No description available.

No description available.

Utgångspunkten för denna kandidatuppsats är att den byggda miljön påverkar våra liv och vårt handlingsutrymme. Den byggda miljön kan möjliggöra eller begränsa olika typer av aktiviteter. Leken är en av de mest grundläggande aktiviteterna i vår sociala utveckling och är en del av den sociala interaktionen med andra människor. Leken förknippas ofta med barn, men leken förkommer och är viktig för alla ålderskategorier.Syftet med studien är att ta reda på hur den byggda miljön kan skapa möjligheter för leken att etableras. Vi undersöker om man kan planera och styra leken och vad leken kan tillföra i ett miljonprogramsområde.Många av de miljonprogramsområden som byggdes i Sverige under 1960- och 70-talet är ofta problemtyngda och den byggda miljön lider av ett eftersatt underhåll. Området Gårdsten i Göteborg är ett område som genomgått ombyggnadsprojekt med inriktning på hållbarhetsfrågor, med de sociala frågorna i huvudfokus. Vi har genomfört en fallstudie av området och genom analys utvärderat hur den byggda miljön skapar förutsättningar för lek. / The basis for this candidate essay is that the built environment affects our lives and our freedom of action. The built environment can facilitate or limit different types of activities. Play is one of the most basic things in our social development and is part of the social interaction with other people. The play is often associated with children, but the play is as important for all age groups.The purpose of this study is to investigate how the built environment can support the opportunity for play to establish itself in one place. We investigate to what extent you can plan and control the play and what play can supply to a Million Programme area.Many of the Million Programme areas that were built in Sweden during the 1960s and 70s are often problematic and the built environment suffers from deferred maintenance. The area Gårdsten in Gothenburg is an area that has undergone renovation projects with a focus on sustainability issues, with social issues in the main focus. We have conducted a case study of the area and through analysis evaluated how the built environment can create conditions for play.

This thesis explores the relationship between surface elevation roughness and tornado frequency throughout the eastern United States. It builds upon previous studies which demonstrated a negative relationship between roughness and tornado frequency for the Great Plains and Arkansas. A generalized linear model with tornado frequency as the response variable and roughness and population density as the predictors is generated. This model demonstrates that increased roughness is associated with decreased tornado frequency at the scale of the entire eastern United States, especially where roughness is greater than 20 meters. The methods are also performed for 13 smaller regions within the eastern US, but an effect of roughness is only confirmed for the regions encompassing the Great Lakes and central Appalachians. From these results, it is concluded that mountain ranges, where roughness exceeds 20 meters, clearly inhibit tornado activity, but there is little evidence that smaller terrain variations have the same effect.

With an increasing demand on natural resources, more efficient prospecting techniques need to be developed. One important tool is geophysical methodologies. As technology develops so do these methods and availability of high-resolution information; however if this information is not properly corrected biased results are achieved. This thesis intends to explore common limitations faced by modern geophysical surveys. Processing and interpreting of geophysical data is often accomplished in frequency domain due to speed and efficiency; however this often leads to non-geologically correct results. A spatial domain filter based on potential field signal curvature analysis is a proposed alternative. By isolating specific curvatures, one is isolating specific frequencies, which are generated by sources at particular depths. The method was applied to synthetic and real-world datasets. Following filtering two analytic routines were applied, which showed that the spatially filtered datasets provided cleaner results. Terrain corrections have always been applied to gravity datasets, but rarely are terrain corrections implemented as a pre-processing step in magnetic survey interpretation. Therefore, interpretations based on anomalies from non-corrected magnetic data may be of non-geological features. In a magnetic survey conducted in the mid-eighties, magnetic lows were associated with alteration; however, at that time of initial interpretation no terrain correction was applied. This dataset was revisited and terrain corrected, which showed that the magnetic lows were associated with unaccounted bathymetry. The Bathurst Mining Camp (BMC) is one of Canada's most important base metal mines, but is threatened by a fluctuating mineral resources market. By using high resolution geophysical surveys potential mineral reserves may be located. However, in order to do so a better understanding of geology is necessary, which is often difficult due to limited outcrops. Through the processing and interpretation of recent geophysical datasets, a revised geological map of a selected portion of the BMC has been developed. / Thesis / Master of Science (MSc)

No / We propose a new dynamic simulation scheme for large-scale flood hazard modelling and prevention. The approach consists of a number of core parts: Digital terrain modelling with GIS data, Nona-tree space partitions (NTSP), Automatic River object recognition and registration, and a flood spreading model. The digital terrain modelling method allows the creation of a geometric real terrain model for augmented 3D environments with very large GIS data, and it can also use information gathered from aviation and satellite images with a ROAM algorithm. A spatial image segmentation scheme is described for river and flood identification and for a 3D terrain map of flooding region growth and visualisation. The region merging is then implemented by adopting Flood Region Spreading Algorithm (FRSA). Compared with the conventional methods, our approach has the advantages of being capable of realistically visualising the flooding in geometrically-real 3D environments, of handling dynamic flood behaviour in real-time and of dealing with very large-scale data modelling and visualisation.

Operation of Autonomous Mobile Robots (AMRs) of all forms that include wheeled ground vehicles, quadrupeds and humanoids in dynamically changing GPS denied environments without a-priori maps, exclusively using onboard sensors, is an unsolved problem that has potential to transform the economy, and vastly improve humanity's capabilities with improvements to agriculture, manufacturing, disaster response, military and space exploration. Conventional AMR automation approaches are modularized into perception, motion planning and control which is computationally inefficient, and requires explicit feature extraction and engineering, that inhibits generalization, and deployment at scale. Few works have focused on real-world end-to-end approaches that directly map sensor inputs to control outputs due to the large amount of well curated training data required for supervised Deep Learning (DL) which is time consuming and labor intensive to collect and label, and sample inefficiency and challenges to bridging the simulation to reality gap using Deep Reinforcement Learning (DRL). This dissertation presents a novel method to efficiently train DRL with significantly fewer samples in a constrained racetrack environment at physical limits in simulation, transferred zero-shot to the real-world for robust end-to-end AMR navigation. The representation learned in a compact parameter space with 2 fully connected layers with 64 nodes each is demonstrated to exhibit emergent behavior for Out-of-Distribution (OOD) generalization to navigation in new environments that include unstructured terrain without maps, dynamic obstacle avoidance, and navigation to objects of interest with vision input that encompass low light scenarios with the addition of a night vision camera. The learned policy outperforms conventional navigation algorithms while consuming a fraction of the computation resources, enabling execution on a range of AMR forms with varying embedded computer payloads. / Doctor of Philosophy / Robots with wheels or legs to move around environments improve humanity's capabilities in many applications such as agriculture, manufacturing, and space exploration. Reliable, robust mobile robots have the potential to significantly improve the economy. A key component of mobility is navigation to either explore the surrounding environment, or travel to a goal position or object of interest by avoiding stationary, and dynamic obstacles. This is a complex problem that has no reliable solution, which is one of the main reasons robots are not present everywhere, assisting people in various tasks. Past and current approaches involve first mapping an environment, then planning a collision-free path, and finally executing motor signals to traverse along the path. This has several limitations due to the lack of detailed pre-made maps, and inability to operate in previously unseen, dynamic environments. Furthermore, these modular methods require high computation resources due to the large number of calculations required for each step that prevents high real-time speed, and functionality in small robots with limited weight capacity for onboard computers, that are beneficial for reconnaissance, and exploration tasks. This dissertation presents a novel Artificial Intelligence (AI) method for robot navigation that is more computationally efficient than current approaches, with better performance. The AI model is trained to race in simulation at multiple times real-time speed for cost-effective, accelerated training, and transferred to a physical mobile robot where it retains its training experience, and generalizes to navigation in new environments without maps, with exploratory behavior, and dynamic obstacle avoidance capabilities.

The concept of simulation-based engineering has been embraced by virtually every research and industry sector (Sinha, Liang et al. 2001; Mocko and Fenves 2003). Engineering and science communities have become increasingly aware that computer simulation is an indispensable tool for resolving a multitude of scientific and technological problems. It is clearly desirable to gain a reliable perspective on the behaviour of a system early in the design stage, long before building costly prototypes (Chul and Ro 2002; Letherwood, Gunter et al. 2004; Makarand Datar 2007; Ersal, Fathy et al. 2008; Mueller, Ferris et al. 2009). Simulation tools have become a critical part of the automotive industry due to their ability to reduce the time and money spent in the development process. Terrain is the principle source of vertical excitation to the vehicle and must be accurately represented in order to correctly predict the vehicle response in simulation. In this dissertation, non-deformable terrain surfaces are defined as a sequence of vectors, where each vector comprises terrain heights at locations oriented perpendicular to the direction of travel. The evolution and implications of terrain surface measurement techniques and existing methods for correcting INS drift are reviewed as a framework for a new compensation method for INS drift in terrain surface measurements. Each measurement is considered a combination of the true surface and the error surface, defined on a Hilbert vector space, in which the error is decomposed into drift (global error) and noise (local error). It is also desirable to develop a compact, path-specific, terrain surface representation that exploits the inherent anisotropicity in terrain over which vehicles traverse. In order to obtain this, a set of analytic basis vectors is formed from Gegenbauer polynomials, parameterized to approximate the empirical basis vectors of the true terrain surface. It is also desirable to evaluate vehicle models and tire models over a wide range of terrain types, but it is computationally impractical to store long distances of every terrain surface variation. This dissertation examines the terrain surface, rather than the terrain profile, to maximize the information available to the tire model (i.e. wheel path data). A method to decompose the terrain surface as a combination of deterministic and stochastic components is also developed. / Ph. D.

Le complexe des glissements de terrain de Betsiamites, dans l'estuaire du SaintLaurent, au pourtour du delta de la rivière Betsiamites sur la CôteNord, (Québec. Canada) résulte d'au moins trois épisodes de glissements de terrain. Ces mouvements de masse, d'une ampleur unique au Québec, ont causé le déplacement d'un volume total de dépôts estimé à 2000 millions de m³ (2 km³ ). Cette thèse contient l'analyse détaillée des éléments géologiques et géotechniques relatifs aux événements de mouvements de masse dans la région de Betsiamites. Des levés géophysiques subaériens et sousmarins intégrés aux résultats des données de forages et d'essais géotechniques in situ ont permis de déterminer l'architecture de la région du delta de la rivière Betsiamites et de reconstituer les principaux événements de glissements de terrain. Ainsi, la cicatrice sousmarine du complexe des glissements de terrain de Betsiamites est le résultat d'un premier mouvement de masse daté à 9250 cal BP dont le volume minimal des matériaux déplacés est estimé à 200 millions de m³. Le second événement de glissement de terrain sousmarin daté de 7250 cal BP a mobilisé un volume de dépôts estimé à 1300 millions de m³ sur une superficie de 54 km². Cette thèse démontre aussi que la portion subaérienne de la cicatrice du complexe des glissements de terrain de Betsiamites est le résultat du glissement de Colombier en 1663. Ce dernier glissement s'est déroulé en quatre phases successives dont la première a eu lieu dans le domaine sousmarin et les suivantes dans le domaine subaérien. Le séisme du 5 février 1663 a causé la une première rupture dans l'environnement sousmarin qui a, par la suite, rapidement déclenché deux grandes coulées argileuses et un étalement latéral. Lors des quatre phases de rupture du glissement de Colombier, le volume des dépôts possiblement déplacés pourrait totaliser 530 millions de m³ sur une superficie de 20 km². L'événement de Colombier est l'un des plus importants glissements subaériens documentés au Canada. La présence des cicatrices sousmarines laissées par l'événement de glissement sousmarin de Betsiamites pourrait avoir agit comme facteurs de prédisposition au déclenchement des ruptures de l'événement de Colombier.

Dans les dernières années, beaucoup de progrès a été fait dans le domaine des voitures autonomes. Plusieurs grandes compagnies travaillent à créer un véhicule robuste et sûr. Pour réaliser cette tâche, ces voitures utilisent un lidar pour la localisation et pour la cartographie. Iterative Closest Point (ICP)est un algorithme de recalage de points utilisé pour la cartographie basé sur les lidars. Ce mémoire explore des approches pour améliorer le minimisateur d’erreur d’ICP. La première approche est une analyse en profondeur des filtres à données aberrantes. Quatorze des filtres les plus communs (incluant les M-estimateurs) ont été testés dans différents types d’environnement, pour un total de plus de 2 millions de recalages. Les résultats expérimentaux montrent que la plupart des filtres ont des performances similaires, s’ils sont correctement paramétrés. Néanmoins, les filtres comme Var.Trim., Cauchy et Cauchy MAD sont plus stables à travers tous les types environnements testés. La deuxième approche explore les possibilités de la cartographie à grande échelle à l’aide de lidar dans la forêt boréale. La cartographie avec un lidar est souvent basée sur des techniques de Simultaneous Localization and Mapping (SLAM) utilisant un graphe de poses, celui-ci fusionne ensemble ICP, les positions Global Navigation Satellite System (GNSS) et les mesures de l’Inertial Measurement Unit (IMU). Nous proposons une approche alternative qui fusionne ses capteurs directement dans l’étape de minimisation d’ICP. Nous avons réussi à créer une carte ayant 4.1 km de tracés de motoneige et de chemins étroits. Cette carte est localement et globalement cohérente. / In recent years a lot of progress has been made in the development of self-driving cars. Multiple big companies are working on creating a safe and robust autonomous vehicle . To make this task possible, theses vehicles rely on lidar sensors for localization and mapping. Iterative Closest Point (ICP) is a registration algorithm used in lidar-based mapping. This thesis explored approaches to improve the error minimization of ICP. The first approach is an in-depth analysis of outlier filters. Fourteen of the most common outlier filters (such as M-estimators) have been tested in different types of environments, for a total of more than two million registrations. The experimental results show that most outlier filters have a similar performance if they are correctly tuned. Nonetheless, filters such as Var.Trim., Cauchy, and Cauchy MAD are more stable against different environment types. The second approach explores the possibilities of large-scale lidar mapping in a boreal forest. Lidar mapping is often based on the SLAM technique relying on pose graph optimization, which fuses the ICP algorithm, GNSS positioning, and IMU measurements. To handle those sensors directly within theICP minimization process, we propose an alternative technique of embedding external constraints. We manage to create a crisp and globally consistent map of 4.1 km of snowmobile trails and narrow walkable trails. These two approaches show how ICP can be improved through the modification of a single step of the ICP’s pipeline.