Search

Cyber-physical systems are a crucial part of modern automation applications. These systems are widespread across the production industry and critical infrastructures where a high degree of security, reliability and availability is required. This work investigates possible defense mechanisms against attacks on cyber-physical systems modeled by networked discrete event systems. Based on a threat assessment, attack prevention, attack detection and localization, and attack recovery methods are proposed. The cyber attacks under consideration are stealthy attacks that actively hide their influence and are not detectable by conventional anomaly detection schemes. The attack prevention is based on a controller encryption scheme exploiting the use of homomorphic encryption. The attack detection and localization are realized by introducing unexpected behavior into the transmitted signals and analyzing the timing behavior. The attack recovery reconfigures the controller based on the information gained from the attack localization and Monte-Carlo Tree Search.

Die intelligente Wiederverwendung von Erfahrungswissen in sicherheitskritischen IT-Projekten, das üblicherweise in natürlichsprachlicher Form vorliegt, beispielsweise in Projektberichten, stellt eine Herausforderung in der betrieblichen Praxis dar. An der Schnittstelle von Betriebswirtschaftslehre, Wirtschaftsinformatik und Kerninformatik werden für eine mögliche Lösung drei bisher getrennte Themengebiete integriert: Projektmanagement und Wissensmanagement aus betriebswirtschaftlicher Perspektive sowie Künstliche Intelligenz (KI). Zur Ermöglichung dieser intelligenten Wiederverwendung von Erfahrungswissen werden etablierte Stränge der KI-Forschung, nämlich Ontologien und Case-based Reasoning, genutzt. Obwohl diese Ansätze aus wissenschaftlicher Sicht nicht völlig neu sind, wurden sie betriebswirtschaftlich bisher nur begrenzt erforscht, insbesondere im Kontext von Projekt- und Wissensmanagement. Gleichzeitig wird'wissenschaftliches Neuland'betreten, indem Instrumente der Word2Vec-Technik für die Ähnlichkeitsermittlung zwischen alten und neuen Projekten verwendet werden. Auf diese Weise werden KI-Technologien aus der GOFAI-Sphäre ('Good Old Fashioned Artificial Intelligence') und der'moderneren'KI-Forschung zu Neuronalen Netzen verknüpft, um die effektive Wiederverwendung von Erfahrungswissen aus sicherheitskritischen IT-Projekten zu ermöglichen.

Maintaining the balance between generation and consumption is at the heart of electricity grid operation. A disruption to this balance can lead to grid overloads, outages, system damage, rising electricity costs or wasted electricity. For this reason, accurate forecasting of load behavior is crucial. In this work, two classes of ML-based algorithms were used for load forecasting: the Hidden Markov Models (HMMs) and the Deep Neural Networks (DNNs), both of which provide stable and more accurate results than the considered benchmark methods. HMMs could be successfully used as a stand-alone predictor with a training based on Maximum Likelihood Estimation (MLE) in combination with a clustering of the training data and an optimized Viterbi algorithm, which are the main differences to other HMM-related load forecasting approaches in the literature. Adaptive online training was developed for DNNs to minimize training times and create forecasting models that can be deployed faster and updated as often as necessary to account for the increasing dynamics in power grids related to the growing share of installed renewables. In addition, the flexible and powerful encoder-decoder architecture was used, which helped to minimize the forecast error compared to simpler DNN architectures such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Long Short-Term Memory networks (LSTMs) and others.

Die Ereignisse seit dem Jahr 2020, wie die weltweiten Lockdowns während der COVID-19-Pandemie und das Auflaufen der Ever Given im Suezkanal, unterzogen Supply Chains einem beispiellosen Stresstest. Es wurde deutlich, dass das Management von Supply Chains ein wettbewerbsentscheidender Faktor ist. Eine der wichtigsten, aber auch schwierigsten Herausforderungen dabei ist es, die Transparenz von Supply Chains zu erhöhen. Transparenz ist zugleich eine inhärente Eigenschaft der Blockchain-Technologie. Bisherige Studien schreiben ihr ein enormes Potenzial zu, durch ihre Unveränderlichkeit zur Verbesserung der Resilienz von Supply Chains beizutragen. In diesem Buch werden die Möglichkeiten und Grenzen der Blockchain Technologie für das Supply Chain Management aufgezeigt. Im Rahmen der Forschungsmethodik erfolgt eine umfassende Literaturaufarbeitung sowie Artefaktentwicklung, die auf einem Design Science Research-Ansatz basiert. Dabei verdeutlichen empirische Untersuchungen die gegenwärtigen Herausforderungen des Supply Chain Managements in Deutschland. Der Analytic Hierarchy Process ermöglicht anschließend ein Matching zwischen den Funktionalitäten der Blockchain Technologie und den Anforderungen des Supply Chain Managements. Auf Basis der gewonnenen Erkenntnisse erfolgt eine exemplarische Konzeptentwicklung sowie -validierung durch Programmierung eines an das Supply Chain Management angepassten Blockchain-Konzepts, welches als hybride Blockchain bezeichnet werden kann.

Die Entwicklung des Berichtswesens in Richtung Self-Service BI ermöglicht Anwendern, die Sichten auf Kennzahlen flexibel und bedarfsorientiert anzupassen. Für die präzise inhaltliche Ausgestaltung eines solchen Berichtswesens müssen die Informationsbedarfe in einer Informationsbedarfsanalyse ebenso präzise ermittelt werden. Dabei legen deduktive Vorgaben, eine einheitliche Nomenklatur und effiziente sowie abteilungsübergreifende Koordination bereits den Grundstein für die Qualität des zukünftigen Berichtswesens. Das hergeleitete Referenzmodell der Informationsbedarfsanalyse für Kennzahlen in Handelsunternehmen bietet dafür eine umfassende Unterstützung. Das Modell ist auf den Aufbau eines Berichtswesens bzgl. der Handelskernprozesse mit einer zugehörigen Self-Service-Komponente ausgerichtet und dient Controllern in diesem Bereich somit als inhaltliche und strukturelle Vorlage. Dazu werden auch die zugehörigen Aspekte der Datenmodellierung und Berichtsvisualisierung einbezogen. Die Inhalte werden theoretisch hergeleitet und praxisbezogen demonstriert, um eine wissenschaftlich fundierte und zugleich anwendungsnahe Grundlage bereitzustellen.

This project aimed to develop a compact and lightweight telemetry system that mechanically decouples measurements, enabling versatile applications across different measurement scenarios, particularly for pistons and other moving systems. Utilizing an ESP8266, a printed circuit board design was created from individually tested integrated circuits, which formed the foundation for measuring various physical variables. Given the challenging conditions inside a piston of an internal combustion engine, ensuring high robustness was paramount during the electrical engineering process to prevent failures due to high temperatures and acceleration levels. Alongside the installation process for thermocouples, software for energy-efficient measurement with customizable resolution and optional transmission via Wi-Fi or internal storing was developed. Demonstrating feasibility during full-load operation of a heavy-duty diesel engine, the telemetry system with a high-resolution surface thermocouple was integrated in one piston to establish a method for assigning crank angles, enabling measurement and transmission of surface temperatures with crank angle resolution. Further series of measurements, including the use of alternative fuels, were conducted with a third piston, expanding the findings. The outcome is an innovative telemetry measurement device capable of recording experimental data from moving components, potentially expediting market solutions by reducing development times.

Pain assessment has remained largely unchanged for decades and is currently based on self-reporting. Although there are different versions, these self-reports all have significant drawbacks. For example, they are based solely on the individual's assessment and are therefore influenced by personal experience and highly subjective, leading to uncertainty in ratings and difficulty in comparability. Thus, medicine could benefit from an automated, continuous and objective measure of pain. One solution is to use automated pain recognition in the form of machine learning. The aim is to train learning algorithms on sensory data so that they can later provide a pain rating. This thesis summarises several approaches to improve the current state of pain recognition systems based on physiological sensor data. First, a novel pain database is introduced that evaluates the use of subjective and objective pain labels in addition to wearable sensor data for the given task. Furthermore, different feature engineering and feature learning approaches are compared using a fair framework to identify the best methods. Finally, different techniques to increase the interpretability of the models are presented. The results show that classical hand-crafted features can compete with and outperform deep neural networks. Furthermore, the underlying features are easily retrieved from electrodermal activity for automated pain recognition, where pain is often associated with an increase in skin conductance.

Dieser Band beschreibt die Untersuchung des Heizpotenzials und Emissionsverhaltens von Ladungswechselkonzepten für die Anwendung in Dieselaggregaten. Dafür wurde eine Entwicklungsmethodik erarbeitet, die es erlaubt, Ladungswechselkonfigurationen für den Anwendungsfall Heizen numerisch gegenüberzustellen und zu analysieren. Weiterhin diente die Entwicklungsmethodik zur Auslegung von Nockenwellen mit einem Zusatzhub im Auslass für den Versuch. Die experimentelle Untersuchung belegt, dass neben Ladungswechselkonfigurationen mit einem Zusatzhub im Auslass auch ein Auslass-Phasen eine besondere Heizwirkung aufweist. Dabei erwies sich unter den untersuchten Konfigurationen mit einem Zusatzhub im Auslass eine kombinierte AGR-Strategie, bestehend aus externer AGR mit einem überwiegenden Anteil internem Restgas, hinsichtlich des Heizpotenzials und Emissionsverhaltens als äußerst wirksam. Die gewonnenen Erkenntnisse der Arbeit tragen zu einem besseren Verständnis etablierter Ladungswechselkonzepte, ihrem Einfluss auf das Heizpotenzial und die Thermodynamik moderner Dieselmotoren bei.

Seit Jahrzehnten wird versucht, den Einsatz von Quellcodes zu reduzieren. Vor zirka 10 Jahren wurde hierfür der Low-Code-Begriff geprägt. Im Zusammenhang mit den Erfordernissen einer allgegenwärtigen Digitalisierung versprechen Low-Code basierte Entwicklungen eine agilere-, kompositorische-, visuell orientierte- und vor allem fachgetriebene Softwareentwicklung. Auch der Bedarf an klassischen Softwareentwicklern sollte damit reduziert werden können. In der vorliegenden Monografie setzen sich Andreas Schmietendorf und Michael Knuth mit existierenden Arbeiten im Diskurs eines Low-Code orientierten Software Engineerings, mit der Auswahl zur Entwicklung benötigter Low-Code-Platt -formen, aber auch mit den funktionalen Eigenschaften exemplarisch analysierter Low-Code Produkte auseinander. Darüber hinaus werden die aktuellen Möglichkeiten generativer KI-Ansätze im Diskurs von Low-Code basierten Softwareentwicklungen aufgezeigt.

Sleep is a cyclic physiological phenomenon, an important aspect of human life activity, which, like sport and diet, is a nutritional element that ensures the growth and development of the organism. Under the influence of various factors such as work and study stress and metabolic disorders, more and more people suffer from various types of sleep disorders. Sleep has become an important research topic in recent years. Sleep stage analysis plays an important role in the early detection and treatment of sleep disorders. However, different age groups show different symptoms of sleep disorders, and different sleep disorders show variability in their different sleep stages. The prevalence of sleep disorders is much higher in children than in adults. Although the classification of sleep stages in adults has been well studied, children show markedly different characteristics of sleep stages. Therefore, there is an urgent need for sleep stage classification in children. With the rapid development of intelligent computing technology, artificial intelligence has found wide application in medical research and health sciences in recent years. In the field of sleep medicine, deep learning approaches can efficiently and automatically learn abstracted relevant sleep features from collected sleep data to accurately interpret children's sleep stages accordingly. Compared to traditional sleep data analysis, this saves many manual and time resources for data annotation and helps sleep experts reduce the risk of misdiagnosing sleep disorders based on their prior knowledge. In this context, this book presents several advanced deep learning-based approaches for sleep stage classification in children using time series polysomnography recordings acquired from clinical sensor devices. Significantly improved performance in classifying sleep stages in children suffering from sleep disorders demonstrates the great potential of joint research and development between artificial intelligence and the field of sleep medicine.

The average age of people has increased due to advances in health sciences, which has led to an increase in the elderly population. This is positive news, but it also raises questions about the quality of independent living for older people. Clinicians use Activities of Daily Living (ADLs) to assess older people's ability to live independently. In recent years, portable computing devices have become more present in our daily lives. Therefore, a software system that can detect ADLs based on sensor data collected from wearable devices is beneficial for detecting health problems and supporting health care. In this context, this book presents several machine learning-based approaches for human activity recognition (HAR) using time-series data collected by wearable sensors in the home environment. In the first part of the book, machine learning-based approaches for atomic activity recognition are presented, which are relatively simple and short-term activities. In the second part, the algorithms for detecting long-term and complex ADLs are presented. In this part, a two-stage recognition framework is also presented, as well as an online recognition system for continuous monitoring of HAR. In the third and final part, a novel approach is proposed that not only solves the problem of data scarcity but also improves the performance of HAR by implementing multitask learning-based methods. The proposed approach simultaneously trains the models of short- and long-term activities, regardless of their temporal scale. The results show that the proposed approach improves classification performance compared to single-task learning.

Gait analysis is valuable in medical research and diagnosis, by delivering information that helps in choosing methods of intervention and rehabilitation that are beneficial for a patient. In gait laboratories, cameras or IMUs are often used to gather gait patterns. This thesis explores the possibility of using sensors below the floor as a gait data source. These sensors measure changes in the electrical capacitance to recognise steps. The construction is designed for indoor environments and is hidden under common flooring layer types. Therefore, it is very robust and suitable for practical use in daily clinical routine. A formal framework was developed to represent the measurements, considering the special characteristics of this floor sensor. The data were then used as input for artificial neural networks that were applied on classification and regression tasks. In a feature construction and extraction approach, the spatial spread of footfalls was derived and used with a feed-forward neural network. Then, in a feature learning approach, the time series data was transformed into a local receptive field, and used with a recurrent neural network. Three studies were conducted for the goals to distinguish between people with low and high risk of falling, to estimate age, and to recognise walking challenges as an external gait intervention. The combination of a robust and hidden floor sensor and machine learning opens up the prospect of future applications in health and care.

Supply chain security encompasses measures preventing theft, smuggling, and sabotage through heightened awareness, enhanced visibility, and increased transparency. This necessitates the adoption of a security-by-design paradigm to achieve effective and efficient security measures, yielding additional benefits such as diminished supply chain costs. Given their vulnerability, transportation and logistics service providers play a pivotal role in supply chain security. This thesis leverages systems security engineering and security-by-design to provide a methodology for designing and evaluating security measures for physical transport goods. It formulates nine principles that define security-by-design and establishes a supply chain security framework. An adaptation of the TOGAF architecture development facilitates the creation of secure-by-design enterprise architectures. Security measures are documented using security-enhanced processes based on BPMN. This enables an analysis and compliance assessment to ascertain the alignment of security with business objectives and the adequate implementation of requirements. The culmination of these efforts is exemplified through a case study.

Künstliche Intelligenz hat als Zukunftstechnologie eine enorme Bedeutung für Wirtschaft und Gesellschaft. Durch die Weiterentwicklung von selbstlernender Software stellen sich zahlreiche Fragen nach der Haftung für die durch diese Systeme (mit-)verursachten Schäden. Haftungsrechtlich wird aufgrund der Selbstständigkeit und Autonomie der Systeme eine Verantwortungslücke befürchtet. Der Autor untersucht vor diesem Hintergrund die technischen Grundlagen selbstlernender Software und analysiert die haftungsrelevanten Aspekte dieser Technologien sowohl rechtssystematisch als auch unter Anwendung des geltenden Rechts. Als Lösungskonzept für im Zusammenhang mit selbstlernender Software bestehende Haftungsprobleme wird ein Entwurf einer gesamteuropäischen Gefährdungshaftungsnorm im Kontext von Geschichte und Rechtssystematik vorgestellt. Die Arbeit leistet damit einen Beitrag zur gegenwärtigen Diskussion um die Regulierung künstlicher Intelligenz.

In our increasingly interconnected world, personalized and technology-assisted healthcare has become a rising trend. In an ageing society, technology enables new ways to care for and assist the elderly. Falls pose a major risk and cause of injuries for senior citizens. Technology-backed fall prevention thus has the potential to avoid severe injuries and further loss of independence, but requires continuous monitoring of the body posture in order to identify imminent falls. This work presents a system concept for a wearable wireless body area network (WBAN) for posture monitoring. It shows the principal feasibility of posture recognition from ultra-wideband (UWB) signals based on a large and diverse set of measurements. For a promising classifier-feature-combination, this work demonstrates how reliable posture recognition can be achieved with a limited number of body-mounted nodes, and analyzes its robustness towards potential pitfalls. It concludes with a proposal for a system implementation, and outlines its integration with existing and future aspects of personalized healthcare.

Logistics and supply chain management are considered as the backbone of humanitarian operations, significantly influencing their performance. Consequently, it is not surprising that the academic community addressed performance measurement in humanitarian logistics early on and extensively. However, there exists a significant disparity between the academic findings on performance measurement and their practical implementation within humanitarian organizations. What factors contribute to this gap, and how can we bridge it? This book aims to provide a socio-technical solution through an action researchbased approach that designs and develops a process-driven and IT-supported performance management system for humanitarian logistics. By utilizing an iterative and participatory design methodology, an active involvement of humanitarian organizations in identifying and addressing their practitioner realities, needs, and objectives is ensured. The resulting performance management system has been designed, implemented, and evaluated within three distinct and representative humanitarian organizations. As a result, these research findings hold promise for enhancing the capabilities of humanitarian organizations to measure and manage logistics performance effectively.

Orphan diseases pose diagnostic challenges due to complex pathologies, limited epidemiological data, and clinical experience. The development of artificial intelligence and machine learning methods has the potential to enhance the accuracy of decision support systems, improving diagnosis outcomes for rare disease patients. This research aims to create a repository for characterizing rare diseases by collecting past experiences of diagnosed patients, reducing gaps in symptom interpretation. This interdisciplinary study, in collaboration with medical experts, has resulted in a computer-aided diagnostic support system utilizing statistical analysis and machine learning algorithms. The system incorporates disease profile aggregation, pattern recognition, and information comparison. An interactive data visualization platform has been established to promote intuitive understanding and evaluate system diagnosis with graphics-based disease feature comparison. It supports medical practitioners during the diagnostic process by presenting visually appealing information. The patient-oriented inquiry mechanism efficiently reduces unnecessary questions while providing a reliable diagnosis based on probability. By combining statistical learning with the visualization module, the system can discover disease-related symptom patterns, offering new means for diagnosing rare disorders. The supplementary diagnosis prediction mechanism can be applied effectively to analyze different groups in surveyswith closed-ended questions.

World-wide trends such as globalization, demographic shifts, increased customer demands, and shorter product lifecycles present a significant challenge to the road freight transport industry: meeting the growing road freight transport demand economically while striving for sustainability. Artificial intelligence, particularly machine learning, is expected to empower transport planners to incorporate more information and react quicker to the fast-changing decision environment. Hence, using machine learning can lead to more efficient and effective transport planning. However, despite the promising prospects of machine learning in road freight transport planning, both academia and industry struggle to identify and implement suitable use cases to gain a competitive edge. In her dissertation, Sandra Lechtenberg explores how machine learning can enhance decision-making in operational and real-time road freight transport planning. She outlines an implementation guideline, which involves identifying decision tasks in planning processes, assessing their suitability for machine learning, and proposing steps to follow when implementing respective algorithms.

In recent years, researchers have achieved great success in guaranteeing safety in human-robot interaction, yielding a new generation of robots that can work with humans in close proximity, known as collaborative robots (cobots). However, due to the lack of ability to understand and coordinate with their human partners, the ``co''in most cobots still refers to ``coexistence''rather than ``collaboration''. This thesis aims to develop an adaptive learning and control framework with a novel physical and data-driven approach towards a real collaborative robot. The first part focuses on online human motion prediction. A comprehensive study on various motion prediction techniques is presented, including their scope of application, accuracy in different time scales, and implementation complexity. Based on this study, a hybrid approach that combines physically well-understood models with data-driven learning techniques is proposed and validated through a motion data set. The second part addresses interaction control in human-robot collaboration. An adaptive impedance control scheme with human reference estimation is presented. Reinforcement learning is used to find optimal control parameters to minimize a task-orient cost function without fully knowing the system dynamic. The proposed framework is experimentally validated through two benchmark applications for human-robot collaboration: object handover and cooperative object handling. Results show that the robot can provide reliable online human motion prediction, react early to human motion variation, make proactive contributions to physical collaborations, and behave compliantly in response to human forces.

Utilizing magnetic induction for wireless communication, wireless powering, passive relaying, and localization could enable massive wireless sensor applications with tiny nodes in challenging media, foremost biomedical in-body sensor networks. This work investigates the performance limits of these unique wireless systems with hardly any assumptions. As a foundation, a general system model and an interface to communication theory are developed. A major part of this work identifies two crucial magneto-inductive fading channels: that between randomly oriented coils and that caused by a nearby swarm of resonant passive relay coils. The analysis yields important technological implications. Based thereon, an investigation of wirelessly-powered in-body sensors is conducted, revealing their active and passive data transmission capabilities. Finally, a treatise of magneto-inductive node localization develops algorithms that perform near identified accuracy limits in theory and practice.

Due to their relevance in systems analysis and controller design, this thesis considers the problem of determining input-output properties of linear time-invariant systems. While obtaining a suitable mathematical model describing the input-output behavior of a dynamical system can be a difficult task, data of the system in form of input-output trajectories is often and increasingly available. This thesis therefore introduces three complementary data-driven analysis methods to determine input-output properties directly from data without deriving a mathematical model first. In particular, the results of this thesis include iterative methods, where data is actively sampled by performing experiments on the unknown system, as well as approaches based on available (offline) data. All these approaches are simple to apply, come with low requirements on the data, and provide rigorous theoretical guarantees. Systems analysis not only provides insights into the system and allows to do controller design with guaranteed stability, but it can also validate a given controller or its closed-loop performance. By developing different methods to determine input-output properties directly from data on the basis of a rigorous mathematical analysis, this thesis contributes to a sound mathematical framework for data-driven systems analysis and control theory.

Big data thrives on extracting knowledge from a large number of data sets. But how is an application possible when a single data set is several gigabytes in size? The innovative data compression techniques from the field of machine learning and modeling using Bayesian networks, which have been theoretically developed and practically implemented here, can reduce these huge amounts of data to a manageable size. By eliminating redundancies in location, time, and between simulation results, data reductions to less than 1% of the original size are possible. The developed method represents a promising approach whose use goes far beyond the application example of crash test simulations chosen here.

The solution of the generalized eigenvalue problem is one of the computationally most challenging operations in the field of numerical linear algebra. A well known algorithm for this purpose is the QZ algorithm. Although it has been improved for decades and is available in many software packages by now, its performance is unsatisfying for medium and large scale problems on current computer architectures. In this thesis, a replacement for the QZ algorithm is developed. The design of the new spectral divide and conquer algorithms is oriented towards the capabilities of current computer architectures, including the support for accelerator devices. The thesis describes the co-design of the underlying mathematical ideas and the hardware aspects. Closely connected with the generalized eigenvalue value problem, the solution of Sylvester-like matrix equations is the concern of the second part of this work. Following the co-design approach, introduced in the first part of this thesis, a flexible framework covering (generalized) Sylvester, Lyapunov, and Stein equations is developed. The combination of the new algorithms for the generalized eigenvalue problem and the Sylvester-like equation solves problems within an hour, whose solution took several days incorporating the QZ and the Bartels-Stewart algorithm.

Dieses Buch bietet einen Überblick über Forschungsdatenmanagement und dessen konkrete Umsetzung in der Informatik. Anhand von Personas und Szenarien wird ein Großteil von informatischen Anwendungsfällen und Fragen abgedeckt, um Lehrende sowie Studierende der Informatik dabei zu unterstützen, Forschungsdatenmanagement adäquat zu realisieren. Im ersten Teil des Buchs wird Lehrenden der Informatik anhand von Modulen konkret aufgezeigt, welche Themen der Informatik besonders geeignet sind, um Forschungsdatenmanagement in das Informatikstudium zu integrieren. Der zweite Teil des Buchs erläutert Bestandteile des Forschungsdatenmanagements und veranschaulicht deren Anwendung auf ausgewählte Szenarien. Abschließend werden im dritten Teil Lehrmaterialien (Arbeitsblätter, Musterlösungen, Checklisten und weitere) zur Verfügung gestellt, um den direkten und fachgerechten Einsatz in der universitären Lehre zu stärken.

Der hier vorliegende Band 12 der Schriftenreihe'Studien zu Infrastruktur und Ressourcenmanagement'gibt zusammenfassend die wesentlichen Ergebnisse wieder, die im Rahmen des interdisziplinären Verbundprojektes'Smart Utilities and Sustainable Infrastructure Change'erzielt werden konnten. Im Zentrum dieses Projektes stand die Entwicklung von innovativen agenten-basierten Computermodellen zur Beratung von kommunalen Ver- und Entsorgungsunternehmen im Kontext der Dekarbonisierung der Energieversorgung, des Klimawandels und der mit demographischen Veränderungen verbundenen infrastrukturbezogenen Herausforderungen.

Digitalisation, digital networks, and artificial intelligence are fundamentally changing our lives! We must understand the various developments and assess how they interact and how they affect our regular, analogue lives. What are the consequences of such changes for me personally and for our society? Digital networks and artificial intelligence are seminal innovations that are going to permeate all areas of society and trigger a comprehensive, disruptive structural change that will evoke numerous new advances in research and development in the coming years. Even though there are numerous books on this subject matter, most of them cover only specific aspects of the profound and multifaceted effects of the digital transformation. An overarching assessment is missing. In 2016, the Federation of German Scientists (VDW) has founded a study group to assess the technological impacts of digitalisation holistically. Now we present this compendium to you. We address the interrelations and feedbacks of digital innovation on policy, law, economics, science, and society from various scientific perspectives. Please consider this book as an invitation to contemplate with other people and with us, what kind of world we want to live in!

This thesis concerns the cooperative control of networked vehicles. Autonomous driving is a topic that is currently being discussed with great interest from researchers, vehicle man -ufacturers and the corresponding media. Future autonomous vehicles should bring the passengers to their desired destination while improving both safety and efficiency compared to current human-driven vehicles. The inherent problem of all vehicle coordination tasks is to guarantee collision avoidance in every situation. To this end, autonomous vehicles have to share information with each other in order to perform traffic manoeuvres that require the cooperation of multiple vehicles. The fundamental problem of vehicle platooning is studied extensively which describes the task of arranging a set of vehicles so that they drive with a common velocity and a prescribed distance. Local design objectives are derived that have to be satisfied by the vehicle controllers. In particular, it is shown that the vehicles have to be externally positive to achieve collision avoidance. As an abstraction from real traffic scenarios, swarms of networked vehicles are considered. The main difference be -tween swarming and traffic problems is that a communication structure that has been appropriate in the beginning might become unsuited for the control task due to the relative movement of the vehicles. To solve this problem, this thesis proposes to use the Delaunay triangulation as a switching communication structure.

In August 2022, researchers and developers from Armenia, Chile, Germany, and Japan met at the American University of Armenia for the third edition of the CODASSCA Workshop on Collaborative Technologies and Data Science in Smart City Applications, co-organized with a Summer School on Artificial Neural Networks and Deep Learning. This book presents their contributions on intelligent technologies in data science and human-centered computing.

While modern project management systems support teams during planning and development activities, primarily through performance-related process information, the equally relevant human factors are often insufficiently considered for explaining team dynamics (e.g., the affect of moods in teams). However, understanding team behavioral patterns are crucial for the accurate planning and steady execution of development tasks throughout an ongoing project. A computer-aided feedback concept is described, unifying interdisciplinary foundations and methods from the software engineering, data science, organizational, and social psychology fields for disclosing team dynamics in agile software projects. The concept covers the systematic capture of sociotechnical data combined with descriptive, predictive, and exploratory model-based methods that support understanding behavioural changes during the development process. Design science from information systems research is used in academic and industrial case studies to conceptualize and operationalize the feedback methods into a practical Jira plugin. A concluding evaluation through an action research method in two industrial software projects results in quantitative and qualitative findings regarding the feedback utilization and utility during agile development processes (e.g., team communication changes related to accomplished performances). The case studies underscore the practical relevance for systematic feedback and the need to better understand human factors in software projects.

The interior sound of vehicles is a major criterion when buying a new or used vehicle. For both pure-electric and hybrid vehicles, separately audible tonal components severely influence the pleasantness of the interior sound. Previous studies revealed that for sounds of vehicles with electrified drives, besides the tire-road and wind noise components, mainly higher-frequency tonal components influence the pleasantness. Within the present work, the audibility of those components has been determined by a modified version of an auditory masking model. For those vehicles, disturbing sound components are usually prominent during transient driving conditions. Therefore, the temporal changes of psychoacoustic parameters should be considered for the pleasantness prediction. A long short-term memory neural network depicts the relationship between time series of psychoacoustic parameters and a single value of pleasantness, which has been acquired by conducting auditory experiments with both original and augmented electric and hybrid vehicle interior sounds. The general scope of the dissertation is to develop a model of pleasantness for interior sounds of vehicles with electrified drives using a long shortterm memory model approach. The predictions form the basis for constructive countermeasures or active sound design concepts to improve the pleasantness of the interior sound.

Since the manipulation tasks for robotic systems become more and more complicated, multi-robot cooperation has been attracting much attention recently. Furthermore, under the trend of human-robot co-existence, collision-free motion control is now also desired on multi-robot groups. This dissertation aims to design a novel distributed optimal control framework to deal with multi-robot cooperative manipulation of rigid objects in dynamic environments. Besides object transportation, the control scheme also tackles obstacle avoidance, joint-space performance optimisation and internal force suppression. The proposed control framework has a two-layer structure, with a distributed optimisation algorithm in the kinematic layer for generating proper joint configuration references, followed by a robot motion controller in the dynamic control layer to fulfil the reference. An indirect and a direct distributed optimisation method are developed for the kinematic layer, both of which are computationally and communicationally efficient. In the dynamic control layer, impedance control is employed for safe physical interaction. As another highlight, abundant experiments carried out on a multi-arm test bench have demonstrated the effectiveness of the presented control schemes under various environmental and task settings. The recorded computation time shows the applicability of the control framework in practice.

This book addresses the automatic analysis of speech disorders resulting from a clinical condition (Parkinson's disease and hearing loss) or the natural aging process. For Parkinson's disease, the progression of speech symptoms is evaluated by considering speech recordings captured in the short-term (4 months) and long-term (5 years). Machine learning methods are used to perform three tasks: (1) automatic classification of patients vs. healthy speakers. (2) regression analysis to predict the dysarthria level and neurological state. (3) speaker embeddings to analyze the progression of the speech symptoms over time. For hearing loss, automatic acoustic analysis is performed to evaluate whether the duration and onset of deafness (before or after speech acquisition) influence the speech production of cochlear implant users. Additionally, articulation, prosody, and phonemic analyses show that cochlear implant users present altered speech production even after hearing rehabilitation

This book deals with applying machine learning and mathematical optimization methods for data-driven decision-making. It contributes to research on building machine learning models that capture human behavior and human preferences and that can be used to improve operations and optimization processes. In the field of churn prevention, churn prediction models have primarily been trained on one time slice of data. This work evaluates an approach to train models on data from multiple time slices and identifies two effects that contribute to an improvement in predictive performance: an increase in sample size as well as training on samples from different time slices. The multi-slicing approach makes models more generalizable under changing conditions and allows for predictions that are more accurate. In a field experiment with B2B customers of a convenience wholesaler, this thesis demonstrates how customer churn can be decreased by basing targeting decisions and retention efforts on predicted churn probabilities. In the field of crew scheduling, this work shows how benefits from machine learning and optimization can be combined to deliver better solutions for complex planning problems. For a railway freight carrier, feedback from planners regarding crew schedules is used to train a machine learning model. This book introduces an approach to integrate predicted planner feedback into the optimization process for improving the expected planner acceptance of solutions.

In the age of digitalization and the fourth industrial revolution, predictive maintenance is becoming increasingly important as a proactive maintenance type. Despite the economic benefits that predictive maintenance generates for companies, its practical application is still in its early stages. This is often due to two prevailing challenges. First, there is a deficiency of knowledge about predictive maintenance and its concrete realization. Second, there is a lack of high quality and rich data of historical machine failures. To increase the representativeness of data, data from several similar machines (i.,e. a fleet) should be considered. To foster the effective implementation of predictive maintenance, supportive guidance in the realization of a predictive maintenance project is needed. For this reason, this dissertation presents a process reference model and a development method for fleet prognostics. The process reference model describes a comprehensive and application-independent view of the complete predictive maintenance process. The model is supplemented by the fleet prognostic development method. To address the specific characteristics of the fleet, a systematic process is depicted which provides a means to assess the heterogeneity of the fleet from a data-driven perspective and simplifies the design of an algorithm considering fleet data. Finally, the applicability and value of the research results are demonstrated with three industrial cases