As automation shifts towards human-centric paradigms under Industry 5.0, the ability to objectively assess human physiological and psychological states becomes vital for safe and effective human-machine cooperation. This paper presents the development and integration of two embedded measurement systems: the Human Measurement Device (HMD) and the Stress Inducing Device (SID). The HMD is a modular, real-time physiological data acquisition platform designed for continuous monitoring of key human indicators such as heart rate variability, electrodermal activity, respiration, and foot pressure. Complementing this, the SID elicits controlled stress responses via gamified cognitive tasks, enabling systematic data generation. Together, these tools facilitate synchronized, high-fidelity analysis of human reactions in dynamic, automated environments. Initially applied within automated driving scenarios to assess trust and comfort, the systems are extensible to broader Industry 5.0 applications, including human-robot collaboration and user interface design. The work contributes a scalable, objective methodology for real-time human state assessment, supporting improved safety, performance, and acceptance in human-in-the-loop systems.
Estimation is an important challenge in software engineering, determining cost, effort, and resource planning. This paper presents an adapted Multivocal Literature Review (MLR) synthesising formal and grey literature on four main estimation approaches identified in literature: algorithmic, expert-based, machine learning, and ensemble-based hybrid approaches. This review discusses the strengths and limitations of each approach, identifying the emerging role of ML models, while also examining non-ML based agile approaches such as planning poker and story points. The findings indicate that estimation is both complex and non-deterministic, where no single approach universally applies across all contexts. Ensemble-based hybrid models which employ a variety of estimation techniques in parallel report promising results in terms of accuracy and adaptability.
This study conducts a Multivocal Literature Review (MLR) to explore the strengths and weaknesses of Artificial Intelligence (AI) in software engineering, focusing on four key areas: reliability, sustainability, security and convenience. AI tools improve efficiency and automation, but they also have challenges such as bias, non-determinism and model transparency. There are threats to the sustainability of AI in software engineering, such as high computational energy demands and legal constraints on proprietary and copyrighted data. Security risks can also emerge from AI-generated code, as models may produce insecure implementations that require human oversight to ensure robustness. Even with these challenges, AI seems to be gaining in popularity due to its convenience, enabling developers to streamline workflows and enhance their productivity. This study explores these strengths and weaknesses and provides insights into the future of AI in software engineering.
Process Debt (PD), a concept derived from the Technical Debt, consists of a sub-optimal activity that might have short-term benefits but generates a negative impact in the medium-long term. PD identification is a key phase of PD management as it aims to determine the type of PD, where it is located, and how to estimate its impact. However, PD identification is the most challenging phase, as practitioners find it the most effort-intensive, mainly due to the immaterial nature of the process. In this paper, to mitigate the difficulties of PD identification, we propose a methodology that relies on frameworks compliant with the ISO/IEC 33000 family standard, the reference standard for process assessment. We also provide an exemplar application using the data from a process assessment performed using the Automotive SPICE, a framework compliant with the ISO/IEC 33000 requirements. An excerpt of the case study results is presented, showing the potential for a systematic and effective approach to PD identification.
Cultural Heritage has moved into focus of the European Union as engine for employment, economic drive and development. According to the report of the Horizon 2020 Expert Group on Cultural Heritage “Cultural heritage must be seen as a special, but integral, component in the production of European GDP and innovation, […]”. To guarantee longevity usability of European heritage it is inevitable to look at all options. The role of Cultural Heritage for climate protection and the climate impact on Cultural Heritage are also in the focus of the OMC report “Strengthening Cultural Heritage Resilience for Climate Change”. Ensuring and increasing biodiversity will also be a major task for Cultural heritage site owners/managers. The INTERREG project “Cities. Gardens. Diversity” is working on reducing heat hotspots in historic sites, like the Imperial Palace Vienna district by identifying appropriate measures to reduce the solar entry and increase the biodiversity in the district. All these experiences will be integrated into a Vocational Education and Training programme which will be developed by Burghauptmannschaft Oesterreich (BHOe), who is responsible for the district. These course modules will be integrated in the training offers by The European Heritage Academy (EHA), which was established in 2017, and are accessible online and in person. All further education offers by EHA provide certification by ECQA, participants can obtain their certificate through validation of prior learning, eLearning and participation in course modules before passing an exam.
Programming language selection is a critical decision in software engineering, affecting performance, security, and overall system reliability. In this paper, we present a Multivocal Literature Review (MLR) of four languages – Go, Haskell, Python, and Rust – chosen for their distinct paradigms and widespread adoption in industry. We synthesise findings from both peer-reviewed works and trusted grey literature to evaluate these languages across three key dimensions: Safety Security, Resource Efficiency, and Domain Suitability. Our analysis reveals that Go’s lightweight concurrency model excels in cloud-native services, yet lacks proper safety tooling for parallel programming. Haskell’s purely functional design and strong type system foster correctness and verifiability, but lazy evaluation may hinder efficiency. Python dominates data-centric fields due to its extensive ecosystem and simplicity, though it suffers from interpreted overhead and package-vulnerability risks. Rust’s ownership model mitigates memory-related vulnerabilities, offering near-C performance and robust concurrency, albeit with a steeper learning curve. By illustrating each language’s strengths, trade-offs, and optimal usage contexts, this review provides researchers and practitioners with actionable insights for informed language selection for software development.
Our research goal is to identify the ethical elements that autonomous vehicles should possess to coexist harmoniously with other road users in a society where autonomous vehicles are prevalent. Although autonomous vehicles are driverless, ethical issues still exist. In particular, autonomous vehicles are a system in themselves, and at the same time, they are merely an object that interacts with other road users in a social system. In previous research, we have proposed a hierarchy of ethics: macroethics, mesoethics, and microethics. Of these, microethics is most relevant to the behavior of systems in society. In this paper, we examine ethics, particularly microethics, through documents related to the test scenario category for autonomous vehicles. As a result, not only is it necessary for a system to be able to make appropriate predictions, but it also needs to have the function of anticipation before it can make those predictions.
Integration of AI-based autonomous systems in real-time environments is advancing rapidly. Technologies enhance ADAS and broaden autonomous vehicle capabilities, emphasizing trust and user acceptance. Achieving safe lane-keeping performance requires demonstrators that operate in real time. This paper introduces "RaspiCar", research platform tailored to evaluate control algorithms within embedded systems. Traditional research emphasizes accuracy, while "RaspiCar" explores AI safety and real-time performance. To ensure safety and real-time decisions, architecture with ZeroMQ is integrated, enabling implementation on resource-limited hardware. The platform's real-time performance is evaluated using PilotNet for steering inference and LaneNet and SegNet for lane segmentation, selected for their complementary strengths. Additionally, a safety mechanism is introduced to enforce emergency stops, mitigating hardware faults and model errors. We demonstrate that real-time AI-driven lane detection runs reliably and safely on low-power platforms. "RaspiCar" provides a scalable framework for advancing real-time autonomy in robotics and industry, enhancing the reliability of lane detection.
In July 2024, the European Union's (EU) Corporate Sustainability Due Diligence Directive (CS3D) was published in the EU Official Journal. This European Supply Chain Act requires companies to monitor their upstream and downstream business partners regarding a set of criteria, such as compliance with human rights, work safety, and environmental regulations. Once in force, companies will be obligated to examine their operations and assess their suppliers' risks of violations. For larger firms with thousands of suppliers, this represents a significant challenge. However, the directive is introduced at a time when supply chain issues are much broader than compliance aspects. The current global situation is characterized by resource shortages and supply disruptions in all industries. Thus, the mandatory monitoring required by this new regulation also provides an opportunity for companies to make their supply chains more resilient in the future. This paper describes the design of a corporate sustainability due diligence process assessment model and depicts its usage as a capability and compliance assurance mechanism embedded within Scrioo, an AI-powered supply chain risk management platform.
The increasing complexity of automotive systems necessitates robust security measures to protect against potential threats. ISO/ SAE 21434 mandates that Threat Analysis and Risk Assessment (TARA) is performed during the concept phase. However, cybersecurity must be maintained throughout the entire product lifecycle, including production, operation, maintenance, and decommissioning. Consequently, the TARA will be continuously updated with each newly detected vulnerability. Therefore, new vehicle engineering requires strategies to optimize the development of vehicle-level TARAs, which then serve as the initial TARA reference. This study focuses on a multi-layer, non-project-based TARA strategy for the development of future vehicles. Such reference TARA will rely on catalogs based on the most recent TARA updates. Our TARA approach focuses on dividing the architecture into layers based on attack steps to ensure the identification and mitigation of potential security threats in automotive systems. The attacks are also classified in a manner that links them to architecture and assets. Therefore, a generic non-project template was created using a TARA-specific tool, designed to be the initial TARA for the development of every new vehicle.
This paper presents a methodology for integrating Functional Mockup Units (FMUs), based on the Functional Mock-up Interface (FMI) standard, into real-time industrial control systems. Originally designed for design-time simulation, FMUs pose challenges in deterministic execution and system safety when deployed in operational environments. To address this, a modular C-based wrapper has been developed to manage FMU behavior within real-time constraints. Planned integration of Linux seccomp filters will further sandbox FMUs to prevent unsafe system calls. Preliminary validation using simplified test cases, including a low-pass filter and RS flip-flop, demonstrates the feasibility of this approach. The methodology supports enhanced testing, improved system flexibility, and lays the foundation for real-time Digital Twin implementations. Future work includes the completion of sandboxing, extensive performance evaluations, and support for multi-FMU co-simulation. This approach aims to enable safe, deterministic FMU integration across diverse industrial domains, supporting the advancement of model-based design and operational intelligence.
In 2018 we presented an agile approach to process management applicable to both small and large organizations. Instead of a centralized one-size-fits-all process itself, a risk-based framework of process requirements is provided centrally whilst the valid processes themselves are defined and maintained at the most local level. This paper presents the advantages, costs, constraints, shortfalls and key findings identified in applying it for more than six years to an organization in constant change. It points out potential improvements and illustrates best practices in applying this approach.
Generative Artificial Intelligence (GenAI) technologies have rapidly permeated personal applications and industries owing to unprece-dented efficiency gains. This swift adoption has been accompanied by accelerated technological evolution and market proliferation of diverse generative solutions. Consequently, users struggle to evaluate GenAI-enabled products against their specific requirements, as traditional software quality assessment frameworks inadequately address the unique characteristics of AI systems, often prioritizing technical metrics over user-required alignment, which is only apparent through practical applications. This research comprehensively examined current evaluation approaches, enabling users to objectively assess GenAI products. Through a systematic literature review (SLR) of 42 studies, we identified and analyzed methodologies that facilitate meaningful comparisons of GenAI solutions based on requirement fulfillment. Our findings reveal four primary evaluation approaches (benchmark-based, modelbased, human-assisted, and automated) and establish a three-layer quality taxonomy that distinguishes traditional software, AI-enabled, and GenAI-specific quality attributes. Our analysis emphasizes user-centered quality evaluation paradigms that bridge the gap between technical performance metrics and actual user values across diverse application contexts.
This study explored how sustainable urban development can be supported through emerging technologies, process improvement, and citizen commitment employed in the adoption of Positive Energy Districts (PEDs). The PED concept is a complex undertaking that requires a systemic, holistic, and integrated approach for its design, implementation, and operation as it addresses multiple interconnected aspects. This aim of the study has been to clarify different PED aspects by carrying out a factor analysis exploring sociocultural, technological, economic, environmental, political, legal, ethical, and demographic (STEEPLED) dimensions. Additionally, the study highlights the main PED opportunities and challenges. The outcomes of this study aid in the understanding of critical factors in the creation, implementation, and operation of PEDs.
Artificial Intelligence (AI) has changed software development processes, including requirement engineering and testing. While AI, especially generative AI, excels in writing specifications and code, it creates a black box that does not disclose its internal processing. This is the consequence of the perceptron architecture with its hidden layers. Explainable AI (xAI) tries to shed light into the darkness of these layers, but this paper presents another approach. AI engines are prompted by defined and well-known processes and functions from Quality Function Deployment (QFD) to transparently create a software product, because the AI answers can be validated against QFD deployment. This paper explains the principles of QFD software development and testing process using generative AI and presents an example of automated testing.
The process assessment model (PAM) version 4.0 of Automotive Software process improvement and capability determination (ASPICE) enables assessors to rate the processes, which are in the scope of the project to be assessed, according to NPLF scheme of ASPICE. The scope of the project usually contains the process SUP.1 - Quality Assurance. The goal of this process is to ensure that output information items (OII) and processes comply with the defined criteria and found deviations are solved in time (Quality Assurance). On the other hand, the project is assessed by an assessor (ASPICE Assessment). This task usually is carried out by a different person than the quality engineer (QE) and shall be independent from the project. This paper shows how the assessor the QE the project team and management shall work together to get the best results for the assessed project based on the experience of the authors.
The Automotive SPICE® (ASPICE) assessment model is maintained by the Verband der Automobilindustrie e.V. (VDA). The current version PAM 4.0 ( http://www.automotivespice.com ), released in November 2023, covers almost every aspect during the development of embedded systems. The process reference model includes engineering relevant processes clustered in process groups for system, software, electronic hardware, and others. The necessity of electromechanics (EM) is not considered in the current version of the PAM. Thus, this paper shows how the needs of EM can be covered by ASPICE based on industry experience.
This DevOps Implementation Guide provides a comprehensive framework for integrating DevOps into the software development lifecycle, incorporating aspects from the ISO/IEC 29110 Basic Profile. By focusing on standardized practices, this guide encourages the integration of tools that enable a more efficient and controlled transition to a DevOps culture. This guide provides a comprehensive framework that spans planning and operation in production environments, emphasizing automation, cross-functional collaboration, technology, and measurement as essential pillars. Activities are categorized into the DevOps process phases: planning, coding, building, testing, release, deployment, operations, and monitoring, each paired with recommended practices and tools that facilitate effective implementation. The results of implementing the guide in a sample of undergraduates in software engineering and master's degree students confirm the usability and usefulness of this guide.
Most car manufacturers require their suppliers to achieve at least capability level 2 in an Automotive SPICE® assessment. The VDA Guidelines provide additional expectations and interpretations for assessors regarding generic practices. However, organizations still struggle to interpret these generic practices, especially process performance objectives (GP 2.1.1). In this paper, members of the SoQrates Initiative from major automotive organizations present additional expectations, practical examples, and best practices for generic practices at level 2, based on numerous assessments and process improvement initiatives. The results provided should not be considered the only truth, but rather as support for assessors and organizations in their interpretation and implementation.
The evolution of connected vehicle technologies, particularly vehicleto-everything (V2X) communications, presents new challenges in ensuring safety and security. Functional safety and cybersecurity, once considered separate domains, are now deeply interconnected, working together to protect modern transportation autonomous driving systems from both unintentional failures and intentional threats. In V2X environments, a vulnerability in one area can directly impact the other - for example, a cyber-attack can compromise the functionality of safety-critical systems. This paper analyses the overlap, interdependencies, and potential conflicts between ISO 26262 and ISO/SAE 21434 in the V2X context, the two leading standards that address safety and security in automotive systems. It highlights scenarios where security measures may unintentionally hinder security functions and vice versa. Through comparative examination, it identifies conceptual and procedural gaps that arise when both standards are applied in parallel, especially in distributed, real-time V2X communication scenarios. A joint consideration of both domains is essential to enable resilient and trustworthy V2X communications that meet the complex demands of future connected mobility.
