University of Siegen
Permanent URI for this communityhttps://dspace.ub.uni-siegen.de/handle/ubsi/7
Browse
107 results
Search Results
- Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Utilization of the Brinkman penalization to represent geometries in a high-order discontinuous Galerkin scheme on octree meshesWe investigate the suitability of the Brinkman penalization method in the context of a high-order discontinuous Galerkin scheme to represent wall boundaries in compressible flow simulations. To evaluate the accuracy of the wall model in the numerical scheme, we use setups with symmetric reflections at the wall. High-order approximations are attractive as they require few degrees of freedom to represent smooth solutions. Low memory requirements are an essential property on modern computing systems with limited memory bandwidth and capability. The high-order discretization is especially useful to represent long traveling waves, due to their small dissipation and dispersion errors. An application where this is important is the direct simulation of aeroacoustic phenomena arising from the fluid motion around obstacles. A significant problem for high-order methods is the proper definition of wall boundary conditions. The description of surfaces needs to match the discretization scheme. One option to achieve a high-order boundary description is to deform elements at the boundary into curved elements. However, creating such curved elements is delicate and prone to numerical instabilities. Immersed boundaries offer an alternative that does not require a modification of the mesh. The Brinkman penalization is such a scheme that allows us to maintain cubical elements and thereby the utilization of efficient numerical algorithms exploiting symmetry properties of the multi-dimensional basis functions. We explain the Brinkman penalization method and its application in our open-source implementation of the discontinuous Galerkin scheme, Ateles. The core of this presentation is the investigation of various penalization parameters. While we investigate the fundamental properties with one-dimensional setups, a two-dimensional reflection of an acoustic pulse at a cylinder shows how the presented method can accurately represent curved walls and maintains the symmetry of the resulting wave patterns.Source Type:Article3 12 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Pulse based Time-of-Flight range sensing(2018); ; Pulse-based Time-of-Flight (PB-ToF) cameras are an attractive alternative range imaging approach, compared to the widely commercialized Amplitude Modulated Continuous-Wave Time-of-Flight (AMCW-ToF) approach. This paper presents an in-depth evaluation of a PB-ToF camera prototype based on the Hamamatsu area sensor S11963-01CR. We evaluate different ToF-related effects, i.e., temperature drift, systematic error, depth inhomogeneity, multi-path effects, and motion artefacts. Furthermore, we evaluate the systematic error of the system in more detail, and introduce novel concepts to improve the quality of range measurements by modifying the mode of operation of the PB-ToF camera. Finally, we describe the means of measuring the gate response of the PB-ToF sensor and using this information for PB-ToF sensor simulationSource Type:Article2 19 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Simulationssystem zur Interaktion mit realen Fischen unter Verwendung von Analyse-durch-Synthese-Verfahren(2019)The concept of virtual reality (VR) has become very popular and the epitome of progress. Most VR systems are designed for humans. In the present work, on the other hand, a VR system for fish is developed, which is used in behavioral research. Through the interaction of a virtual with a real fish, new possibilities are created and existing work ows are simplified. The aim of this work is the development of a test stand for visual, interactive behavioral research experiments. During this a virtual 3D fish is used as a stimulus during mate choice experiments and is able to interact with the real test fish. As a basis, a system architecture was designed in the context of this work, which was derived from robotics and adapted to the special use of test rigs for visual fish-computer interaction. The developed fish animation system enables a simple way of design and an user-friendly animation of virtual fish stimuli. In order to realize an interaction, the position of the real fish is tracked and the fish movement is exactly reconstructed. The optical tracking system tracks the 3D position of the fish in real time and generates a simple geometric reconstruction of the experimental animals. The developed method for compensation of the light refraction occurring at the aquarium increases the position accuracy. In addition, methods for automatic initialization, simple camera calibration and for increasing system reliability with wavy water surfaces are presented. For the reconstruction of the fish movement, analysis-by-synthesis methods are used, with which the geometric properties of the real fish are transferred into model parameters of the virtual model. In this way, recorded motion sequences can be processed for direct use in the animation system. In order to increase the precision, a method for the compensation of light refraction has also been developed for the reconstruction part. The complete system has been successfully tested in several experiments with Sailfin mollies at the University of Siegen and has been partially made available as open source software to the research community.Source Type:Doctoral Thesis3 8 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Wearable-based affect recognition(2019)Advances in wearable-based sensor technology like smartphones and watches allow to monitor users in a minimally intrusive way. At the point of writing, wearables are, for instance, used to count steps or estimate burned calories. Recently, a first generation of smartwatches entered the consumer market offering data driven insights into affective states. Over the course of this thesis physiological and motion data recorded using wearables have been employed to detect the affective state (e.g., stress or amusement) of users. The contributions made are threefold: First, a comprehensive literature review of the state-of-the art in wearable-based affect recognition was conducted. Second, concluding from this review a lack of publicly available multimodal datasets was identified. This gap was closed by recording, benchmarking, and publishing a lab study dataset for WEarable Stress and Affect Detection (WESAD). Third, a field study was conducted recording physiological and motion data as well as affective labels from 11 healthy subjects. Prior to the field study guidelines for smartphone-based labelling apps were formulated and they were evaluated using the field study data. Furthermore, data and labels acquired during the field study were used to train both feature-based and latest end-to-end trainable machine learning classifiers, detecting affective states on different scales. Both types of classifiers performed on par (averaged F1 score across scales: ~ 45%). Hence, potential pitfalls for wearable-based affect recognition were discussed in detail and implications for further research were provided.Source Type:Doctoral Thesis5 14 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Execution environment for integrated real-time systems based on software-defined networking(2019)Today there exists a wide range of industrial systems that are based on federated architectures, which means that the each computing node in the system is exclusively assigned to one function. Due to the increasing computing capability of a single processor and the increasing amount of computing processors on a single platform, extensive research on integrating multiple functions with different criticality levels on a shared platform was carried out. For example, in the avionic domain, the development trend has moved from federated to integrated architectures. The ARINC 653 standard was released, which defines the execution environment for hosting several avionic software functions within a single computing node. ARINC 653 was successfully implemented (e.g., Airbus A380) and achieved its primary goals (cost and weight reduction, enabling modular certification). However, the existing execution environments based on an integrated architecture support only static system configurations. In specific domains like the railway industry, dynamic system adaptation is required during runtime, which affects both the application execution environment and the data communication mechanisms. In this dissertation, our focus is on an execution environment based on an integrated architecture, which guarantees the safe integration of mixed-criticality applications and also addresses the system reconfiguration problem. In order to close the research gap, we introduce an execution environment for integrated real-time applications by leveraging the Software-Defined Networking (SDN) paradigm. We extend the temporal and spatial isolation mechanisms from the application layer to the execution environment, so that the integrated applications share the computing node without interference. For the data communication of the integrated applications, we propose a virtual switch supporting temporal and spatial isolation between data flows and leverage the SDN paradigm to address the reconfiguration requirements of data flows. Besides, we also address the controlled import and export of messages between data flows in the proposed virtual switch. For the deterministic communication requirements of hard real-time applications, we propose a virtual switch that is IEEE 802.1Qbv and IEEE 802.1Qci capable according to the Time Sensitive Networking (TSN) standard, in order to close the research gap of virtual switching guaranteeing bounded delay with low jitter in an integrated architecture. In the proof-of-concept implementations, we demonstrate the non-interference between applications in the execution environment by fault injection. In our virtual switch demonstrators, we evaluate the fundamental isolation mechanisms and determinism of message switching, while measuring the caused overhead for message transmission as well as controlled data exchange, where the measured overhead in the proposed virtual switch is less than 10 μs.Source Type:Doctoral Thesis3 13 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Fully-automated plant recognition systems in challenging controlled and uncontrolled environments using classical and Deep Learning methods(2019)Similar to other sectors, present-day agriculture relies on new advances in different fields such as machine learning, computer vision, robotics, botany, etc. In the modern world, new scopes have been introduced to agriculture, either directly or indirectly, to meet human needs, preserve the natural and environments and resources for the future. As an example, the sustainability of growth is dependent on a drop in cost under a particular threshold, and modernization of agriculture, in different aspects, is a demand to accelerate the process toward an acceptable growth. In order to improve agricultural productivity and increase benefits, one necessity is to transition from traditional methods to modern methods and availability of smart machines. In this way, it is feasible to build systems based on automation and control concepts and utilize precise algorithms for carrying out different tasks with fewer hands-on farms and protecting natural resources for the next generations. Hence, experts in robotics and electrical engineering are also involved with new aspects of agriculture and farming. Accordingly, while researchers have been forced to compete for increasing precision and profitability in agricultural activities and improve present methods with respect to natural environments, it is also necessary to serve on new major fronts: accurate mitigation of weeds in fields, optimum water consumption, reducing labor costs and number of workers, 24-hour remote control of fields, etc. Hence, it is necessary to provide more useful information about plant species and apply the extracted information for further purposes. Accurate recognition of plants is an essential part of such information. This task cannot be neglected as it supports not only farmers but also botanists and environmentalists. By considering the workplaces of farmers and botanists, it is feasible to divide the workspaces into two main subsets: controlled environments like laboratories with static conditions and uncontrolled environments like outdoor environments with dynamic conditions. Despite the importance of plant recognition, a considerable number of works has been proposed for recognizing plant species in stationary conditions based on constant background, light condition, the position of leaves, presence of single leaves, etc. In the real world, such constraints and assumptions do not lead to promising results. Therefore, consideration of other factors is essential to build efficient systems for natural plant recognition. In this research, both workspaces have been considered to develop well-mechanized plant recognition systems. This work employs the modern combined methods for local feature extraction and precise recognition of plant species. To fulfill the goals in the controlled environment, six different plant recognition systems are developed and evaluated by conducting various experiments. It is noteworthy that the modern combined methods have been adopted as the foundation of the first phase of the natural plant recognition systems in the uncontrolled environment. However, the story changes in outdoor environments and there is no fixed condition for taking images of plants and leaves. In uncontrolled environments, environmental and non-environmental factors affect the photographing process. Light intensity and illumination are two crucial environmental factors that have an impact on images, and these factors may vary over time. Images taken from one particular scene or object are not the same if it is captured in the morning or the evening. Furthermore, weather affects the color intensity in outdoor environments as the color of leaves depends on temperature, light and water supply, and changes to these factors are also inevitable with the change of month and season. Non-environmental factors like background and distance have also effects on the performance of plant recognition systems. Backgrounds of images of natural plants taken in outdoor environments are generally more complicated in comparison to backgrounds in controlled environments. Meanwhile, the distance, either short or long, between camera and plant is undoubtedly another big challenge in uncontrolled environments. In addition, there is no certainty that the images contain only one single leaf or there will be a number of leaves within images. To develop a more efficient natural plant recognition system, we ride the wave of the tsunami of deep learning and build a novel plant recognition system based on a convolutional neural network. Due to the promising result and the superior performance, the system is then deployed as the main core of a mobile real-time system. To evaluate the system, a mobile robot and a semi-robot have been equipped with cameras to navigate and explore the outdoor environment in two different years, 2017 and 2018. While exploring, an image is captured and automatically processed by the deep natural plant recognition system to visualize the species of the target plant as a real-time system. The final results show that the real-time mobile plant recognition system can identify natural plant species independently of the used camera, distance, time of day and other environmental and non-environmental factors in uncontrolled environments.Source Type:Doctoral Thesis1 3 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Grammar-based compression for strings and trees(2019)The goal of grammar-based compression is to represent a string by a small context-free grammar that produces only this string. Such a grammar is called a straight-line program (SLP). Grammar-based compression is a powerful tool to efficiently store data and process the compressed representation without decompressing it. In the first part of this work, we study the grammar-based compressors LZ78, BiSection, Repair, Greedy and LongestMatch, which are among the most popular compressors in this area. In the seminal work "The smallest grammar problem" by Charikar et al., the authors derived lower and upper bounds on the approximation ratios of several grammar-based compressors including the algorithms mentioned above. Unfortunately, for none of the compressors the presented bounds matched. Here, we close the gaps for LZ78 and BiSection. For RePair and Greedy we improve the lower bounds. Moreover, we improve a result of Arpe and Reischuk which relates grammar-based compression for arbitrary alphabets and binary alphabets. In the second part of this work, we consider grammar-based compression for trees. The main principle is similar, because the goal is to represent a tree by a small linear context-free tree grammar that produces only this tree. Such a tree grammar is called a tree straight-line program (TSLP). As a main contribution, we present two algorithms that produce a TSLP of size O(n/log n) for any given tree with n nodes and a constant set of node labels, where we assume that the maximal number of children of a node in the tree is also bounded by a constant. Additionally, the obtained TSLP has logarithmic depth. We show that those properties can be achieved in logarithmic space, or alternatively, in linear time. Similar results on the worst-case size of SLPs are well known. We use our constructions for two applications: First, we apply TSLPs to the problem of transforming arithmetical formulas into equivalent circuits of size O((n*log m)/log n) and depth O(log n), where n is the size of the formula and m is number of different variables occurring in the formula. As a second application, we present a binary encoding of unlabeled binary trees based on grammar-based tree compression. We prove that this encoding is worst-case universal and thus asymptotically optimal for certain tree sources.Source Type:Doctoral Thesis7 6 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Efficient rendering and simulation of fluid transport and phase transitions in SPH-based fluids(2019)Particle based fluid simulation using the smoothed particle hydrodynamics (SPH) method has gained much attention in recent years. Due to its flexible discretization, inherent mass preservation and its ability to stably simulate free surface flows and complex interactions with boundaries, SPH has found its way into many fields of research. While many phenomena including the transport of substances such as salt or dye, and melting and solidification can already be described in SPH, interactions with the air phase are commonly omitted. Typically, SPH liquids are rendered only in terms of their surface. However, with the growing complexity of simulations there arises a need for complementary rendering and visualization techniques that take transport of substances into account. The goal of this work is to improve fluid animation of surface dynamics and the rendering and visualization of fluid transport. First, a simulation of evaporation and condensation of SPH based fluids is introduced. Therefore, the air phase is simulated on a coarse grid and exchanges mass with the particle based liquid phase. Condensation only takes place on surfaces of rigid objects and is realized using textures into which mass can be condensed and from which particles can be generated. In order to achieve high visual detail of condensed liquids at surfaces, an implicit surface model is developed that allows to render moving liquid droplets at sub-particle detail including dynamic contact angles. Second, an efficient adaptive volume ray casting of SPH-based scalar fields is developed. In order to achieve fast spatial access to particle data, particles are mapped to cells of a view-aligned perspective grid. By applying a sampling error analysis to the volume rendering equation inside of each grid cell, the sampling rate can locally be adjusted according to a user-defined screen space error tolerance yielding substantial speedups without sacrificing image quality. Third, this work presents a vector field visualization of advective-diffusive flows of scalar quantities. Therefore, the advective, diffusive and total flow are each decomposed into a scalar quantity and a velocity component of transport. By introducing the novel visual metaphor of stream feathers, all flow components can be simultaneously visualized allowing for an intuitive insight into complex flow scenarios.Source Type:Doctoral Thesis2 18 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Language recognition in the sliding window model(2019)In many streaming applications recent elements in the stream are considered more important than older elements. In the sliding window model we are given an unbounded stream of elements and the goal is to maintain a data structure which allows performing a certain query (e.g. computing a numerical quantity or verifying a property) on the set or sequence of the last n elements. The number n is called the window size, which can be either a fixed number or controlled online. The challenge is to devise streaming algorithms which avoid maintaining the window explicitly using Θ(n) space. This thesis considers the language recognition problem in the sliding window problem: Given a formal language (in other words, a property) and a stream of symbols, maintain a small data structure which allows testing whether the current window, i.e. the suffix of length n, belongs to the language (satisfies the property). The main question that we aim to answer is: Which languages admit sliding window algorithms using sublinear space? The first main result is a space trichotomy (constant, logarithmic, linear) for the space complexity of regular languages in the fixed- and the variable-size sliding window model, together with language-theoretic descriptions of the space classes. We also study the uniform setting where the regular language is considered as part of the input. On this basis we extend these results in various directions: (i) randomness, (ii) approximation, and (iii) subclasses of context-free languages. We prove a quatrochotomy for the randomized space complexity of regular languages. Concerning approximation, we present a constant-space sliding window property tester for every regular language, which distinguishes between words in the language and words that have large Hamming distance from the language. Finally, we give partial results on context-free languages over sliding windows and extend the space trichotomy for regular languages to the class of visibly pushdown languages.Source Type:Doctoral Thesis2 8 - Some of the metrics are blocked by yourconsent settings
Publication Access status: Open Access , Algorithmic aspects of grammar-compressed trees(2019)We introduce forest straight-line programs (FSLPs) as a compressed representation for unranked trees (forests). These are compared to other compression formalisms for forests, namely top dags and TSLPs (tree straight-line programs) for fcns (first-child next-sibling) encoding. We show that FSLPs are equally succinct to TSLPs for fcns encodings but converting them to top dags requires a blowup of the alphabet size. All of these formalisms are treated uniformly using an algebraic setting. We then use FSLPs to implement various data structures and algorithms: We can answer in polynomial time if two FSLPs produce equal forests modulo associativity and/or modulo commutativity. Allowing linear preprocessing time, we implement a data structure that allows navigation steps on FSLPs (go to the first/last child, the next/previous neighbor or to the parent, or print the symbol at the current location) to occur in constant time. Allowing polynomial time preprocessing, we extend this to include a subtree equality check that works in constant time. We also study the effects of unorderedness on compression and derive various bounds. In the best case, compression using DAGs can be improved from $n$ to $log n$ when trees are allowed to be reordered, i.e. a ratio of $n / log n$. For TSLPs/FSLPs instead of DAGs, we show a ratio of $n * log log n / log^2 n$. We show how to evaluate a form of visibly one-counter automata on FSLPs in polynomial time by implementing them as an algebra. Finally, we implement an algorithm that produces a top dag and is worst-case optimal, i.e. it always finds a top dag that compresses to at least $n/log n$, while also retaining other beneficial properties from previous top dag compressors.Source Type:Doctoral Thesis2 3
