Publikationer från Högskolan i Skövde
Not a member yet
11110 research outputs found
Sort by
Implementation of a robotic cell for high precision battery assembly and disassembly
No full textSince the Industrial Revolution, advances in robotics and automation have enhanced reliability, efficiency and precision in manufacturing. Recently, interest in integrating collaborative robots with vision systems to optimize production tasks has grown. At the same time, increasing environmental concerns have driven the introduction of circular economy principles, promoting reuse and recycling of components at the end of life. This thesis presents the design and construction of an automated station for assembling and disassembling modules. A collaborative robot with vision system is integrated capable of detecting and manipulating components within the station, and 3D printed fixtures are used to facilitate safe and correct handling. Experimental tests demonstrate the station’s ability to perform both processes without human intervention. The resulting model offers a scalable solution for implementing cells circularity in sustainable industrial applications.
Disinformation and generative language models - Challenges and opportunities in the information flow : A technical and ethical analysis of AI-driven disinformation and counter-disinformation
No full textStudien sätter fokus på hur generativa språkmodeller kan optimeras för att bekämpa AI-genererat textbaserad desinformation. Studien hade som uppdrag att först och främst dyka in och förstå vad AI-genererad desinformation innebär och hur spridningen går till samt dess konsekvenser. Tekniska och etiska utmaningar identifierades under studien som står mot optimeringar av generativa språkmodeller för att bekämpa spridningen av AI-generad textbaserad desinformation. Ett flertal akademiska texter och vetenskapliga publikationer har inkluderats i studien för att tydliggöra hur forskningen ser på fenomenet och hur långt dessa forskningar har kommit. Studien har ett kvalitativt tillvägagångssätt och har inkluderat akademiska texter och semi-strukturerade intervjuer för att analysera elementen som identifierades i studien och finna ett svar på forskningsfrågan.Det finns övrigt digitalt material (t.ex. film-, bild- eller ljudfiler) eller modeller/artefakter tillhörande examensarbetet som ska skickas till arkivet.There are other digital material (eg film, image or audio files) or models/artifacts that belongs to the thesis and need to be archived.</p
Detection of microorganisms and genes conferring antibiotic resistance in human plasma using nanopore sequencing with MinION for early sepsis diagnosis
No full textSepsis is a serious condition caused by a dysregulated immune reaction. It is a common cause of death and can have many long-lasting health consequences for the survivors. The lack of early and appropriate treatment options decreases the chances of survival. The current sepsis diagnostic is based on time-consuming blood culturing (BC), with low sensitivity. Therefore, the ongoing research on sepsis focuses largely on finding new biomarkers and methods for early sepsis diagnosis. One of the promising biomarkers is cell-free DNA (cfDNA), which can provide information about the current condition as well as the prognosis of the patient. This project aimed to analyse plasma samples, non-spiked and spiked with microbial DNA, through sequencing with the MinION device to evaluate the effectiveness of nanopore sequencing for detecting different species and genes conferring antibiotic resistance. To achieve this, the plasma isolated from healthy human blood was spiked with different amounts of microbial DNA. The total DNA was extracted from the spiked and non-spiked plasma samples. The extracted DNA was sequenced using the MinION device and a Flongle flow cell. The obtained data was analysed with EPI2ME wf-metagenomics, and the Bacterial and Viral Bioinformatics Resource Center (BV-BRC) Taxonomic Classification (TC) and Metagenomic Read Mapping (MRM) tools. The microorganisms were identified at the species and genus levels. The identification of genes conferring antibiotic resistance was not as efficient as species identification, allowing for the detection of a fraction of the expected genes. The reason could be the low quality of reads arising from high fragmentation of DNA. It might be due to a low concentration of extracted DNA, resulting in highly fragmented DNA during library preparation. An optimisation of the procedure or the use of a different sequencing kit might improve the results. More research, however, is needed to assess that
Optimizing real-to-synthetic data ratios for enhanced remaining useful life prediction
No full textAccurate prediction of a system’s remaining useful life (RUL) is critical for maintenance planning and cost-effective operations. However, real run-to-failure data are usually scarce, which limits model training. Synthetic data generation through augmentation is a promising solution that can enrich training sets by creating realistic failure examples. This thesis investigates what is the optimal ratio of real and augmented data to improve RUL model performance. To achieve this, a series of experiments was conducted using LSTM-based models for the evaluation on the NASA C-MAPSS turbofan engine datasets (FD001 and FD004), and for the augmentation Gaussian Noise Injection and Conditional Generative Adversarial Networks (cGANs) were used. The results show that the optimal data mix depends on dataset complexity. For the simpler FD001 set, adding augmented data to the original dataset until the synthetic data represents 40–45% of the whole dataset yielded the lowest prediction error. It is not as clear for the more complex FD004 set, where the best mixture is somewhere in a wider span of roughly 10–50% synthetic data, with no single ratio clearly dominating. Importantly, the augmentation method (Gaussian vs. cGAN) had little effect on this optimal range. These and additional findings provide a good starting point for balancing real and augmented data in RUL tasks, however, further research is certainly needed to confirm these trends in other scen
Investigating the current methods and challenges for digital forensic investigations in smart homes : A systematic literature review
No full textWith the continuous growth and adoption of smart homes, it is inevitable that smart home systems and devices become more exposed to crimes and criminals. Due to this fact, researchers and forensic experts have started to explore how to forensically utilise these systems and devices as witnesses and evidence sources to crimes that transpire in the home. The purpose and aim of this systematic literature review was therefore to identify which methods can be used to conduct a digital forensic investigation in smart home environments, and which challenges exist that might negatively affect, or hinder, the investigation. The search for previous research was performed with one defined search term, in six different databases, which ultimately resulted in the acceptance of 54 articles. These 54 articles were then qualitatively analysed using thematic analysis/coding, in order to produce this study’s results, and answered the research question that this study set out to answer. The qualitative analysis ultimately resulted in the identification of fifteen separate subthemes, across the two pre-defined themes. Five of these subthemes related to methods that can be deployed in digital forensic investigations within smart home environments, while the other ten related to challenges that may complicate or hinder investigations. Furthermore, the results ultimately revealed that the most common method was physical acquisition, and that the least common methods were software acquisition and cloud forensics, whilst the most common challenges were data location, and data issues, and the least common challenge was privacy issues.
Exploring the victim's role in digital forensics : A scoping review of academic focus and gaps
No full textI takt med att digital teknik blir alltmer integrerad i det moderna samhället, växer också den digitala forensikens roll för att upprätthålla lag och ordning. För närvarande fokuserar den akademiska världen på nya tekniker och ramverk och deras potentiella implementering i digitala forensiska utredningar. Detta är utan tvekan en viktig faktor för digital forensik, men det finns en risk att man inte ser skogen för träden, med andra ord att man missar viktiga faktorer. I den här studien har betydlsen av att brottsoffret känner sig berört demonstrerats med hjälp av reparativ rättvisa. Implicita faktorer som brottsoffren själva bryr sig om identifieras, dessa faktorer är sekretess, utredningens längd och engagemang i utredningen. Explicit eller implicit oro för brottsoffer söktes i forskningsartiklar från fem databaser. Resultaten av denna scoping review visar att forskning som explicivt handlar om brottsoffer är extremt sällsynt. Forskning som implicit berör offren är en minoritet. Tillsammans representerar explicit och implicit koncern ungefär en fjärdedel av de granskade artiklarna. Denna under-representation kan betraktas som en lucka i forskningen som behöver åtgärdas i framtiden. As digital technology becomes increasingly integrated into modern society, the role of digital forensics in maintaining law and order grows alongside it. Currently, the focus of academia is, emerging techniques and frameworks, and their potential implementation in digital forensic investigations. While this is undoubtedly an important factor in digital forensics, it risks missing the forest for the trees and overlooking other critical elements. In this study, the importance of victim concern has been demonstrated using restorative justice. Implicit factors that victims themselves are concerned with are identified, these factors being privacy, duration of the investigation and engagement with the investigation. Explicit or implicit victim concern was searched for in research articles from five databases. The results of this scoping review shows that research explicitly concerned with the victims, is extremely rare. Research implicitly concerned is a minority. Together, explicit and implicit concern represent about a fourth of the reviewed articles. This under-representation, could be considered a gap in the research that needs to be addressed in the future.
Context-aware trajectory prediction and collision detection : A case study using NAS-optimized context-aware LSTM for cyclist-vehicle collision detection at roundabouts
No full textUrban roundabouts present complex interaction scenarios where Advanced Driver Assistance Systems (ADAS) often generate excessive false alarms, undermining driver trust and system effectiveness. This master’s thesis investigates whether integrating spatiotemporal context into a deep sequential model can improve collision-risk estimation and reduce unnecessary warnings in cyclist–vehicle encounters. Building on the hybrid Decision Tree + Random Forest approach of Atif et al. (2025), we implement a Long Short-Term Memory (LSTM) network whose architecture is fine-tuned via Neural Architecture Search (NAS). The LSTM ingests enriched trajectory data—incorporating traffic density, proximity cues, and lane alignment— from stereo-camera recordings at Gothenburg roundabouts (nearly two million observations). Performance is evaluated against the baseline using five-fold cross-validation, focusing on accuracy, recall, precision, and false-alarm rate. Results show the NAS-optimized LSTM achieves 95% accuracy and reduces false positives by 46% compared to the Random Forest baseline, while maintaining balanced recall (0.93). To ensure transparency, we apply SHAP and LIME explainability methods, revealing that collision predictions emerge from subtle temporal patterns rather than isolated static features. While the Random Forest excels in spatial interpretability, the LSTM’s sequence-level reasoning drives its superior false-alarm reduction. We discuss methodological trade-offs, generalizability to diverse urban environments, and ethical considerations—such as data privacy under GDPR and the risk of driver over-reliance on automation. These findings suggest that context-aware deep learning, when paired with explainable AI, can meaningfully enhance ADAS reliability and support safer, more trustworthy urban mobility. Det finns övrigt digitalt material (t.ex. film-, bild- eller ljudfiler) eller modeller/artefakter tillhörande examensarbetet som ska skickas till arkivet.There are other digital material (eg film, image or audio files) or models/artifacts that belongs to the thesis and need to be archived.</p
Machine learning-based prediction and feature importance analysis of thermocouple temperatures in industrial furnaces
No full textThis thesis presents a machine learning-based framework for predicting thermocouple temperatures and analysing feature importance in industrial furnaces. Utilizing high-frequency time-series data from two furnaces with six thermal zones, the study explores the effectiveness of both linear (Lasso, Ridge) and nonlinear (Random Forest, XGBoost, LSTM) models in forecasting localized and aggregated temperature behaviours. Results indicate that nonlinear models, particularly XGBoost, achieve superior accuracy and execution efficiency, making them ideal for real-time deployment in furnace operations. Feature interpretability is addressed using SHAP and LIME, which highlight critical process variables such as fuel flow rates, zone-level average temperatures, and line control speed. These insights provide valuable input for process optimization and predictive maintenance. An adaptive thresholding mechanism is implemented using residual-based confidence intervals, enabling dynamic and statistically grounded detection of sensor anomalies. The hierarchical approach by structuring models at the zone level, furnace level, and whole unit level, give engineers the flexibility to pinpoint anomalies with precision and explore thermocouple (TC) changes in specific areas. This adds a layer of adaptability that’s crucial for troubleshooting and operational decision-making. Additionally, the study evaluates the utility of transfer learning for model reuse across zones, finding it beneficial for data-limited settings. To enhance operational usability, an interactive Power BI dashboard is developed for real-time monitoring, visualization, and anomaly exploration. This end-to-end framework demonstrates how predictive modeling, interpretability, and visualization can be integrated to support smarter, data-driven furnace operations in industrial environments
Implementation of an inverter test setup for electric vehicles : Semi-automated testing with an interactive interface
No full textGiven the growing tendency of electric mobility within the automotive sector, this thesis presents the implementation of a semi-automated inverter test bench meant for load testing and extracting performance-related figures from this crucial electric vehicle (EV) drivetrain component. The primary goal is to propose a flexible and practical platform, capable of providing the means to draw conclusions on production viability and flaws of inverter boards. The testing infrastructure integrates real-time automatized data acquisition, configuring the necessary parameters from a centralized LabVIEW Guided User Interface (GUI). The connection with lab instruments and sensors is achieved through a LAN network, given its practicality and effectiveness for this level of implementation. In particular, the focus of the research question is on solder layer quality and the effects it may have on the results. To study this phenomenon, electrical loads of different nature were used, including real electric motors and Resistive-Inductive (RL) loads. Comparing each case’s results, the impact of soldering joint defects is demonstrated to some extent, also exposing the limited capabilities of the used methods (based on voltage difference and on-state current measuring, temperature and X-ray scans). Local temperature increases were observed, proving the point of some related studies that will be later reviewed. On the other hand, the employed real EV inverter highlighted the importance of matching load characteristics with the design of the board, as well as the importance of optimized control methods. The study concludes that the proposed test bench solution offers a practical and scalable solution for load performance testing and early solder fault detection on the electric mobility field. Future work following this basis could lead to fully automated setups and integrated, higher-precision measurement methods capable of bringing more straightforward conclusions
Strategic flexibility in gaming firms : The role of resource management
No full textStrategic flexibility is a critical capability for firms to respond to turbulent business environment in a timely manner through resource deployment. However, the knowledge of how firms could better use their internal resources to develop strategic flexibility remains limited. Moreover, most existing studies on resource-related antecedents of strategic flexibility focus on financial resources. To address these gaps, this study aims to examine how resource management contributes to the development of strategic flexibility. Specifically, it explores the role of resource management in achieving strategic flexibility and investigates how human resource and technological resource activities influence strategic flexibility. The qualitative method was adopted and data was gathered by seven semi-structured interviews with 14 informants from seven small gaming firms in Sweden. Our empirical findings suggest that resource management (e.g. resource leverage and resource structure) may enable strategic flexibility. While human resource activities may have limited influence on strategic flexibility, technological resource activities (e.g., technological acquisition, upgrade, shift, and operation) may have positive influence on strategic flexibility, especially its proactive form. Furthermore, this study offers novel insights into the role of strategic orientations in enabling strategic flexibility, indicating that technological and market orientations may jointly contribute to strategic flexibility. The findings also suggest the integration of human resource and technological resource activities may positively influence strategic flexibility. This study advances the understanding of resource management in the development of strategic flexibility and provides valuable insights for managers and practitioners on how to effectively leverage their human and technological resources to sustain competitive advantage.Det finns övrigt digitalt material (t.ex. film-, bild- eller ljudfiler) eller modeller/artefakter tillhörande examensarbetet som ska skickas till arkivet.There are other digital material (eg film, image or audio files) or models/artifacts that belongs to the thesis and need to be archived.</p