Fortiss

Boundary artefacts are shared artefacts that support collaboration by allowing different groups to interpret the same information in different ways. Software development activities benefit from them, as a single artefact can support stakeholders across different organisational boundaries. When these artefacts contain inconsistencies, such as incorrect information, practitioners’ trust in them may decrease, leading to inefficiencies in task execution. This study developed and evaluated a guideline to support the creation of boundary artefacts in software engineering contexts. We conducted a longitudinal, multi-phase study embedded in an industrial setting. The guideline was developed based on a literature review and prior findings from a previous case study and was then submitted for practitioner evaluation. A post-implementation analysis of the guideline was carried out after a period without researcher intervention. Our guideline consists of 10 principles grouped into three categories: (1) Scope: stakeholders, boundaries, and terminology; (2) Structure: artefact format, transference, granularity, and additions; and (3) Management: evaluation, ownership, governance, and integration. Practitioner evaluations suggested that these principles support the creation of reliable, predictable, and functional boundary artefacts. However, practitioners also noted challenges during use, including the time-consuming nature of the activity and difficulties in understanding the concept of boundary artefact. Overall, the guideline was well received. After the non-intervention period, it was adopted as a standard by the partner company for artefacts such as security testing, standards documentation, and requirements specifications. Adoption challenges persisted, including cultural barriers and comprehension issues. Further applications across different artefacts could clarify how the principles influence their reliability, functionality, and predictability.

Earth Faults are a common occurrence on Distribution Power Grids, and determining the exact fault location in a compensated MV grid structure is a cumbersome process, which can negatively impact grid reliability. It is imperative for the operators in this scenario to find out fast where an earth fault has occured on the grid. This paper proposes a correlation-based earth fault localisation (EFL), which makes a spatial comparision of stored RMS Voltage signature database from distributed measurement points with obtained measurements. Furthermore, we present another RMS-EFL approach based on Gated Graph Neural Networks, which consider the topological structure of the grid as an input to the model. The assessment of both approaches in a real MV grid topology with simulated faults shows that the proposed methods can provide accurate results and that those are robust with respect to parameters that are frequently hard to obtain in practical deployments.

CAD models using a boundary representation (BREP) are widely used across industrial applications. However, extracting and interpreting information from these models typically requires expert knowledge. In this paper, we present a GraphRAG-based system that lowers this barrier by enabling natural language access to detailed geometric and topological information. The approach builds upon an ontological representation of geometries, which is accessible via the semantic query language SPARQL. Large language models are used to translate user questions into formal queries and generate natural language replies, allowing for flexible, domain-aware interaction with the geometric data. The effectiveness of the system is demonstrated through a systematic evaluation using diverse CAD models and a range of natural language queries. Our system empowers domain experts to explore and analyze complex CAD models without requiring expertise in query languages or ontology design. By reducing the complexity of geometric information retrieval, the GraphRAG-based system enables more accessible, automated, and intelligent use of CAD data in industrial applications.

Human machine interaction is a huge source of inspiration in today's media art and digital design, as machines and humans merge together more and more. Its place in art reflects its growing applications in industry, such as robotics. However, those interactions often remains too technical and machine-driven for people to really engage into. On the artistic side, new technologies are often not explored in their full potential and lag a bit behind, so that state-of-the-art research does not make its way up to museums and exhibitions. Machines should support people's imagination and poetry in a seamless interface to their body or soul. We propose an artistic sound installation featuring neuromorphic body sensing to support a direct yet non intrusive interaction with the visitor with the purpose of creating sound scapes together with the machine. We design a neuromorphic multihead human pose estimation neural sensor that shapes sound scapes and visual output with fine body movement control. In particular, the feature extractor is a spiking neural network tailored for a dedicated neuromorphic chip. The visitor, immersed in a sound atmosphere and a neurally processed representation of themselves that they control, experience the dialogue with a machine that thinks neurally, similarly to them.

Robot evolution systems in which bodies and brains evolve in tandem can be significantly improved by extending them with the ability to learn. Technically, this means that 'newborn' robots are given the opportunity to optimize their inherited brain to control the inherited body adequately. Robots are in an underdeveloped 'infant' stage during this learning stage since their brains and fitness are still being improved. An open issue with regard to this infancy period is that of 'fertility': Should the robot be eligible for mating during the learning stage? This paper explores two distinct approaches from the literature, based on the Triangle of Life (TOL) model, where infant robots cannot produce offspring, and the Morphological Innovation Protection (MIP) mechanism, where they can. The main contribution is a new algorithm, TOL with infant fertility (TOL+IF), inspired by MIP. Experimental comparisons with TOL and MIP show that the new method is superior. TOL+IF is successful not only in producing robots with much higher fitness but also in maintaining the population diversity at higher levels and in evolving different interesting morphologies.

Testing Advanced Driving Assistance Systems (ADAS), such as lane-keeping functions, requires creating road topologies or using predefined benchmarks. However, the test cases in existing ADAS benchmarks are often designed in specific formats (e.g., OpenDRIVE) and tailored to specific ADAS models. This limits their reusability and interoperability with other simulators and models, making it challenging to assess ADAS functionalities independently of the platform-specific details used to create the test cases.This paper evaluates the interoperability of SensoDat, a benchmark developed for ADAS regression testing. We introduce OpenCat, a converter that transforms OpenDRIVE test cases into the Catmull-Rom spline format, which is widely supported by many current test generators. By applying OpenCat to the SensoDat dataset, we achieved high accuracy in converting test cases into reusable road scenarios. To validate the converted scenarios, we used them to evaluate a lane-keeping ADAS model using the Udacity simulator. Both the simulator and the ADAS model operate independently of the technologies underlying SensoDat, ensuring an unbiased evaluation of the original test cases. Our findings reveal that benchmarks built with specific ADAS models hinder their effective usage for regression testing. We conclude by offering insights and recommendations to enhance the reusability and transferability of ADAS benchmarks for more extensive applications.

Test Input Generators (TIGs) are crucial to assess the ability of Deep Learning (DL) image classifiers to provide correct predictions for inputs beyond their training and test sets. Recent advancements in Generative AI(GenAI) models have made them a powerful tool for creating and manipulating synthetic images, although these advancements also imply increased complexity and resource demands for training. In this work, we benchmark and combine different GenAI models with TIGs, assessing their effectiveness, efficiency, and quality of the generated test images, in terms of domain validity and label preservation. We conduct an empirical study involving three different GenAI architectures (VAEs, GANs, Diffusion Models), five classification tasks of increasing complexity, and 364 human evaluations. Our results show that simpler architectures, such as VAEs, are sufficient for less complex datasets like MNIST. However, when dealing with feature-rich datasets, such as ImageNet, more sophisticated architectures like Diffusion Models achieve superior performance by generating a higher number of valid, misclassification-inducing inputs.

The energy system in Germany consists of a large number of distributed facilities, including millions of PV plants, wind turbines, and biomass plants. To understand and manage this system efficiently, accurate and reliable information about all facilities is essential. In Germany, the Marktstammdatenregister (MaStR) serves as a central registry for units of the energy system. The reliability of this data is critical for the registry's usefulness, but few validation studies have been published. In this work we provide a review of existing literature that relies on data from the MaStR and thereby show the registry's importance. We then build a data and testing pipeline for relevant data of the registry, with a focus on the two aspects of facility's location and size. All test results are published online in a reproducible workflow. Hence, this work contributes to a reliable data foundation for the German energy system and starts an open validation process of the Marktstammdatenregister from an academic perspective.