GenAI has a potential to enhance the learning and teaching processes in engineering education. For instance, GenAI feedback on students' task performance can be effective depending on when such feedback is provided. However, little is known about how engineering faculty and instructors discover such potential within the scope of their instruction when they try out the technology for the first time. To this end, this study purported to describe an engineering instructor's and seven teaching assistants' initial experiences of integrating GenAI into their undergraduate engineering course and the corresponding changes in students' formative exercise performance. An embedded descriptive single case study design was employed. The corresponding research data included four interviews conducted at the beginning, middle and end of an academic semester, and students' formative exercise performance. Overall, after GenAI integration, students' formative exercise performance increased, and a critical and reflective practice of learning about how to integrate GenAI into instruction provided informative insights. Still, technology integration stayed at the level of replacing other instructional methods or increasing the efficiency of solving coding problems. It turned out to be exciting and surprising for students to be able to use GenAI in course work even though their use of the technology weakened over time. Our findings suggest that engineering teaching staff's initial experimental experiences with GenAI integration can be informative and provide context-specific practical insights. Therefore, it is reasonable for higher education institutions to encourage such experiences especially when there is a lot of unknown regarding an emerging technology.
The increasing use of Large Language Models (LLMs) offers significant opportunities across the engineering lifecycle, including requirements engineering, software development, process optimization, and decision support. Despite this potential, organizations face substantial challenges in assessing the risks associated with LLM use, resulting in inconsistent integration, unknown failure modes, and limited scalability. This paper introduces the LLM Risk Assessment Framework (LRF), a structured approach for evaluating the application of LLMs within Systems Engineering (SE) environments. The framework classifies LLM-based applications along two fundamental dimensions: autonomy, ranging from supportive assistance to fully automated decision making, and impact, reflecting the potential severity of incorrect or misleading model outputs on engineering processes and system elements. By combining these dimensions, the LRF enables consistent determination of corresponding risk levels across the development lifecycle. The resulting classification supports organizations in identifying appropriate validation strategies, levels of human oversight, and required countermeasures to ensure safe and transparent deployment. The framework thereby helps align the rapid evolution of AI technologies with established engineering principles of reliability, traceability, and controlled process integration. Overall, the LRF provides a basis for risk-aware adoption of LLMs in complex engineering environments and represents a first step toward standardized AI assurance practices in systems engineering.
There is a pressing need for better development methods and tools to keep up with the growing demand and increasing complexity of new software systems. New types of user interfaces, the need for intelligent components, sustainability concerns, etc. bring new challenges that we need to handle. In the last years, model-driven engineering (MDE), including its latest incarnation, i.e. low/no-code development, has been key to improving the quality and productivity of software development, but models themselves are becoming increasingly complex to specify and manage. At the same time, we are witnessing the growing popularity of vibe coding approaches that rely on Large Language Models (LLMs) to transform natural language descriptions into running code at the expense of potential code vulnerabilities, scalability issues and maintainability concerns. While many may think vibe coding will replace model-based engineering, in this paper we argue that, in fact, the two approaches can complement each other and provide altogether different development paths for different types of software systems, development scenarios, and user profiles. In this sense, we introduce the concept of \textit{vibe-driven model-based engineering} as a novel approach to integrate the best of both worlds (AI and MDE) to accelerate the development of reliable complex systems. We outline the key concepts of this new approach and highlight the opportunities and open challenges it presents for the future of software development.
The rapid rise of LLMs over the last few years has promoted growing experimentation with LLM-driven AI tutors. However, the details of implementation, as well as the benefit in a teaching environment, are still in the early days of exploration. This article addresses these issues in the context of implementation of an AI Teaching Assistant (AI-TA) using Retrieval Augmented Generation (RAG) for Trinity College Dublin's Master's Motion Picture Engineering (MPE) course. We provide details of our implementation (including the prompt to the LLM, and code), and highlight how we designed and tuned our RAG pipeline to meet course needs. We describe our survey instrument and report on the impact of the AI-TA through a number of quantitative metrics. The scale of our experiment (43 students, 296 sessions, 1,889 queries over 7 weeks) was sufficient to have confidence in our findings. Unlike previous studies, we experimented with allowing the use of the AI-TA in open-book examinations. Statistical analysis across three exams showed no performance differences regardless of AI-TA access (p > 0.05), demonstrating that thoughtfully designed assessments can maintain academic validity. Student feedback revealed that the AI-TA was beneficial (mean = 4.22/5), while students had mixed feelings about preferring it over human tutoring (mean = 2.78/5).
Context: Autism spectrum disorder (ASD) leads to various issues in the everyday life of autistic individuals, often resulting in unemployment and mental health problems. To improve the inclusion of autistic adults, existing studies have highlighted the strengths these individuals possess in comparison to non-autistic individuals, e.g., high attention to detail or excellent logical reasoning skills. If fostered, these strengths could be valuable in software engineering activities, such for identifying specific kinds of bugs in code. However, existing work in SE has primarily studied the challenges of autistic individuals and possible accommodations, with little attention their strengths. Objective: Our goal is to analyse the experiences of autistic individuals in software engineering activities, such as code reviews, with a particular emphasis on strengths. Methods: This study combines Social-Technical Grounded Theory through semi-structured interviews with 16 autistic software engineers and a survey with 49 respondents, including 5 autistic participants. We compare the emerging themes with the theory by Gama et al. on the Effect of Neurodivergent Cognitive Dysfunctions in Software Engineering Performance. Results: Our results suggest that autistic software engineers are often skilled in logical thinking, attention to detail, and hyperfocus in programming; and they enjoy learning new programming languages and programming-related technologies. Confirming previous work, they tend to prefer written communication and remote work. Finally, we report a high comfort level in interacting with AI-based systems. Conclusions: Our findings extend existing work by providing further evidence on the strengths of autistic software engineers.
Queer students often encounter discrimination and a lack of belonging in their academic environments. This may be especially true in heteronormative male-dominated fields like software engineering, which already faces a diversity crisis. In contrast, disciplines like humanities have a higher proportion of queer students, suggesting a more diverse academic culture. While prior research has explored queer students' challenges in STEM fields, limited attention has been given to how experiences differ between the sociotechnical, yet highly heteronormative, field of software engineering and the socioculturally inclusive humanities. This study addresses that gap by comparing 165 queer software engineering and 119 queer humanities students experiences. Our findings reveal that queer students in software engineering are less likely to be open about their sexuality, report a significantly lower sense of belonging, and encounter more academic challenges compared to their peers in the humanities. Despite these challenges, queer software engineering students show greater determination to continue their studies. These insights suggest that software engineering could enhance inclusivity by adopting practices commonly seen in the humanities, such as integrating inclusive policies in classrooms, to create a more welcoming environment where queer students can thrive.
FMs, particularly LLMs, are increasingly used to support various software engineering activities (e.g., coding and testing). Their applications in the software engineering of CPSs are also growing. However, research in this area remains limited. Moreover, existing studies have primarily focused on LLMs-only one type of FM-leaving ample opportunities to explore others, such as vision-language models. We argue that, in addition to LLMs, other FMs utilizing different data modalities (e.g., images, audio) and multimodal models (which integrate multiple modalities) hold great potential for supporting CPS software engineering, given that these systems process diverse data types. To address this, we present a research roadmap for integrating FMs into various phases of CPS software engineering, highlighting key research opportunities and challenges for the software engineering community. Moreover, we discuss the common challenges associated with applying FMs in this context, including the correctness of FM-generated artifacts, as well as the inherent uncertainty and hallucination associated with FMs. This roadmap is intended for researchers and practitioners in CPS software engineering, providing future research directions using FMs in this domain.
PurposeSevere scarcity of natural river sand (RS), exacerbated by environmental protection policies and extraction constraints, has significantly impacted aggregate supply for railway concrete. While manufactured sand (MS) offers a substitute for RS in railway applications, its widespread adoption in high-strength railway prestressed structures is challenged by lack of drying shrinkage and creep research data on concrete.Design/methodology/approachHigh-strength manufactured sand concrete (MSC) was prepared using MS with varying lithologies and stone powder contents. Its drying shrinkage and creep behaviors were evaluated in accordance with the Chinese standard GB/T 50082. The deformation mechanism was analyzed by combining nano-scratch testing.FindingsCompared to RS concrete, MSC from all tested lithologies showed higher drying shrinkage but lower creep deformation. The drying shrinkage rose steadily with increased stone powder content, while the creep strain displayed a distinct non-linear trend, decreasing first before rising. To prepare low-deformation MSC, select high-strength MS and limit stone powder content not greater 10%. Nano-scratch tests indicated that harder MS particles suppress microcracking at the interfacial transition zone (ITZ), improving the creep resistance. The predictive models for drying shrinkage and creep were also developed by incorporating coefficients for stone powder and lithology effects.Originality/valueThese findings serve as a foundation for the application of MSC in railway prestressed structures, offering both theoretical and practical guidance.
Transportation engineering, Railroad engineering and operation
Ladan Gholami, Pietro Ducange, Alberto Gotta
et al.
Accurate prediction of receiver state is vital for optimizing network performance in urban settings, where rapid spatial variations in channel conditions pose significant challenges to communication quality. This paper presents a Machine Learning-based framework for predicting channel states in Unmanned Aerial Vehicle-assisted mmWave communication networks. Given that mmWave signals are susceptible to blockage by buildings and other urban structures, predicting receiver conditions at a specific location can be determined by directly deploying the geometric features describing the built-up environment surrounding the receiver. A set of geometrical features is extracted and used as input to train the adopted learning models, namely Decision Tree, Linear Decision Tree (LDT), Random Forest, Support Vector Machine, and Deep Neural Network (DNN), to estimate the probability of three distinct receiver states: Line-of-Sight, Non-Line-of-Sight, and Blocked. Experimental results indicate that the DNN-based model achieves the highest prediction accuracy and robustness, while the LDT provides computational efficiency and straightforward explainability. To improve the interpretability of the black-box DNN model, we employ the SHapley Additive exPlanations (SHAP) method, which identifies the most influential environmental features in state probability prediction. Furthermore, we enrich the standard 3GPP model by incorporating the top SHAP-ranked features, leading to notable performance improvements.
Telecommunication, Transportation and communications
In this paper, we study the downlink precoding of spatial non-stationary (SnS) extremely large-scale multiple-input multiple-output (XL-MIMO) in the near-field channel with uniform spherical wave (USW) and mixed line-of-sight and non-line-of-sight environments. First, we present a novel precoding scheme by replacing the digital beamforming (BF) in the conventional hybrid BF with user-subarray pairing network. The presented scheme utilizes the SnS channel characteristic efficiently, and can facilitate the high-speed implementation and the deployment of low-resolution digital-to-analog converters. Next, we study the optimizations to maximize both the sum-rate and min-rate by designing the power allocation, user-subarray pairing network, and analog BF jointly. For the sum-rate maximization, we first reformulate the corresponding nonconvex problem by the quadratic transformation to facilitate the further processing. Then, the alternating optimization framework is utilized to optimize the variables alternately. Concretely, the Riemannian conjugate gradient method, projected gradient ascent method, and mathematical programming with equilibrium constraints alternating direction method are employed to optimize the analog BF, power allocation, and user-subarray pairing network, respectively. For the min-rate maximization, we extend the aforementioned solutions by proper modifications. Finally, the effectiveness of the proposed optimization algorithms is verified through computer simulations.
Telecommunication, Transportation and communications
Este estudo investiga a relação entre a ampliação da capacidade aeroportuária e a distribuição das operações ao longo do dia, com ênfase no impacto da alocação de slots sobre a eficiência operacional. A análise considera dados de 26 aeroportos brasileiros entre 2019 e 2024, utilizando indicadores como o CUI (Índice de Utilização da Capacidade) e a razão entre taxa de pico anual e capacidade horária instalada. Por meio de modelos de efeitos fixos aplicados a dados em painel, constatou-se que o aumento no número de passageiros é o principal fator associado à intensificação do uso da infraestrutura. Embora a expansão da capacidade tenha contribuído para reduzir a pressão nos horários de pico, não houve impacto significativo na redistribuição das operações ao longo do dia. Os resultados indicam que, além de investimentos em infraestrutura, é necessário aprimorar os mecanismos de alocação de slots e considerar políticas de incentivo à demanda em horários de menor movimento. A pesquisa reforça a importância de uma gestão coordenada entre infraestrutura e regulação para o uso eficiente e sustentável dos aeroportos brasileiros.
Margaret-Anne Storey, Rashina Hoda, Alessandra Maciel Paz Milani
et al.
Mixed methods research is often used in software engineering, but researchers outside of the social or human sciences often lack experience when using these designs. This paper provides guiding principles and advice on how to design mixed method research, and to encourage the intentional, rigorous, and innovative application of mixed methods in software engineering. It also presents key properties of core mixed method research designs. Through a number of fictitious but recognizable software engineering research scenarios, we showcase how to choose suitable designs and consider the inevitable trade-offs any design choice leads to. We describe several antipatterns that illustrate what to avoid in mixed method research, and when mixed method research should be considered over other approaches.
Lola Burgueño, Davide Di Ruscio, Houari Sahraoui
et al.
Model-Driven Engineering (MDE) provides a huge body of knowledge of automation for many different engineering tasks, especially those involving transitioning from design to implementation. With the huge progress made in Artificial Intelligence (AI), questions arise about the future of MDE, such as how existing MDE techniques and technologies can be improved or how other activities that currently lack dedicated support can also be automated. However, at the same time, it has to be revisited where and how models should be used to keep the engineers in the loop for creating, operating, and maintaining complex systems. To trigger dedicated research on these open points, we discuss the history of automation in MDE and present perspectives on how automation in MDE can be further improved and which obstacles have to be overcome in both the medium and long-term.
Deep learning models are usually black boxes when deployed on machine learning platforms. Prior works have shown that the attributes (e.g., the number of convolutional layers) of a target black-box model can be exposed through a sequence of queries. There is a crucial limitation: these works assume the training dataset of the target model is known beforehand and leverage this dataset for model attribute attack. However, it is difficult to access the training dataset of the target black-box model in reality. Therefore, whether the attributes of a target black-box model could be still revealed in this case is doubtful. In this paper, we investigate a new problem of black-box reverse engineering, without requiring the availability of the target model's training dataset. We put forward a general and principled framework DREAM, by casting this problem as out-of-distribution (OOD) generalization. In this way, we can learn a domain-agnostic meta-model to infer the attributes of the target black-box model with unknown training data. This makes our method one of the kinds that can gracefully apply to an arbitrary domain for model attribute reverse engineering with strong generalization ability. Extensive experimental results demonstrate the superiority of our proposed method over the baselines.
Multimodal GPTs represent a watershed in the interplay between Software Engineering and Generative Artificial Intelligence. GPT-4 accepts image and text inputs, rather than simply natural language. We investigate relevant use cases stemming from these enhanced capabilities of GPT-4. To the best of our knowledge, no other work has investigated similar use cases involving Software Engineering tasks carried out via multimodal GPTs prompted with a mix of diagrams and natural language.
In this paper, we study a autonomous aerial vehicle (AAV) inspection system. In this system, the AAV flies to all inspection points in a certain area for patrol inspection, and the energy of inspection AAV is limited. Our goal is to optimize the communication quality of the AAV by planning the inspection sequence and flight trajectory, so as to ensure that the AAV can complete the inspection task and minimize the outage time subject to limited energy of the AAV. To solve this problem, we propose a hybrid algorithm, which consists of simulated annealing (SA) algorithm and Dueling Double Deep Q Network (D3QN) algorithm. The SA algorithm is used to obtain the inspection sequence of the AAV with the most energy saving. On this basis, the D3QN algorithm is used to optimize the flight trajectory of the energy-limited inspection AAV. To prove the effectiveness of the sub-optimal solution obtained by our proposed algorithm, we use several algorithms as a comparison. Numerical results show that the proposed algorithm is effective in optimizing the communication quality of the inspection AAV with limited energy, and its performance is improved by about 15%-50% compared with other benchmarks.
Telecommunication, Transportation and communications
Abstract Layered rock slope exists widely. Because of its special slope structure, it is prone to bending deformation and toppling failure, which is a serious threat to engineering construction and safety operation. At present, the research of layered rock slope still has great innovation potential. During the construction of Wudongde Hydropower Station on Jinsha River, safety and stability problems such as slope geological structure development, face rock unloading and relaxation, and even slip and large deformation were encountered. Through field exploration, it is found that the rock and soil stratification of the slope on both sides of Wudongde Hydropower Station is highly obvious. At present, there is a lack of research on-site long-term displacement monitoring of layered rock high-steep slope, especially for layered slope in complex hydrogeology and construction environment. In order to strengthen the research on the deformation and stability of layered rock slope, this paper analyzes the measured displacement data of Wudongde hydropower station slope, and establishes three-dimensional geological finite element model with the help of numerical simulation software. The stability of the slope is calculated by combining the finite difference method and the strength reduction method. Finally, the evolution mechanism of the deformation of the layered rock slope is explained according to the geological structure characteristics. The main conclusions of this paper are as follows: the layered slope in the dam reservoir area is prone to deformation under the combined action of long-term construction disturbance and fissure water seepage, and the construction disturbance has a strong influence on the artificial excavation area below 1070 m, and the maximum rock mass deformation and surface displacement in the artificial excavation area of the slope reach 92.2 mm and 312.5 mm, respectively. However, the influence of construction disturbance on the natural mountain above 1070 m is limited, the valley deformation of the natural mountain on the left bank of the reservoir area is higher than that on the right bank, and the cumulative deformation is still less than 20 mm. The influence of seepage on the displacement of the area with higher elevation at the top of the slope is more obvious, and the influence of excavation and other disturbances on the displacement of the artificial excavation area with lower elevation is more obvious. The deformation of the river valley in the water cushion pond behind the dam increases slowly, and the change trend of the field deformation data is mostly consistent with that of the numerical calculation. The horizontal shrinkage of the mountains on both sides shows a contraction trend on the whole, and the maximum horizontal shrinkage calculated by numerical simulation is close to 20 mm, which is located at the elevation of 990 m.
Cosmina-Cristina Ratiu, Christoph Mayr-Dorn, Alexander Egyed
Engineering processes for safety-critical systems describe the steps and sequence that guide engineers from refining user requirements into executable code, as well as producing the artifacts, traces, and evidence that the resulting system is of high quality. Process compliance focuses on ensuring that the actual engineering work is followed as closely as possible to the described engineering processes. To this end, temporal constraints describe the ideal sequence of steps. Checking these process constraints, however, is still a daunting task that requires a lot of manual work and delivers feedback to engineers only late in the process. In this paper, we present an automated constraint checking approach that can incrementally check temporal constraints across inter-related engineering artifacts upon every artifact change thereby enabling timely feedback to engineers on process deviations. Temporal constraints are expressed in the Object Constraint Language (OCL) extended with operators from Linear Temporal Logic (LTL). We demonstrate the ability of our approach to support a wide range of higher level temporal patterns. We further show that for constraints in an industry-derived use case, the average evaluation time for a single constraint takes around 0.2 milliseconds.