Search

This work focuses on the Limited Information Shared Control and its controller design using potential games. Through the developed systematic controller design, the experiments demonstrate the effectiveness and superiority of this concept compared to traditional manual and non-cooperative control approaches in the application of large vehicle manipulators.

In this work, the first simulation model of oxygen depolarized cathodes (ODC), which are silver catalyst-based gas diffusion electrodes, is presented that considers the phase equilibrium of the gas-liquid interface and structure-related inhomogeneities in electrolyte distribution. By means of the model it has been identified that mass transport of water and ions in the liquid phase is a crucial factor for electrode performance and how it is influenced by the electrode structure.

During the evolvement of autonomous driving technology, obtaining reliable 3-D environmental information is an indispensable task in approaching safe driving. The operational behavior of automotive radars can be precisely evaluated in a virtual test environment by modeling its surrounding, specifically vulnerable road users (VRUs). Such a realistic model can be generated based on the radar cross section (RCS) and Doppler signatures of a VRU. Therefore, this work proposes a high-resolution RCS measurement technique to determine the relevant scattering points of different VRUs.

In this study the technical and economic feasibility of a 380 kV, 5 kA resistive type superconducting fault current limiter was investigated. Conceptual designs were developed with superconducting tapes cooled by liquid nitrogen and arranged in bifilar coils. A cryostat was designed using FEM simulations taking into account the non-linear voltage distribution. For the complete system including current limiting reactors and cooling system, the investment and operation costs were calculated.

A gyrotron is used in magnetically confined plasma experiments for heating, current drive, plasma stabilization and plasma diagnostics. This work presents the first design and construction of a multi-frequency / multi-purpose coaxial-cavity pre-prototype gyrotron operating at (136)/170/204 GHz with an output power of 2 MW. It is the first step towards operating frequencies up to 240 GHz using the coaxial-cavity gyrotron technology.

Optical measurement methods are becoming increasingly important for high-precision production of components and quality assurance. The increasing demand can be met by modern imaging systems with advanced optics, such as light field cameras. This work explores their use in the deflectometric measurement of specular surfaces. It presents improvements in phase unwrapping and calibration techniques, enabling high surface reconstruction accuracies using only a single monocular light field camera.

The approach of a laser scanning headlamp for adaptive light functions includes six laser sources which are deviated on a micro mirror. The oscillation of that mirror generates an extended light distribution on a white light converter which is then projected on the road. This work describes that approach and presents methods to evaluate the homogeneity of headlamp light distributions.

In this thesis, the use of voltage-source active filters for harmonic mitigation of grid-connected power converters is proposed. Based on a modeling of this novel converter configuration, an operating strategy is derived to accomplish this task with low control bandwith, low passive filtering effort, and high efficiency with only a small increase in installed semiconductor power.

The design of approaches for future market-based energy network operation faces the technical challenge of needing to coordinate a vast number of network participants spatially and temporally, in order to balance energy supply and demand, while achieving secure network operation. To meet this challenge, the research field of transactive control emerged. Within this work a new transactive control approach for coupled electric power and district heating networks is presented.

Reinforcement Learning is a promising tool to automate controller tuning. However, significant extensions are required for real-world applications to enable fast and robust learning. This work proposes several additions to the state of the art and proves their capability in a series of real world experiments.

Image processing combines the disciplines of cameras – image-based sensors – with the processing of the sensor data – the images. From this follows the particular attraction of this field. The conference proceedings at hand of the “Image Processing Forum”, which took place on 24.-25.11.2022 in Karlsruhe as a common event of the Karlsruhe Institute of Technology and the Fraunhofer Institute of Optronics, System Technologies and Image Exploitation, contain the articles of the contributions.

In this work, the task of wide-area indoor people detection in a network of depth sensors is examined. In particular, we investigate how the redundant and complementary multi-view information, including the temporal context, can be jointly leveraged to improve the detection performance. We recast the problem of multi-view people detection in overlapping depth images as an inverse problem and present a generative probabilistic framework to jointly exploit the temporal multi-view image evidence.

High-temperature superconductors have distinct advantages compared to conventional conductors. Below their critical temperature, superconductors have immeasurably low ohmic losses. To maintain the superconducting state, superconductors require constant cooling. This study aims at identifying the environmental impacts of the application of superconductors in future grid technologies such as superconducting power cables.

Radar is an important part of autonomous driving and allows an appropriate environment recognition. The contribution of this work is a new arrangement of capacitively coupled patches and a holistic view on substrates under different environmental conditions. Thus, an optimized design can meet reliability and bandwidth requirements.

This work presents a real-time dynamic pricing framework for future electricity markets. Deduced by first-principles analysis of physical, economic, and communication constraints within the power system, the proposed feedback control mechanism ensures both closed-loop system stability and economic efficiency at any given time. The resulting price signals are able to incentivize competitive market participants to eliminate spatio-temporal shortages in power supply quickly and purposively.

To keep up with the ever-increasing data transmission speed needs, data center interconnects are scaling up to provide multi-Tbit/s connectivity. These links require a high number of WDM channels, while the associated transceivers should be compact and energy efficient. Scaling the number of channels, however, is still limited by the lack of adequate optical sources. In this book, we investigate novel chip-scale frequency comb generators as multi-wavelength light sources for Tbit/s WDM links.

Knowledge of the cycle-frequency changes in the machine currents is of great advantage for many methods of controlling modern, highly utilized electrical machines. But how can these changes in machine currents be reliably measured during operation? Is the very short time of the individual switching states sufficient? And how can the changes in the currents be described mathematically and interpreted physically? This work provides an answer to these questions.

LiDAR (Light detection and ranging) represents one of the key technologies for autonomous driving. For this reason, this work is concerned with the question of eye safety of the laser-based technology. For this purpose, investigations on laser-tissue interactions, laser-induced damage thresholds and mechanisms by multi-pulse trains were performed as well as assessments according to the laser safety standard in Europe (DIN EN 60825) and the USA (ANSI Z.136) were applied.

Model-based methods are increasingly being used for a safe and efficient operation of battery cells. For modeling battery cells, fractional models which are characterized by non-integer derivative orders have become established due to their electrochemical interpretability. In this work, methods for parameter and derivative order identification of fractional systems are derived without restriction with respect to the excitation at the begin of identification.

In the THz frequency range, superconducting detectors coupled to planar antennas are well suited for the use in the field of accelerator beam diagnostics. This work studies the coupling efficiency and spectral sensitivity of different antenna designs. Based on the conclusions, two different detector systems for spectroscopic measurements are developed and verified with measurements at accelerator-based light sources.

In this work, discrete-time and continuous-time methods that integrate flexible reference trajectory representations into Adaptive Dynamic Programming approaches are presented and analyzed for the first time. Moreover, theoretical conditions on the system state are derived that ensure the persistent excitation property, which is crucial for the convergence of the adaptation. Real-world applications of the presented adaptive optimal trajectory tracking control methods reveal their potential.

This work presents the development and application of high-speed fluorescent thermal imaging for quench analysis in high-temperature superconductors (HTS). Using a fluorescent coating, with a temperature-dependent light emission, temperature changes can be calculated over 2D surfaces. The technique uncovered peculiar transient effects in novel HTS tape architectures and also helped to verify and better understand hot spot development in both insulated and non-insulated, HTS–wound pancake coils.

In classical approaches for the torque control of Permanent Magnet Synchronous Machines the torque references are converted into current references by static lookup tables which consider power losses. This procedure is dynamically suboptimal, interdependent and strongly machine-dependent. This work addresses the question: How can a Model Predictive Controller be designed to simultaneously optimize the objectives torque reference tracking and power loss minimization?

By providing a high resolution at high distances in a broad field of view, LiDAR sensors are a key element in automated driving cars. However, in harsh environmental conditions, particles from rain or dust attach on the outer surface of the sensor and cause a degradation of the overall performance. This work introduces several methods to detect and characterize those particles, realizing the integration of an intelligent and fully automated cleaning system to restore the required performance.

In this work, two methods are developed and evaluated to integrate and test functional chains of automotive lighting functions on simulation-based test systems. The focus is initially on an approach for the efficient use of light for the display of high contrasts in virtual night driving with a real driver assistance camera. Furthermore, a test method is developed to evaluate pixel-based lighting functions on a hardware-in-the-loop test system based on metrics. In this work, two methods to stimulate and test automotive front lighting functions on simulation-based test benches are developed and evaluated. The first approach deepens the idea to reach necessary high contrasts for virtual night driving with a real driver assistance camera by an efficient use of reflecting light. Furthermore, a test method is shown to test pixel-based lighting functions based on metrics on a hardware-in-the-loop test bench.

In recent years, OFDM has become more and more important for radar applications. A disadvantage, however, is the need for high sampling rates, which is exactly the starting point of the frequency comb OFDM radar scheme. With its help, the bandwidth in the radar channel can be enhanced without increasing the sampling rates at the same time. In this way, a high range resolution can be achieved without using fast converters with high sampling rates.

The aim of this work is the development of a Radar system for consumer applications. It is capable of tracking multiple people in a room and offers a touchless human-machine interface for purposes that range from entertainment to hygiene.

In this work, spatio-spectrally coded multispectral light fields, as taken by a light field camera with a spectrally coded microlens array, are investigated. For the reconstruction of the coded light fields, two methods, one based on the principles of compressed sensing and one deep learning approach, are developed. Using novel synthetic as well as a real-world datasets, the proposed reconstruction approaches are evaluated in detail.

In this work, the stack operation of solid oxide cells (SOCs) is analyzed by a simulation model. For this purpose, all coupled physical and chemical sub-processes are represented within a two-dimensional geometry of a single stack level. After parameterization with experimental data and extensive validation, the model is applied to three research issues in SOC technology.

Especially pedestrians rely on the interaction with other road users. The introduction of automated driving systems withdraws this communication from drivers and thus, vehicles need to interact with pedestrians. Therefore, multiple vehicle-pedestrian-communication concepts are evaluated. A driver intention prediction allows pedestrians to learn these newly introduced signals, while driver and vehicle communicate simultaneously.

This work introduces a new specification and verification approach for dynamic systems. The introduced approach is able to provide type II error free results by definition, i.e. there are no hidden faults in the verification result. The approach is based on Kaucher interval arithmetic to enclose the measurement in a bounded error sense. The developed methods are proven mathematically to provide a reliable verification for a wide class of safety critical systems.

The work deals with aspects of image and video exploitation for monitoring, surveillance and security applications in non-cooperative application situations. The term "non-cooperative" refers to scenarios in which, for example, the recording parameters (backgrounds, lighting conditions, etc.) are not or hardly known. The goal is to detect man-made modifications (artifacts) in natural environments.

The research reported in this thesis focuses on the decision making aspect of human-machine cooperation and reveals new insights from theoretical modeling to experimental evaluations: Two mathematical behavior models of two emancipated cooperation partners in a cooperative decision making process are introduced. The model-based automation designs are experimentally evaluated and thereby demonstrate their benefits compared to state-of-the-art approaches.

In the first part of this dissertation, a framework for the design of a CNN for FPGAs is presented, consisting of a preprocessing algorithm, an augmentation technique, a custom quantization scheme and a pruning step of the CNN. The combination of conventional image processing with neural networks is shown in the second part by an example from robotics, where an image-based visual servoing process is successfully conducted for a gripping process of a robot.

The proposed method for the configuration of renewably powered energy supply systems as isolated grids backed by energy storage technology and based on statistical models shows the economic viability of such a system. The energy supply system was configured by implementing cost models using an optimization software. Validation was achieved through comparison with measurements taken from two different existing energy supply areas.

The atrial substrate undergoes electrical and structural remodeling during atrial fibrillation. Detailed multiscale models were used to study the effect of structural remodeling induced at the cellular and tissue levels. Simulated electrograms were used to train a machine-learning algorithm to characterize the substrate. Also, wave propagation direction was tracked from unannotated electrograms. In conclusion, in silico experiments provide insight into electrograms' information of the substrate.

This work aimed at designing, studying and producing the first prototypes of KIDs tailored for fusion plasma polarimetric diagnostics. Diamond was considered for the first time as substrate material for low-temperature superconducting detectors given its unmatched optical, radiation hardness and thermal qualities, properties necessary for working environments potentially saturated with radiation. This work represents a first step toward the optimization and final application of this technology.

This work addresses the automated generation of physical-based models and model-based observers. We develop port-Hamiltonian methods, which for the first time allow a complete and consistent automation of these two processes for a large class of interconnected systems.

Although laser cutting of metals is a well-established process, there is considerable potential for improvement with regard to various requirements for the manufacturing industry. First, this potential is identified and then it is shown how improvements could be made using machine learning. For this purpose, a database was generated. It contains the process parameters, RGB images, 3D point clouds and various quality features of almost 4000 cut edges.

A design process for HTS DC cables was developed for high current applications. Based on the design process, a 35 kA HTS DC cable demonstrator was developed. The superconducting elements of the demonstrator were manufactured and tested individually at 77 K. Afterwards, the demonstrator cable was assembled and tested at 77 K. The assembled demonstrator successfully reached 35 kA at 77 K and self field conditions.

As the share of renewable generation increases in electric grids, the traditionally heat driven operation of combined heat and power plants (CHPs) reaches its limits. Thermal storage is required for a flexible operation of CHPs. This work proposes three novel methods to use a heating grid as thermal storage by exploiting its thermal dynamics. These include the first approach proving global optimality, a novel linear formulation of grid dynamics and an easily real world applicable approach.

This work addresses inverse dynamic games, which generalize the inverse problem of optimal control, and where the aim is to identify cost functions based on observed optimal trajectories. The identified cost functions can describe individual behavior in cooperative systems, e.g. human behavior in human-machine haptic shared control scenarios.

Major challenges in nanoparticle and nanocomposite development are the control of particle size and shape, and to achieve uniform particle dispersion. In this book, the application of optical coherence tomography (OCT) for nanocomposite and nanoparticle characterization is investigated. Industrial requirements are robustness, small system cost and size, and an open path towards parallelization. We design and investigate silicon photonic integrated OCT systems that comply with these requirements.

This report presents an update of the experimental achievements published in the review “State- of-the-Art of High-Power Gyro-Devices and Free Electron Masers”, Journal of Infrared, Millime-ter, and Terahertz Waves, 41, No. 1, pp 1-140 (2020) related to the development of gyro-devices (Tables 2-34). Emphasis is on high-power gyrotron oscillators for long-pulse or continuous wave (CW) operation and pulsed gyrotrons for any applications.

This work proposes a concept based on haptic-shared control to assist humans in taking over a task from an automation, such as at the end of highly automated driving. The description of the interaction by time-variant differential games allows a model-based controller design as well as the estimation of the human's current readiness to take over. Several experiments demonstrate more performant and safe transitions with the new approach.

Due to its dynamic range, the human eye can adapt to a wide variety of light situations within a very short time. If the dynamic range of the eye is insufficient, glare occurs. There is no suitable objective measurement method to describe the effects on visual performance and its course. This was developed and validated as part of this work.

Currently, dynamically reconfigurable systems are not used in automotive and there is no process model for their development. The focus of this dissertation is to explore methods and approaches for the development of such systems. One major architectural driver is autonomous driving, another is functional high integration on central computing platforms. Taking these into account, dynamic reconfiguration is classified and explored.

Novel THz device concepts and signal processing schemes are introduced and experimentally confirmed. Record-high data rates are achieved with a simple envelope detector at the receiver. Moreover, a THz communication system using an optoelectronic receiver and a photonic local oscillator is shown for the first time, and a new class of devices for THz transmitters and receivers is investigated which enables a monolithic co-integration of THz components with advanced silicon photonic circuits.

Ultrasound Tomography is an emerging technology for medical imaging that is quickly approaching its clinical utility. Research groups around the globe are engaged in research spanning from theory to practical applications. The International Workshop on Medical Ultrasound Tomography 2019 brought together scientists to exchange their knowledge and discuss new ideas and results in order to boost the research in Ultrasound Tomography.

This work deals with the design, fabrication and quality control of microstructured lenses for headlamp projection systems. An algorithm is described to structure smooth lens surfaces. The data transfer for manufacturing steel tools for the injection molding process is explained. A reverse engineering process used to qualify the tools is presented.