : (auteur) Map inference techniques aim at using GPS trajectories collected from a fleet of vehicles, to infer geographic information and enrich road map databases. In this paper, we investigate whether a Convolutional Neural Network can detect traffic signals on a raster map of features computed from a large dataset of GPS traces. Experimentation revealed that our model is able to capture traffic signal pattern signature on this very specific case of unnatural input images. Performance indices are encouraging but need to be improved through a more refined tuning of the workflow. Numéro de notice : C2018-053 Thématique : GEOMATIQUE Nature : Communication nature-HAL : ComAvecCL&ActesPubliésIntl DOI : sans Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=91340

Documents numériques

en open access CNN for traffic signal inference ... - pdf éditeur Adobe Acrobat PDF

Spatial big data and machine learning in GIScience, Workshop at GIScience 2018, Melbourne, Australia, 28 August 2018 / Martin Raubal (2018)    

Public Titre : Spatial big data and machine learning in GIScience, Workshop at GIScience 2018, Melbourne, Australia, 28 August 2018 : Proceedings Type de document : Actes de congrès Auteurs : Martin Raubal, Editeur scientifique ; Shaowen Wang, Editeur scientifique ; Mengyu Guo, Editeur scientifique ; David Jonietz, Editeur scientifique ; Peter Kiefer, Editeur scientifique Congrès : Workshop Spatial big data and machine learning (28 aout 2018; Melbourne, Australie), Auteur Editeur : Zurich : Eidgenossische Technische Hochschule ETH - Ecole Polytechnique Fédérale de Zurich EPFZ Année de publication : 2018 Importance : 53 p. Format : 21 x 30 cm Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Bases de données localisées [Termes descripteurs IGN] apprentissage automatique [Termes descripteurs IGN] données localisées [Termes descripteurs IGN] données massives [Termes descripteurs IGN] interface homme-machine [Termes descripteurs IGN] trajectoire Note de contenu : PART 1: TRAJECTORIES - Stay-Move Tree for Summarizing Spatiotemporal Trajectories / Eun-Kyeong Kim - Detection of Unsigned Ephemeral Road Incidents by Visual Cues / Alex Levering, Kourosh Khoshelham, Devis Tuia, and Martin Tomko - Convolutional Neural Network for Traffic Signal Inference based on GPS Traces / Y. Méneroux, V. Dizier, M. Margollé, M.D. Van Damme, H.Kanasugi, A. Le Guilcher, G. Saint Pierre, and Y. Kato - Using Stream Processing to Find Suitable Rides: An Exploration based on New York City Taxi Data / Roswita Tschümperlin, Dominik Bucher, and Joram Schito - Classification of regional dominant movement patterns in trajectories with a convolutional neural network / Can Yang and Gyozo Gidófalvi PART 2: COGNITION & HCI - Spatial Big Data for Human-Computer Interaction / Ioannis Giannopoulos - Unsupervised Clustering of Eye Tracking Data / Fabian Göbel and Henry Martin - Collections of Points of Interest: How to Name Them and Why it Matters / Gengchen Mai, Krzysztof Janowicz, Yingjie Hu, Song Gao, Rui Zhu, Bo Yan, Grant McKenzie, Anagha Uppal, and Blake Regalia PART 3: SPATIAL PATTERNS - A Multi-scale Spatio-temporal Approach to Analysing the changing inequality in the Housing Market during 2001-2014 / Yingyu Feng and Kelvyn Jones - Automated social media content analysis from urban green areas – Case Helsinki / Vuokko Heikinheimo, Henrikki Tenkanen, Tuomo Hiippala, Olle Järv, and Tuuli Toivonen - Long short-term memory networks for county-level corn yield estimation / Haifeng Li, Yudi Wang, Renhai Zhong, Hao Jiang, and Tao Lin - Conditional Adversarial Networks for Multimodal Photo-Realistic Point Cloud Rendering / Torben Peters and Claus Brenner Numéro de notice : 17546 Thématique : GEOMATIQUE Nature : Actes En ligne : http://spatialbigdata.ethz.ch/index.php/proceedings/ Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=91338

Contient

• Convolutional neural network for traffic signal inference based on GPS traces / Yann Méneroux (2018) 

Documents numériques

en open access Spatial big data and machine learning in GIScience 2018 - pdf éditeur Adobe Acrobat PDF

Monitoring rock glaciers by combining photogrammetric and GNSS-based methods / Fabian Neyer (2017)  

Public Titre : Monitoring rock glaciers by combining photogrammetric and GNSS-based methods Type de document : Thèse/HDR Auteurs : Fabian Neyer, Auteur Editeur : Zurich : Schweizerischen Geodatischen Kommission / Commission Géodésique Suisse Année de publication : 2017 Autre Editeur : Zurich : Eidgenossische Technische Hochschule ETH - Ecole Polytechnique Fédérale de Zurich EPFZ Collection : Geodätisch-Geophysikalische Arbeiten in der Schweiz, ISSN 0257-1722 num. 99 Importance : 174 p. Format : 21 x 30 cm ISBN/ISSN/EAN : 978-3-908440-45-1 Note générale : bibliographie thesis submitted to attain the degree of doctor of sciences of ETH Zurich Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications photogrammétriques [Termes descripteurs IGN] Alpes centrales [Termes descripteurs IGN] champ de vitesse [Termes descripteurs IGN] collocation par moindres carrés [Termes descripteurs IGN] déformation de la croute terrestre [Termes descripteurs IGN] données localisées 3D [Termes descripteurs IGN] glacier [Termes descripteurs IGN] image terrestre [Termes descripteurs IGN] matrice de covariance [Termes descripteurs IGN] pergélisol [Termes descripteurs IGN] reconstruction 3D [Termes descripteurs IGN] rocher [Termes descripteurs IGN] série temporelle [Termes descripteurs IGN] Suisse Index. décimale : 33.60 Applications photogrammétriques - usage combiné de la photogrammétrie et de la lasergrammétrie Résumé : (auteur) Rock glaciers are creeping landforms of perennially frozen ground and belong to the permafrost creeping phenomena. They are mainly composed of rock debris that accumulate in areas of high natural erosion. Ice particles between the rocks cause the moving accumulation in steep terrain to dynamically flow downslope. In the Alpine region, these morphological landforms mainly occur at north-facing mountain slopes in high altitudes above the forest boundary and are known for their sensitivity to climate change. For several decades, rock glaciers have been monitored for scientific aims, while advances in surveying technologies increased the interest in such studies since the 1990s. Modern technologies in remote sensing (e.g., airborne imagery or satellite-based measurement techniques) are often combined with measurements from field campaigns, i.e., measurements taken directly on a rock glacier (e.g., GNSS, laser-scanning, ground temperature measurements, etc). The high-level goal is to enhance the process understanding, especially with respect to the changing climate: various studies indicate an extended risk of slope failures in steep frozen bedrock due to the global temperature increase. Early recognition of increased activities help to inform local authorities in the endangered areas about the potential hazard before such an event. The present work is part of the X-Sense project (Nano-Tera.ch), with an interdisciplinary team of scientists that build and operate new low-cost devices for data acquisition, develop new data processing pipelines and algorithms for evaluation, and also gain new insight of natural processes in these regions. Autonomous measurement systems, developed within other work packages in the X-Sense project, observe different permafrost creep areas with high resolution in space and time. Combined with multi-year observations, the derived surface motions are used to obtain an improved process understanding. This work focuses on the photogrammetric image processing in order to retrieve precise surface displacement estimates. More precisely, image sequences, acquired with two permanently installed commercial digital single-reflex cameras, are used to measure topographic changes in the observed permafrost area. By the combination with high resolution GNSS positioning results, the goal is to obtain precise time series of moving rock boulders at different positions within the field of view. Challenges arising from the combination of different data sets, the development of an automatic processing pipeline, and an improvement of the processing strategy in general, are the main tasks of this thesis. The study site is the bordering area above the Grabengufer rock glacier (Mattervalley VS, Switzerland), known as the Grabengufer rock slide. Local topographic conditions allowed only a partially good installation geometry for the photogrammetric reconstruction. With respect to a 3D reconstruction without the use of GNSS coordinates, an accuracy increase of about one order of magnitude could be achieved in case these high-precision solutions were integrated. More specifically, respective standard deviations for the East, North, and Height components of 6, 5, and 2 cm were achieved. The stated accuracy, maintained throughout the measurement period of nearly four years (summer months), was obtained in an area of approximately 80m×80 m, with a mean distance of 80 m from the two cameras. Position time series of moving rock boulders were filtered using the principles of collocation. Analyzing the correlation characteristics of the stochastic signal, an optimal correlation length was computed and used to extract relevant signals from the noise contaminated time series. Velocity was directly estimated as a derived quantity in the collocation process. Furthermore, the techniques of the adaptive collocation approach is presented. This iterative method uses the principles of a dynamically adjusting anisotropic covariance metric. In an example of 2-dimensional velocity fields it is shown that regional compression and extension areas can be extracted. Results indicate that the observed permafrost area has experienced a mean annual acceleration of about 0.1m/Year between the years 2013 and 2015. During the late summer months of 2015, a prominent temporal acceleration was observed. The mean displacement rate was found to be 0.67m/year, whereas the 3-dimensional displacement is dominated by a translation following the gliding surface. An area in the front of the observed field of view was found to have higher displacement rates, especially during the late summer months, thus it detaches from the otherwise relatively homogeneous flow field. The methods and principles presented in this work show the potential of monitoring permafrost surface displacements using permanently installed optical cameras in combination with positioning results from permanently mounted GNSS stations. These principles can easily be transfered to other monitoring applications and thus contribute to a better understanding of such processes. Note de contenu : 1 Introduction 2 Study Site 3 Image-Based Displacement Estimation 4 Object Point Reconstruction 5 Collocation for Time Series Analysis 6 Results 7 Conclusions 8 Outlook Numéro de notice : 17458 Thématique : IMAGERIE Nature : Thèse étrangère Note de thèse : thèse : Sciences : ETHZ : 2016 En ligne : https://www.sgc.ethz.ch/sgc-volumes/sgk-99.pdf Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=89683

Réservation

Réserver ce document

Exemplaires (1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
17458-01	33.60	Livre	Centre de documentation	Photogrammétrie - Lasergrammétrie	Disponible

Advanced modeling and algorithms for high-precision GNSS analysis / Kan Wang (2016)  

Public Titre : Advanced modeling and algorithms for high-precision GNSS analysis Type de document : Thèse/HDR Auteurs : Kan Wang, Auteur Editeur : Zurich : Eidgenossische Technische Hochschule ETH - Ecole Polytechnique Fédérale de Zurich EPFZ Année de publication : 2016 Collection : Dissertationen ETH num. 23188 Note générale : bibliographie thesis submitted to attain the degree of doctor of sciences of ETH Zurich Langues : Anglais (eng) Descripteur : [Termes descripteurs IGN] ambiguïté entière [Termes descripteurs IGN] antenne GPS [Termes descripteurs IGN] centre de phase [Termes descripteurs IGN] données BeiDou [Termes descripteurs IGN] données Galileo [Termes descripteurs IGN] données GPS [Termes descripteurs IGN] double différence [Termes descripteurs IGN] erreur systématique [Termes descripteurs IGN] GPS en mode différentiel [Termes descripteurs IGN] horloge [Termes descripteurs IGN] phase GNSS [Termes descripteurs IGN] positionnement cinématique [Termes descripteurs IGN] récepteur GNSS [Termes descripteurs IGN] récepteur trifréquence [Termes descripteurs IGN] résolution d'ambiguïté [Termes descripteurs IGN] retard ionosphèrique [Termes descripteurs IGN] Suisse [Termes descripteurs IGN] trajet multiple [Vedettes matières IGN] Traitement de données GNSS Résumé : (auteur) In the recent ten years, the Global Navigation Satellite System (GNSS) processing has experienced a fast development in many areas including the increasing number of frequencies, the higher quality of positioning instruments, e.g. the receiver clocks and the satellite clocks, and more integrated modeling and calculation strategies. This thesis includes investigations of different modeling and parameterization methods in modern GNSS positioning with the focus on three important positioning error sources: the receiver clock errors, the phase ambiguities and the ionospheric delays. The thesis shows that making use of the high-quality receiver clocks and applying appropriate receiver clock modeling can help to improve the kinematic height estimates, which are highly correlated with the receiver clock parameters. An efficient pre-elimination and back-substitution strategy of epoch parameters with relative clock constraints between subsequent and near-subsequent epochs has been developed to enable processing of, e.g., high-rate data. A detailed analysis of the relationship between the clock quality and the improvement of kinematic heights has been performed. Studies were also conducted to decorrelate the receiver clock parameters, the kinematic heights and the troposphere parameters. Experiments with real data have shown that appropriate deterministic and stochastic clock models can also be helpful to increase the resolution of the estimated Zenith Path Delay (ZPD) parameters without obvious degradation of the stability of the kinematic heights. The second aspect of the thesis focuses on the resolution of triple-frequency phase ambiguities with different linear combinations. A complete analytical investigation of Geometry-Free (GF) and Ionosphere-Free (IF) triple-frequency phase ambiguity resolution with minimized noise level has been performed for different frequency triplets. The analysis was done separately for the best two linear combinations and the third one. Experiments have shown that the fractional parts and the formal errors of the combined ambiguities of the best two linear combinations are relatively small for Galileo E1, E5b and E5a and GPS L1, L2 and L5 triplets, while the third linear combination remains a challenge. Further analysis with the geostationary satellites of the Beidou Navigation Satellite System (BDS) elaborated in the framework of this thesis has also confirmed that the combined ambiguities from the best two GF and IF linear combinations can be fixed by rounding, while the estimated ambiguities on L1 have relatively large deviations from the values obtained from the traditional dual-frequency double-difference ambiguity resolution. Apart from the triple-frequency ambiguity resolution on the double-difference level, the so-called track-to-track ambiguities between different tracks of the same receiver and the same satellite have also been investigated for the best two triple-frequency linear combinations using GPS L1, L2 and L5 as well as Galileo E1, E5b and E5a observations. The outcome demonstrates that elevation-dependent influences on the observations like Phase Center Variations (PCVs), Phase Center Offsets (PCOs) and multipath are important for the fixing of the track-to-track ambiguities. The combined track-to-track ambiguities using the best two linear combinations are also effective in detecting problems in the observation data. The third aspect of the thesis includes the investigation of the differential ionospheric delays and gradients in the region of Switzerland from 1999 to 2013. In differential Global Positioning System (GPS) positioning, the ionospheric delays for short baselines are in most cases small enough to be ignored, except under extreme conditions, e.g., during ionospheric stormy days, and for applications with high integrity requirements, e.g., during approach and landing of aircrafts. This thesis introduces an algorithm using double-difference phase measurements with resolved phase ambiguities and global ionosphere maps provided by the Center for Orbit Determination in Europe (CODE) to extract the single-difference ionospheric delays, and enabling an automatic and robust processing of the data over 15 years. The results show that the daily maximum slant ionospheric gradients calculated from the differential slant ionopheric delays and the baseline lengths from 1999 to 2013 are below the slant ionosphere gradient boundary of the Conterminous United States (CONUS) ionospheric anomaly threat model. Numéro de notice : 17250 Thématique : POSITIONNEMENT Nature : Thèse étrangère Note de thèse : dissertation : sciences : ETH Zurich : 2016 En ligne : http://dx.doi.org/10.3929/ethz-a-010610972 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=81986

Autonomous navigation in complex nonplanar environments based on laser ranging / Philipp Andreas Krüsi (2016)  

Public Titre : Autonomous navigation in complex nonplanar environments based on laser ranging Type de document : Thèse/HDR Auteurs : Philipp Andreas Krüsi, Auteur Editeur : Zurich : Eidgenossische Technische Hochschule ETH - Ecole Polytechnique Fédérale de Zurich EPFZ Année de publication : 2016 Note générale : bibliographie A thesis submitted to attain the degree of doctor of sciences of ETH Zurich Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes descripteurs IGN] algorithme ICP [Termes descripteurs IGN] carte de nuage de points [Termes descripteurs IGN] données lidar [Termes descripteurs IGN] données localisées 3D [Termes descripteurs IGN] navigation autonome [Termes descripteurs IGN] robot mobile [Termes descripteurs IGN] semis de points [Termes descripteurs IGN] télémètre laser [Termes descripteurs IGN] télémétrie laser aéroporté [Termes descripteurs IGN] télémétrie laser terrestre [Termes descripteurs IGN] temps réel [Termes descripteurs IGN] vision par ordinateur Résumé : (auteur) This thesis addresses the problem of autonomous navigation with ground robots in complex environments, which may be characterized as nonplanar and nonstatic. The goal of the presented research is to enable reliable navigation over large distances in generic indoor and outdoor environments, independent of external localization sources such as a global positioning system (GPS). Focusing on these challenges, algorithms for all building blocks of autonomous navigation—localization, mapping, terrain assessment, motion planning, and motion control—are developed, implemented, integrated, and finally evaluated in extensive field experiments. Sensor-based perception of the environment is a basic requirement for localization and mapping. We propose to use a high-frequency three-dimensional (3D) laser scanner as the main exteroceptive sensor. The advantages of this technology lie in the high density and accuracy of the provided measurements, and their independence of lighting and weather conditions. We develop a highly scalable system for six-dimensional (6D) localization and 3D mapping based on iterative closest point (ICP) matching. A topological/metric map representation, where metric information is kept in spatially constrained local submaps representing vertices in a graph, allows to build consistent large-scale maps without requiring global optimization. Long-term application in dynamic and changing environments is enabled by integrating methods for identifying dynamic objects in the scene and for continuously updating existing submaps. Planning feasible and safe motions for a robotic vehicle requires distinguishing obstacles from traversable terrain. We develop two different algorithms for terrain assessment. The first method is targeted at real-time obstacle detection in the vicinity of the robot. Assuming locally planar terrain, a grid-based obstacle map is built by analyzing the raw laser scans. The second approach is based on dense point cloud maps (which can be obtained from the ICP mapping system) and suitable for planar and nonplanar environments. The algorithm computes the geometry and the traversability of the terrain “on demand” at specific query locations, avoiding any artificial discretization or explicit surface reconstruction. The desired terrain characteristics are estimated based on statistics on the local distribution of map points. Given a specific navigation task, motion planning can be defined as the problem of reasoning about how to act based on the knowledge about the environment. This thesis addresses both local obstacle avoidance and global planning over large distances. Our approach to local planning consists of computing a set of candidate trajectories, which are shaped around nearby obstacles or along a given reference path, and enforced to satisfy the robot’s kinematic constraints. The optimal local trajectory is chosen by evaluating the motion alternatives in terms of guidance towards the goal and traversability of the underlying terrain. For global motion planning, we develop an algorithm embedding the proposed point-cloud-based terrain assessment method, which allows trajectories to be directly planned on 3D point cloud maps. The approach is designed to be suitable for generic nonplanar environments, including rough outdoor terrain, multi-level facilities, and more complex geometries. Piecewise continuous trajectories are computed in the full 6D space of robot poses, while strictly considering the vehicle’s kinematic and dynamic constraints. We apply sampling-based planning algorithms to generate an initial trajectory connecting the desired start and goal poses. Subsequently, the trajectory is locally optimized according to a generic cost function, which may include path length, path curvature, and roughness of the traversed terrain. While enforcing the hard constraints to remain satisfied (terrain contact, traversability, kinodynamic feasibility), the trajectory is iteratively deformed until a local minimum of the cost function is reached. We develop two complete systems for autonomous navigation, integrating these approaches. Combining the ICP-based localization and mapping framework with local obstacle detection and local motion planning, we implement a framework for autonomous route following, commonly referred to as teach and repeat (T&R). After a manually controlled teach run, where a graph of local submaps is built, the robot is able to automatically repeat the learned route, using the recorded maps for localization. Unlike classical T&R systems, our framework is suitable for application in dynamic environments, where the integrated obstacle avoidance scheme allows to detect and circumnavigate obstacles appearing on the reference path. In addition to the T&R approach, we present a second navigation system, integrating the point-cloud-based terrain assessment and global planning algorithms with ICP-based localization and mapping. Given a graph of point cloud maps—typically recorded in a manually controlled survey run—the framework enables navigation within the mapped area without being restricted to known routes. Motion control is implemented by a trajectory tracking controller with integrated real-time collision checking. Together with continuous map updates and frequent replanning of the global trajectory, these techniques enable autonomous navigation in nonplanar, nonstatic environments. Finally, we describe the characteristics of the mobile robot ARTOR, which was set up for the purpose of testing and evaluating the developed algorithms under realistic conditions. ARTOR consists of a six-wheeled, electrically powered base vehicle equipped with sensors, computers, and communication gear. The proposed autonomous navigation algorithms were integrated on the robot and tested in extensive field experiments, demonstrating reliable, GPS-independent navigation over large distances and under greatly varying environmental conditions, in unstructured off-road terrain, multi-level environments, and dynamic urban areas. Numéro de notice : 17367 Thématique : IMAGERIE/INFORMATIQUE Nature : Thèse étrangère Note de thèse : PhD thesis : Sciences : ETH Zurich : 2016 En ligne : http://dx.doi.org/10.3929/ethz-a-010656081 Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=84243

Co-location of geodetic observation techniques in space / Benjamin Männel (2016)  

Permalink

Real-time depth-image-based rendering for viewpoint-variable display on mobile devices / Shuoran Yang (2016)  

Permalink

Study and development of a laser based alignment system for the compact linear collider / Guillaume Stern (2016)  

Permalink

Theoretical and practical aspects of high-rate GNSS geodetic observations / Simon Häberling (2016)  

Permalink

Algorithms for vision-based path following along previously taught paths / Deon George Sabatta (2015)  

Permalink

Data-driven feature learning for high resolution urban land-cover classification / Piotr Andrzej Tokarczyk (2015)  

Permalink

Depth, anisotropy, and water equivalent of snow estimated by radar interferometry and polarimetry / Silvan Leinss (2015)  

Permalink

To die or not to die: Forest dynamics in Switzerland under climate change / Nicolas Bircher (2015)  

Permalink

Inverse Procedural Modelling and Applications / Julien Weissenberg (2014)  

Permalink

Mapping literature : spatial data modelling and automated cartographic visualisation of fictional spaces / Anne-Kathrin Weber (2014)  

Permalink