6949 research outputs found
Sort by
Sensor fusion in localization, mapping and tracking
No full textMaking autonomous driving possible requires extensive information about the surroundings as well as the state of the vehicle. While specific information can be obtained through singular sensors, a full estimation requires a multi sensory approach, including redundant sources of information to increase robustness. This thesis gives an overview of tasks that arise in sensor fusion in autonomous driving, and presents solutions at a high level of detail, including derivations and parameters where required to enable re-implementation. The thesis includes theoretical considerations of the approaches as well as practical evaluations. Evaluations are also included for approaches that did not prove to solve their tasks robustly. This follows the belief that both results further the state of the art by giving researchers ideas about suitable and unsuitable approaches, where otherwise the unsuitable approaches may be re-implemented multiple times with similar results. The thesis focuses on model-based methods, also referred to in the following as classical methods, with a special focus on probabilistic and evidential theories.
Methods based on deep learning are explicitly not covered to maintain explainability and robustness which would otherwise strongly rely on the available training data. The main focus of the work lies in three main fields of autonomous driving: localization, which estimates the state of the ego-vehicle, mapping or obstacle detection, where drivable areas are identified, and object detection and tracking, which estimates the state of all surrounding traffic participants. All algorithms are designed with the requirements of autonomous driving in mind, with a focus on robustness, real-time capability and usability of the approaches in all potential scenarios that may arise in urban driving.
In localization the state of the vehicle is determined. While traditionally global positioning systems such as a Global Navigation Satellite System (GNSS) are often used for this task, they are prone to errors and may produce jumps in the position estimate which may cause unexpected and dangerous behavior. The focus of research in this thesis is the development of a localization system which produces a smooth state estimate without any jumps. For this two localization approaches are developed and executed in parallel. One localization is performed without global information to avoid jumps. This however only provides odometry, which drifts over time and does not give global positioning. To provide this information the second localization includes GNSS information, thus providing a global estimate which is free of global drift. Additionally the use of LiDAR odometry for improving the localization accuracy is evaluated.
For mapping the focus of this thesis is on providing a computationally efficient mapping system which is capable of being used in arbitrarily large areas with no predefined size. This is achieved by mapping only the direct environment of the vehicle, with older information in the map being discarded. This is motivated by the observation that the environment in autonomous driving is highly dynamic and must be mapped anew every time the vehicles sensors observe an area. The provided map gives subsequent algorithms information about areas where the vehicle can or cannot drive. For this an occupancy
grid map is used, which discretizes the map into cells of a fixed size, with each cell estimating whether its corresponding space in the world is occupied. However the grid map is not created for the entire area which could potentially be visited, as this may be very large and potentially impossible to represent in the working memory. Instead the map is created only for a window around the vehicle, with the vehicle roughly in the center. A hierarchical map organization is used to allow efficient moving of the window as the vehicle moves through an area. For the hierarchical map different data structures are evaluated for their time and space complexity in order to find the most suitable implementation for the presented mapping approach.
Finally for tracking a late-fusion approach to the multi-sensor fusion task of estimating states of all other traffic participants is presented. Object detections are obtained from LiDAR, camera and Radar sensors, with an additional source of information being obtained from vehicle-to-everything communication which is also fused in the late fusion. The late fusion is developed for easy extendability and with arbitrary object detection algorithms in mind. For the first evaluation it relies on black box object detections provided by the sensors. In the second part of the research in object tracking multiple algorithms for object detection on LiDAR data are evaluated for the use in the object tracking framework to ease the reliance on black box implementations. A focus is set on detecting objects from motion, where three different approaches are evaluated for motion estimation in LiDAR data: LiDAR optical flow, evidential dynamic mapping and normal distribution transforms. The thesis contains both theoretical contributions and practical implementation considerations for the presented approaches with a high degree of detail including all necessary derivations. All results are implemented and evaluated on an autonomous vehicle and real-world data. With the developed algorithms autonomous driving is realized for urban areas
GIS-based workflow for identifying renewable heat potentials and an optimal strategy for LowEx district heating
No full textTo reach climate protection goals, Germany aims to significantly reduce emissions through the heat transition. Providing renewable heat sources in urban areas presents one of the greatest challenges due to the limited space, yet high energy demand. This thesis addresses the issue by creating a workflow based on GIS (Geographic Information Systems) to quickly identify socalled “sub-areas” where low temperature (LowEx) district heating could be suitable. The heating concept shall connect sources of shallow geothermal energy within an observed area and enable a more effective distribution of heat. Photovoltaics (PV) on roofs as well as parking lots shall counter the increased power demand of the decentralized heat pumps. The workflow involves estimations of annual technical potentials and the identification of sub-areas using ArcGIS Pro, Polysun, and Earth Energy Designer. This was created based on and applied to the district of Neu-Schwachhausen. Bremen. Scenarios (S) of 2022, 2030, 2038-a, and 2038-b were observed, varying the availability of the selected technologies (borehole heat exchanger and PV) and the level of heat demand for space and water heating. The sub-areas were evaluated regarding overall quality based on selected criteria. Results show that such a concept of LowEx DH is the ideal heating strategy for the district if the circumstances in S-2038-b can be achieved. Most neighborhoods would be technically suitable, especially due to the 300 m deep boreholes and heat demand reductions. Considering the cumulated values, the required power for heat pumps could be covered by PV alone in many of the areas, underlining the necessity of efficient energy storage in reality. Solar parking lots have a less substantial impact overall compared to roof potentials, though a detailed feasibility study is recommended to confirm this. This thesis also shows the importance of sharing local sources between neighborhoods, in addition to using large sports fields, to efficiently maximize area usage. Prioritizing these elements would lead to higher energy self-sufficiency in the subareas. In further research, the workflow can be extended by adding an economic parameter, seasonal fluctuations, and improved automation
Entwurf und Anwendung verteilter Hardwarearchitekturen in der digitalen Signalverarbeitung
No full textDiese Dissertation beschäftigt sich mit dem Entwurf von Schaltungen zu Algorithmen aus der digitalen Signalverarbeitung. Das Augenmerk gilt dabei der Implementierung auf mehreren Untersystemen, aus denen das Gesamtsystem zusammengesetzt wird. Hintergrund für dieses Vorgehen sind die Vorteile, die aus der funktionalen Partitionierung oder Datenpartitionierung hervorgehen. Bekannt sind vor allem die höhere Ausbeute bei der Halbleiterfertigung und Technologieunabhängigkeit bei separat hergestellten Untersystemen aus den Chipletarchitekturen großer Prozessorhersteller. Das Prinzip wird in dieser Arbeit auf andere Signalverarbeitungsalgorithmen angewandt und es wird untersucht, welche Nachteile und Herausforderungen entstehen. Als Anwendungen dienen ein massive MIMO Entzerrer und ein Radarsignalverarbeitungssystem. An den beiden Beispielen werden die funktionale Partitionierung sowie die Datenpartitionierung gezeigt und es werden ASIC- und FPGA-Prototypen vorgestellt. Als kritisch für die Etablierung von Chipletarchitekturen gilt die Vertrauenswürdigkeit, da mehrere Instanzen an der Fertigung des Gesamtsystems beschäftigt sind. Diese Arbeit nimmt eine Sicherheitsanalyse vor und beleuchtet die Unterschiede zwischen verteilten und nicht verteilten Systemen bezüglich der Vertrauenswürdigkeit und Cybersicherheit in Chipletarchitekturen
Canopy gap dynamics in mangrove forests: exploring global patterns and drivers
No full textMangrove forests are located at the interface between land and sea in tropical and subtropical lati-tudes. They provide several valuable ecosystem services, including carbon sequestration, habitat for diverse fauna, coastal protection, and serving as a source of fuel and timber for coastal com-munities. However, the unique location of mangrove forests in a highly dynamic environment, makes them susceptible to disturbances which leads to the formation of canopy gaps.
Mangrove canopy gaps may counteract senescence, contributing to maintaining the mangrove for-est in a rejuvenated and regenerated state. The rejuvenation and regeneration of canopy gaps have implications for the integrity of the numerous valuable ecosystem services that mangrove forests offer. Despite the socio-ecological implications of canopy gaps, knowledge regarding their global and local extent, drivers, occurrences, densities, and closure rates remains limited.
This thesis addresses the knowledge limitation by conducting a comprehensive investigation of the distribution patterns and dynamics of canopy gaps in mangrove forests on both global and local scales. The investigation employs a multifaceted approach, encompassing extensive literature reviews, remote sensing techniques, and predictive models, to explain the patterns of canopy gaps formation, closure dynamics and reveal their underlying drivers while validating their regenera-tion capacity.
Canopy gaps are found in 133 mangrove patches distributed across 35 countries spanning Ameri-ca, Africa, Asia, and Oceania. Significant variations in canopy gap sizes, canopy gap densities, and percentage of canopy gaps coverage in mangrove patches across different regions were ob-served. The occurrence of canopy gaps on a global scale is mainly driven by lightning strikes, and precipitation of the coldest quarter, while their density is driven by lightning strikes, the precipita-tion of the wettest and driest months, and the maximum temperature of the warmest month. Overall, these climatic factors have the potential to act synergistically thus contributing to canopy gap occurrences and density within a given mangrove forest patch.
On a local scale, the thesis showed clustered spatiotemporal patterns in South Africa’s largest mangrove forest at uMhlathuze (80% of the total mangrove coverage in the country) near Rich-ards Bay. Beachwood canopy gaps primarily exhibited random patterns with some spatial cluster-ing, along with random temporal patterns. The patterns at both sites support the hypothesis that lightning strikes, insects or pathogen attacks or competition potentially contribute to canopy gap formation. Spatial distribution of canopy gaps was linked to high canopy at both uMhlathuze and Beachwood, supporting the lightning strikes hypothesis. Canopy gaps at uMhlathuze remained open for at least 23 years. In contrast, no canopy gap at Beachwood had closed over the time span of 18 years that the study covers. These findings highlight the need for active (re-)establishment of canopy gaps, as the very slow natural regeneration might result in loss of valua-ble ecosystem services provided by mangroves, such as carbon sequestration and long-term stor-age of carbon.
Furthermore, the canopy gap closure dynamics and factors influencing their closure across 10 countries and two biogeographical realms—the Atlantic East Pacific and Indo West Pacific were investigated. At higher latitudes above the equator a pattern of relatively shorter canopy gap clo-sure durations and increased annual percentage of canopy gap closure was observed. Approxi-mately 70-100% of the canopy gaps had undergone closure in eight countries. Conversely, during the timeframe encompassed by the study, over 70% of the canopy gaps in Australia exhibited a persistent lack of closure for at least 18 years. Canopy gap closure duration was found to be sig-nificantly influenced by the mean temperature of the wettest quarter of the year. Similarly, the annual percentage of canopy gaps closing was significantly influenced by mean temperature of the wettest quarter, mean temperature of the driest quarter, pH, and salinity. This highlights the potential impact of climatic environmental parameters on canopy gap closure dynamics.
The thesis emphasizes the significance of prioritizing the (re-)establishment of mangrove forest areas with non-closing gaps. This is crucial for ensuring the integrity of long-term carbon storage, coastal protection, habitats for diverse fauna, and a sustainable timber supply that supports local livelihoods under climate change.
Overall, this thesis contributes to providing baseline data on mangrove areas with canopy gaps and their potential drivers, both globally and on regional and local scales. It highlights the factors influencing canopy gap closures and mangrove areas facing a lack of regeneration, emphasizing the urgent need for human assistance in (re-)establishing those canopy gaps
Eignen sich agentenbasierte Simulationsmodelle zur Value-at-Risk Prognose an Finanzmärkten?
No full textIn dieser Thesis wird die Vorhersagekraft von Value-at-Risk-Prognosen analysiert, welche von agentenbasierten Modellen erstellt werden. Um eine allgemeine Approximation der Modellparameter zu erhalten, wird die erste Kalibrierung mit der Methode der simulierten Momente durchgeführt. Für eine Abschätzung der ABM am aktuellen Zeitrand, wird anschließend eine Rolling-Window Maximum-Likelihood-Kalibrierung sowie eine Rolling-Window sequenzielle Monte-Carlo-Schätzung angewandt. Die VaR-Prognosen werden dann durch eine weitere zeitliche Iteration der Modelle generiert. Die Ergebnisse zeigen, dass ABM nicht nur für VaR-Prognosen geeignet, sondern auch in der Lage sind, die Prognosegüte gängiger Benchmarkmodelle zu übertreffen. Insbesondere kann festgestellt werden, dass ABM in hochvolatilen Rezessionsphasen besser abschneiden als die in der Praxis dafür verwendeten GARCH-Modelle, wodurch ABM die beste Wahl für VaR-Prognosen in Zeiten mit hohem Verlustrisiko sind
Macht. Gemeinwesenarbeit. Community Organizing? Wie Selbstorganisation und Handlungsfähigkeit von Nachbar*innen in der GWA gestärkt werden kann.
No full textDie vorliegende Masterarbeit bietet zunächst einen geschichlichen Überblick über die Entwicklung der GWA sowie die Rezeption von Community Organizing in Deutschland. Mittels qualitativer Forschung wird darauf aufbauend untersucht, inwieweit das im zweiten GWA-Qualitätsstandard „Stärkung von Handlungsfähigkeit und Selbstorganisation“ genannte Community Organizing, bzw. CO-Elemente in der GWA-Praxis - speziell in Niedersachsen – heute zur Anwendung kommt. Zudem wird herausgearbeitet, welche Rahmenbedingungen für die Stärkung von Handlungsfähgkeit und Selbstorganisation sowohl aus der Sicht der Hauptamtlichen als auch aktiven Stadtteilbewohner*innen als hinderlich bzw. förderlich wahrgenommen werden. Im Rahmen der Analyse gab es zudem weitere Erkenntnisse in Bezug auf die Haltung der Hauptamtlichen in Bezug auf Ehrenamt und Aktivierung
Der Briefroman als Medium weiblicher Kritik in der Aufklärung – Einblicke in Caroline Rudolphis »Gemälde weiblicher Erziehung«
No full textPublikation der U Bremen und U Wien internationalen Studierendenkonferenz »Debattieren, Opponieren, Protestieren – Interdisziplinäre Perspektiven auf sprachliche Praktiken des Widersprechens« 2023.586
›Wir‹ gegen ›die Anderen‹ – Operationalisierung des kollektiven Identitätsbegriffs für die linguistische Diskursforschung am Beispiel europäischer Integrationsdiskurse
No full textPublikation der U Bremen und U Wien internationalen Studierendenkonferenz »Debattieren, Opponieren, Protestieren – Interdisziplinäre Perspektiven auf sprachliche Praktiken des Widersprechens« 2023.384
Multimodal and collaborative interaction for visual data exploration
No full textAs we generate and encounter vast amounts of data every day, the need to support human-data interaction increases. This dissertation investigates how different interaction modalities and devices can support data experts to visually explore and make sense of data, individually and collaboratively. Through a series of empirical studies applying mixed methods, I study how experts interact and wish to interact with spatio-temporal data on tablets and large vertical displays at the workplace.
While data exploration and sensemaking usually take place on a desktop computer, there is a diverse range of computing devices that provide novel ways of interacting beyond the standard mouse and keyboard input devices used for WIMP (windows, icons, menus, pointers) interfaces. Therefore, I explore how different interaction modalities, such as touch, speech, pen, and mid-air gestures, can support exploratory and sensemaking tasks on interactive surfaces. The starting point of this dissertation is a visualization design study with social policy researchers to study data-driven work in the context of a real-world scenario. Through this design study, the dissertation contributes the first formal evaluation of co-creation as a methodology for visualization design as well as the data and task abstractions derived from the social policy research domain. After defining the design requirements of the domain experts, the dissertation focuses on the interaction design of visualization systems that support the data-driven tasks. A comparative evaluation on visual data exploration between desktop and tablet-based workplaces revealed that experts apply different interaction strategies to solve similar tasks across devices, making use of different views and interaction techniques.
Following up on the single-user scenario, I examine how pairs of experts interact with visualization systems, and with each other, in the context of co-located and synchronous work. The dissertation presents how experts wish to interact on large vertical displays through an elicitation study with touch, pen, speech, and mid-air gestures. The dominance of speech interaction among user preferences leads to an in-depth exploratory study on the role of speech in collaborative work. Despite its challenges, speech interaction benefits awareness and is evenly present in closely coupled and loosely coupled collaboration. Overall, participants prefer interacting unimodally, changing modalities depending on the task and the distance from the display. The dissertation contributes the characterization of interaction patterns and strategies on multimodal visual systems, in individual and collaborative scenarios. Based on the findings on performance, user experience, and interaction choices, the dissertation provides a series of design considerations for multimodal and collaborative systems to support data exploration and sensemaking, together with two systems that enable individuals and small groups to perform such tasks
