In software engineering research it is common to analyze existing projects to develop new concepts or to validate assumptions about software development. In the SmartSHARK project we provide a database about open source development for this purpose. In our database you can find data about source code history (commits, software metrics, bugs), issue trackers (Jira, Bugzilla, Github Issues), pull requests and code reviews, mailing lists, and CI systems (Travis CI, GitHub Actions). For more information, please visit the project homepage(https://smartshark.github.io).
Contact: Email: Prof. Dr. Steffen Herbold
The comparability and reproducibility of research results of empirical work in software engineering is a largely open problem. In this project, we establish a solid foundation for comparable and externally valid research on software fault prediction. Our approach is based on three pillars. The first pillar is data quality. Previous studies on data quality have not considered the essential problem of misclassified data. However, this introduces significant noise in the data, which affects not only the fault prediction models themselves, but also their quality assessment. The second pillar is replications.
Because of the inconsistencies highlighted by previous replications, we believe that broader replications of the state of the art are needed. Large portions of the state of the art have never been replicated or systematically compared with other approaches.
The third pillar is guidelines for research on software fault prediction. If future research does not avoid anti-patterns that negatively impact the validity of results, our work will only help in the short term to avert a replication crisis.
Contact: Email: Prof. Dr. Steffen Herbold
Internet marketplaces such as Amazon or eBay are dominating more and more areas of the retail trade as "digital checkpoints". One exception so far has been the trade with durable, highly varied and customizable consumer goods. The effect of the configurable product on the consumer's own four walls can only be checked after delivery. The ARBAY project brings the consultation and configuration of individualized goods into the living room. Virtual and augmented reality technologies are used in the project to create new digital distribution channels for highly varied goods. The aim is to develop a sales platform that extends the principle of known sales platforms for the sale of these goods. The main focus of the ISSE project is the development of a semantic product model as a basis for the platform.
E-Mail: Dr. Christoph Knieke
The aim of the project BioTope is to develop a basic technology and engineered methodology to facilitate the creation of emergent services that allow for self-adaptive system platforms. The rules of composition are not centrally and statically predetermined by the
Platform, but can be dynamically configured and demand-driven. This should be achieved and maintained in an Open IoT ecosystem that Integrates data and services to create processes to satisfy user requirements. The main motivation here is user requirements, which drives any new behaviour created in the system. The ecosystem tends to create the balance between user needs and provided data and services. All this is governed by system guarantees that ensures the correct flow of the ecosystem.
E-Mail: Eric Douglas Nyakam Chiadjeu
E-Mail: Christian Schindler
Software ecosystems are complex system groups of interacting, distributed individual systems that require continuous, autonomous optimization. In DevOpt we understand an emergent, distributed system as a three layers architecture: Local IoT ecosystems negotiate their work configuration / resource usage under framework conditions. A control layer can correct local decisions through comprehensive, distributed optimization. A DevOps layer enables analysis, maintenance and further development through human intervention.
DevOpt aims at the development of controlled emergent systems through distributed modelling, local negotiation of device configurations / resource usage, increased development efficiency through model-based design, as well as combination of emergence and DevOps. The demonstration / evaluation takes place as electric grid scenarios using self-learning predictive SmartMeters. The ISSE will implement a component-based emergence framework for the local layer and environments, which enables local optimization and mutual dynamic use of resources. Furthermore, the emergent integration should be monitored and optimized with regard to functional and non-functional requirements.
The project "Timing Analysis and Steering Development" researches methods for estimating and safeguarding runtime behavior in embedded control systems with hard real-time requirements. This is done in the context of actual development projects. The focus is on the specification, modeling and measurement of runtime behavior at system level.
This project is about improvement and maintenance of V-Modell XT Bund. The V-Modell XT Bund is a company-specific adaptation of the V-Modell XT to the federal authorities. In this project it is developed, maintained and maintained according to V-Modell XT.
E-Mail: Dr. Christoph Knieke
Weit e. V. has commissioned the Institute for Software and Systems Engineering to provide the V-Modell XT in the form of a public REST interface. After the fundamental revision of the V-Modell XT metamodel, an interface will be offered that is specified with OpenAPI and implemented in Java.
First software developments of Weit e. V. implementing the V-Modell XT REST API are the newly developed Web Assistant (formerly Project Assistant) and the Digital Projects App(https://dipa.online). The revision of this software ecosystem is intended to offer a simple and contemporary entry into project management with V-Modell XT.
In this context, there is research interest in the software life cycle as well as the software architecture with the background that there are already several editions in the form of metamodels and derivations for different user groups of the V-Modell.
E-Mail: Dr. Christoph Knieke
Environmental perception systems of autonomous vehicles are nowdays using extensive AI-based algorithms. Established techniques and methods to proof correctness regarding safety are reaching its limits. Even if a lot of driving scenarios and millions of kilometers are used for testing, an overall safety assurance during design time is not possible. The goal for the project SafeWahr is to duly detect violations of safety critical specifications and uncertainities of AI-based environmental perception systems of autonomous vehicles. In case a violation was detected the autonomous vehicle will then continue its driving task with restricted functionality in a so called fail-operational mode.
One approach for handling situations, which are not known during design time is to partially shift the safety assurance to the operating time. Ultimately, some kind of "operation time certification" is aimed. For this purpose an operation time monitoring architecture for self-diagnosis will be developed in SafeWahr. Within this operation time monitoring architecture three types of monitors will be included: (1) A Situation Monitor, which determines if the current situation was considered during design time, (2) a Validity Monitor, which determines if the AI-based perception is safe regarding its results and (3) a Function Monitor, which determines if the target function acts corretly regarding safety specifications.
E-Mail: Adina Aniculaesei
The autoMoVe project focuses on highly modular vehicle concepts based on a universally usable basic vehicle module with interchangeable vehicle superstructures, e.g. for passenger or goods transport. Taking into account the requirements of the selected application scenarios, the vehicle concepts are developed with a focus on vehicle design, vehicle functions, energy management and control as well as software architectures.
Furthermore, a development and simulation platform will be implemented to support the virtual development work. Here the functions of the developed vehicle concepts are demonstrated. In addition, individual innovative subsystems are physically implemented and tested.
E-Mail: Iqra Aslam
The Mobility Laboratory is a cross-sectional internal project that aims to connect the current dynamically adaptive software platform with safeguarding mechanisms. The platform to be developed should be used in as many projects as possible. Furthermore, the mobility laboratory is a place where mainly students come together to work on topics of autonomous driving and machine learning. The laboratory has two RaspberryPi vehicles, a 1:8 model vehicle and an indoor positioning system.
E-Mail: Adina Aniculaesei
The mobil-e-Hub project aims to meet the challenges of increasing logistics traffic on the last mile to the customer, especially due to e-commerce. The technological focus and central project innovation of mobil-e-Hub is a new logistics system that can tie drones with transport boxes via carrier systems to (electric) vehicles - the mobile e-hubs - for passenger mobility, e.g. public transport buses. The drones themselves autonomously pick up the boxes optimized for food transport at automated picking stations, touch down on the vehicles equipped for this purpose and take off directly at the delivery point to autonomously hand over the box to the customer.
There are legal and technological challenges in drone operation as a delivery service. ISSE is addressing the technological challenges. To enable reliable, robust and safe operation of the delivery drone system, an online monitoring system for the drones is being developed and implemented. For this purpose, the dependability cage approach developed at ISSE for runtime monitoring of functional requirements of autonomous vehicles is adapted to flight systems. In addition, for optimal control planning of an e-mobility system, energy management is crucial, therefore artificial intelligence methods are used to predict the energy demand. Challenging is the coupled energy management between the electric carrier vehicle and the delivery drone considering the logistics system requirements (e.g., time), the path information, the current local conditions (e.g., temperature), and the previous observations. The information from the energy management can be linked to the online monitoring system to ensure that sufficient energy is available to reach the target and make a planned landing.
Contact: E-Mail: Adina Aniculaesei
The project of “Zukunftslabor Mobilität” (in English: Future Mobility Lab) focuses on the application-oriented research work by using different mobility-related technologies. In four well-defined interdisciplinary fields (Collaborative Research Fields (CRF 1-4)), potential environmentally and socially compatible mobility solutions such as connection of systems, human users and infrastructures relying on the digitalization technology will be investigated and developed, with consideration of concrete use cases about future mobility in Lower Saxony of Germany as well as related methods and approaches for the development of innovative mobility solutions.
In the scope of this project, TU Clausthal is involved together with other project partners in CRF 2, which focuses on relevant topics of smart mobility data handling. As known, the data plays a significantly important role as the basis of autonomous driving (CRF 1) and additionally as the precondition of the development and implementation of appropriate mobility services and corresponding business models (CRF 4). In CRF2, scientists from the research fields of communication technology, information and software engineering work closely together to develop methods to realize a safe and reliable data acquisition, evaluation and fusion of mobility data. For this purpose, the approaches for reliable (safety, security and privacy) data collection and processing, which fulfill the legal framework like DSGVO, will be developed and demonstrated with consideration of both phases for system development and operation. The approaches for the development of a reliable data architecture, methods and approaches for the standardization of data handling and fusion, as well as the assessment and assurance of data quality will be also developed and tested in the project.
Contact: e-mail: Abhishek Buragohain
Elements of safety-relevant solutions, such as fire doors, differ very significantly from machines in industrial plants in terms of their costs and connection to the energy supply, a variety of challenges need to be tackeld to turn safety-relevant systems into Cyber Physical Systems. The project deals with the necessary developments. These range from the design of specific sensors and actuators, which can also be produced cost-effectively in small quantities, to IT solutions for condition monitoring and automated functional testing of safety-relevant systems, and the establishment of methodological competencies for the development process of such Cyber Physical Systems.
Contact: Stefan Wittek
Der Einsatz von KI soll dazu dienen, die spezifischen Kosten der Ölförderung zu senken, ohne Kompromisse in der Sicherheit, des Umweltschutzes oder der Integrität einzugehen. Die Studie soll vor diesem Hintergrund aufzeigen, welche Einsatzbereiche für KI in einem maturen Ölfeld bestehen und wie diese wirtschaftlich zu bewerten sind. Die Einsatzmöglichkeiten zur Mehrwertbestimmung künstlicher Intelligenz im maturen Ölfeld sollen in einer Literaturstudie durchgeführt werden. Betrachtungsgegenstand ist hierbei der abgegrenzte Bereich von der Sonde selbst (inkl. Pumpe) bis zur Übergabestation inkl. Pipelinesystem. Hierbei werden die Einsatzmöglichkeiten und Potentiale von KI-Techniken aus den Bereichen Knowledge Discovery in Databases, Predictive Maintenance, Zeitreihenprognose, Modellbildung, modellbasierten und modellfreien Optimalsteuerungen und -regelungen untersucht und beschrieben.
Ansprechpartner: Stefan Wittek
The project "AI-based flood warning system" is a practical application of AI forecasting systems to safety-critical areas. The application domain is the catchment area of Goslar with the existing sensor infrastructure and its historical data. The task of the artificial intelligence is to observe the current status of weather such as precipitation, soil moisture, solar radiation over a past period of time and put it in relation to the current water level in the settlement. A possible correlation is assumed, so that dangerous, sudden peaks can be predicted at an early stage. The forecasting method is thus a core element for the coordination of a series of structural measures for the prevention of sudden flooding events.
Der Einsatz von KI im Kontext der prädiktiven Instandhaltung von Abwasserkanälen birgt Vorteile, sowohl im Sinne der Kostenersparnis als auch der Vorbeugung von Totalausfällen und daraus ggf. resultierenden Verunreinigungen des Grundwassers. Ziel ist es hierbei, bisher analog ausgeführte Tätigkeiten, so beispielsweise die Klassifikation von Schäden am Kanal, in Zukunft automatisiert erfolgen zu lassen. Darauf aufbauend lässt sich insbesondere eine Strategie ableiten, mit der sich der zukünftige Instandhaltungsprozess optimieren lässt. Zum Zwecke der Nutzerfreundlichkeit wird zusätzlich ein 3D-Modell des gesamten Netzes entwickelt, mit der sich sowohl die Klassifikation, als auch die Prognose des Zustandes visuell einsehen lässt. Die geplanten Schritte hierbei reichen von der theoretischen Erarbeitung eines Konzepts, bis hin zu einem einsatzfähigen Prototypen. Die eingesetzten Techniken umfassen klassische statistische Methoden, bis hin zu maschinellen Verfahren im Sinne des Predictive Maintenance und der Klassifikation.
Ansprechpartner: Benjamin Acar
NFDI4Ing brings together the engineering communities and fosters the management of engineering research data. The consortium represents engineers from all walks of the profession. It offers a unique method-oriented and user-centred approach in order to make engineering research data FAIR – findable, accessible, interoperable, and re-usable.The consortium represents researchers from all engineering disciplines. It offers a unique method-oriented and user-centered approach to make research data FAIR - discoverable, accessible, interoperable and re-usable.
NFDI4Ing defines a total of 7 archetypes for scientists as users of research infrastructure.The ISSE is mainly active in the BETTY archetype: engineering research software. This includes in particular the code of simulation models, and questions concerning the integration of heterogeneous models, as well as their approximation with the help of AI methods.
Contact: Stefan Wittek
In the north of Germany, the Harz, with its annual precipitation of 1200 mm on the Clausthal plateau and its storage capacity, through the numerous tailings ponds and dams, supplies regions as far away as Bremen with drinking water. Due to this importance in drinking water supply, but also due to the existing underground infrastructure of the mining heritage and due to its central and cross-border location, the Harz region is of outstanding water management importance in Germany. AI-supported water management can help to mitigate the extreme weather events that occur due to climate change, such as heavy rainfall events with flood disasters or long periods of drought. Possible solutions should both show the user options for action and make the use of conventional, numerical models more efficient and effective. The Harz region is to be developed as a digital twin and serve as a basis for the application of complex simulations over a wide area. The focus of the modelling is on an AI-hybrid model that combines both conventional and innovative approaches and thus generates added value in terms of water quality and quantity.
Contact: Benjamin Acar
AKI is a collaborative project, between the Institute of Software and System Engineering, Öko-Institut e.V. and the Federal Office for the Safety of Nuclear Waste Disposal. The Achivement of the Project is to research the use and application of artificial intelligence, i.e., AI, in the selection process in the field of repository research. Foremost, the structuring and evaluation criteria of the possible applications of the AI are to be carried out. This evaluation shall be considered concerning scenarios or geoscientific questions. The result should be a transparent and explainable comparison, classification and evaluation of the possibility to apply and usefulness of machine learning methods in the field of geosciences. Thereby, research gaps in the geosciences shall be identified, and R&D needs shall be named. The essential use of AI will be made in interacting with a so-called expert system. One essential transparency criterion is that artificial intelligence will take over particularly time-consuming and repetitive tasks, but the human decision-maker should make the final decision in the so-called expert system.
Contact: Dimitri Bratzel
This project aims to develop subject-specific and interdisciplinary study and learning content on artificial intelligence, particularly in the field of machine learning, for all status groups in universities and to enable its use by external stakeholders (i.e. the interested public) across universities.
universities and to enable their use by external stakeholders (i.e. interested members of the public). As a decentralized platform for these coordinated development activities, an AI hub will be established, which will enable coordinated development and usage planning as well as the implementation of didactically secured and innovative learning and mediation concepts of methods and application-oriented tools in the context of data-driven modeling, analysis and simulation for different target groups and promote AI-based innovations through participatory formats.
Contact: Stefan Wittek
The aim of the research project is to develop methods, precautionary concepts and transferable practical tools for extreme events for water supply companies and the water management administration. In this context, relevant sub-areas of water supply from water extraction and treatment to operation and water use are addressed, as are the different types of raw water: groundwater, spring water and surface water. In representative model regions of the German long-distance and surface water supply, an exemplary implementation takes place. This allows further long-distance and surface water suppliers or local water suppliers to adapt their respective different local requirements in a practical way. Ultimately, the project should also provide impulses for rule-making, e.g. in risk management and for cooperation with state authorities in extreme situations. The project should also raise awareness for the challenges of water supply during extreme events. A modern public relations work shall sensitize water supply companies with comparable challenges as well as interested end customers for the topic of extreme events and show the possible solutions developed in the project.
Contact person: Stefan Wittek
This project is part of the CircularLIB research training group on the cycle-oriented production of lithium-ion batteries. In this program the three universities TU Braunschweig, TU Clausthal and Leibnitz Universität Hannover are cooperating with the Fraunhofer Institute for Surface Engineering and Thin Films IST.
The aim is to research a hybrid AI-based modeling approach for lithium-ion batteries. Recently, machine learning has also been increasingly used in the context of classical engineering disciplines. In particular, deep learning approaches provide fast and relatively accurate modeling approaches based on learning specific tasks from examples. This is of crucial interest especially in the context of high-dimensional complex systems, where the underlying physics is not fully known or the computational effort for simulation with conventional numerical methods is high. Another aspect driving the use of deep learning approaches is the need for fast models that can be used in iterative tasks such as optimization and control. However, data-driven models are typically black-box approaches that are developed based only on data and do not explicitly incorporate physical knowledge or constraints into model development. This can result in such models lacking robustness and accuracy, especially with limited training data.
Recent advances in physically-informed machine learning have led to a number of approaches that are well suited for solving various types of partial differential equations (PDEs). This project aims to develop models based on both data and available physical knowledge for lithium-ion batteries using physically informed machine learning techniques that allow for a hybrid AI-based modeling approach.
Contact: Hamidreza Eivazi Kourabbaslou