Efter flera decennier av utveckling har Virtual Reality (VR) blivit ett modernt och populärt sätt att spela datorspel på. VR-headsets ses idag som ett redskap för att skapa immersion hos spelare. Denna studie menar att ta reda på om externa faktorer kan stärka immersion i VR, och använder sig av spelet Richie's Plank Experience (Toast Interactive, 2016). I det genomförda testet fick tolv testpersoner gå ut på en matta eller en riktig planka med en fläkt som blåser mot dem. Studien visar att externa faktorer kan användas för att öka immersionen hos en användare och att i denna studie var fläkten det som ökade immersionen mest i VR. Den största bidragande faktorn till immersion visade sig vara rädsla, som kan användas varsamt för att stärka immersion. I framtiden kan denna forskning användas för utveckling av bland annat VR-spelarkader som kan utnyttja liknande tekniker för att höja immersion för sina spelare.

Detta examensarbete utforskade utvecklingen av avancerade maskininlärningsmodeller föratt förbättra autonoma transportsystem. Genom att fokusera på identifiering och klassificering av trafikljussignaler, bidrog arbetet till säkerheten och effektiviteten hos självkörandefordon. En granskning av modeller som Single Shot MultiBox Detector (SSD), som objektdetekteringsmodell, InceptionV3 och VGG16, som klassificeringsmodeller, genomfördes,med särskild vikt på deras träning och testningsprocesser.Resultaten, med avseende på valideringsnoggrannhet ’accuracy’ och valideringsförlust(loss), visade att InceptionV3-modellen presterade väl över olika parametrar. Denna modellvisade sig vara robust och anpassningsbar, vilket gjorde den till ett bra val för projektets målom noggrann och pålitlig klassificering av trafikljussignaler.Å andra sidan visade VGG16-modellen varierande resultat. Medan den presterade väl undervissa förutsättningar, visade den sig vara mindre robust vid vissa parametrarinställningar,speciellt vid högre batch-storlekar, vilket ledde till lägre valideringsnoggrannhet och högrevalideringsförlust. / This thesis explored the development of advanced machine learning models to improve autonomous transportation systems. By focusing on the identification and classification of traffic light signals, the work contributes to the safety and efficiency of self-driving vehicles. Areview of models such as the Single Shot MultiBox Detector (SSD), as an object detectionmodel, and InceptionV3 and VGG16, as classification models, was conducted, with particular emphasis on their training and testing processes.The results, in terms of validation accuracy and validation loss, showed that the InceptionV3model performed well across various parameters. This model proved to be robust and adaptable, making it a good choice for the project's goal of accurate and reliable classification oftraffic light signals.On the other hand, the VGG16 model showed varying results. While it performed well undercertain conditions, it proved to be less robust at certain parameter settings, especially at higherbatch sizes, which led to lower validation accuracy and higher validation loss.

This thesis has been written on behalf of Västerås stad with the purpose to test and evaluate methods for automatic classification of land use from orthophotos. Today, only manual classification of land use based on orthophotos is used, which is a time-consuming task. Västerås stad therefore want to find a method to use in the future that is more efficient and provides equal quality. The land use classes used today, and which will also be used for the automatic classification, are: ●      Green area ●      Asphalt ●      Gravel ●      Sand ●      Other hard surfaces ●      Building In the project, two different methods for automatic classification were tested: classification algorithms and deep learning. Deep learning was used to train a model to classify land use from orthophotos according to the previously mentioned classes. The same set of training samples was used for all classification algorithms and for the training of the deep learning model. Accuracy calculations were performed for all classifications. The trained deep learning model achieved an accuracy of 0.843284. The classification algorithms tested were Support Vector Machine, Random Trees, K-Nearest Neighbor, and Maximum Likelihood. These were tested with different sets of input data. The results showed that Support Vector Machine with a normalized Digital Surface Model (nDSM) and Normalized Difference Vegetation Index (NDVI) as supplementary input data to the orthophoto provided the highest accuracy, with an overall accuracy of 0.852886, which is also better than the result from the deep learning model. After the best result was identified, it was generalized to remove small, misclassified segments. After generalization, the overall accuracy increased to 0.882682. The classes gravel and other hard surfaces stood out with particularly low accuracy in all classifications. When comparing the manually performed classification with the automatically created classifications, clear interruptions in the road network are noticeable. This has been identified as being due to trees and vegetation obscuring the ground beneath. This is a problem that seems difficult to solve with current automatic classification methods. / Detta examensarbete har genomförts på uppdrag av Västerås stad med syftet att testa och utvärdera metoder för automatisk klassificering av markanvändning från ortofoton. Idag används endast manuell klassificering av ortofoton, vilket är ett tidskrävande arbete. Västerås stad önskar därför hitta en metod att använda i framtiden som är effektivare men som ger en motsvarande kvalitet. Klasserna som används idag och som ska användas även för den automatiska klassificeringen är: ●      Grönyta ●      Asfalt ●      Grus ●      Sand ●      Övriga hårda ytor ●      Byggnad  I projektet testades två olika metoder för automatisk klassificering: klassificeringsalgoritmer och djupinlärning. Djupinlärning har använts till att träna upp en modell till att klassificera markanvändning från ortofoton enligt de tidigare nämnda klasserna. För alla klassificeringsalgoritmer samt för träningen av djupinlärningsmodellen har samma uppsättning träningsprov använts. Noggrannhetsberäkningar har gjorts för samtliga klassificeringar. Den egentränade djupinlärningsmodellen gav en noggrannhet på 0,843284.  Klassificeringsalgoritmerna som testades var Support Vector Machine, Random Trees, K-Nearest Neighbor och Maximum Likelihood. Dessa har testats med olika uppsättningar av indata. Resultatet visade att Support Vector Machine med en normaliserad digital ytmodell (nDSM) och normaliserat vegetationsindex (NDVI) som kompletterande indata till ortofotot gav den högsta noggrannheten med en övergripande noggrannhet på 0,852886, vilket även är bättre än resultatet från djupinlärningsmodellen. Efter att det bästa resultatet identifierats generaliserades det för att få bort små felklassificerade segment. Efter generaliseringen steg den övergripande noggrannheten till 0,882682.   Klasserna grus och övriga hårda ytor stack ut med särskilt låg noggrannhet i samtliga klassificeringar. Vid jämförelse mellan den klassificering som gjorts manuellt och de klassificeringar som skapats automatiskt märks tydliga avbrott i vägnätet. Detta har identifierats bero på att träd och växtlighet skymmer marken under. Detta är ett problem som tycks svårt att lösa med nuvarande automatiska klassificeringsmetoder.

This report presents an approach to reinforce the security of the L-band Digital Aeronautical Communications System (LDACS) by developing and testing an enhanced protocol model. We have created a protocol model of MAKE, Mutual authentication and Key Exchange, based on the paper "Enhancing Cybersecurity for LDACS: a Secure and Lightweight Mutual Authentication and Key Agreement Protocol" by Suleman Khan, Gurjot Singh Gaba, Andrei Gurtov, in which the research paper addresses the security challenges inherent in LDACS. Using the open-source tool Tamarin Prover, we analysed and simulated the protocol to evaluate its effectiveness against posing threats. In this paper, our methodology involves an understanding of the MAKE protocol's architecture, identifying vulnerabilities and modeling in Tamarin Prover, to strengthen the security of LDACS. We developed two models of the protocol. The test consisted of four different lemmas and revealed partial verification of the two models, but with different outcomes. Some aspects of the model were proven to be true. Therefore, further research needs to be done to successfully validate these lemmas to ensure the robustness and reliability of the analyzed security protocol, MAKE.

No description available.

Electric vehicles play a crucial role in the global transition towards sustainability, particularly highlighted in initiatives like the European Green Deal. With projections indicating a significant increase in electric vehicle adoption worldwide, including a notable surge in the EU and Sweden, the strain on existing electric infrastructure becomes a concern. Managed charging – the process of regulating the charging of electric vehicles in a coordinated manner – emergesas a promising strategy to mitigate this strain, optimizing charging schedules to alleviate peakloads, and reduce the need for extensive grid upgrades. However, naive peak shaving approaches may fall short in addressing systemic issues, prompting the need for smarter solutions based on predictive modelling. This thesis focuses on Dansmästaren, a parking garage designed for mass electric vehicle charging, located in Uppsala, Sweden. Through load shifting techniques, one approach being explored at Dansmästaren aims to avoid grid capacity constraints by strategically scheduling EV charging to off-peak hours. This is being done using smart charging, which utilizes predictive models to predict charging durations for the scheduling of EV charging. This thesis aims to aid such predictive models by constructing a new feature for these models totrain on, namely clusters. These clusters are created using time series clustering, a technique that groups time series to clusters by running a range of algorithms comparing the similarity of different time series to each other using a variety of distance measures. In this case, the study uses data collected during three months in the form of time series, split by charging sessions, to construct the clusters. The performance of these clusters are then tested using deep learning as a predictive model to evaluate whether or not, and to which degree, the construction of clusters helped the predictive model achieve better results. Different approaches and algorithms are tested and evaluated for the time series clustering with the intention of getting the best possible performance — here meaning the specific construction of clusters resulting in the best performance increase for overall predictions. Different approaches were also tested and evaluated for the deep learning model, although not to the same extent, since the time series clustering is the focus of this thesis. In the end, a predictive performance increase of up to 17% was achieved by the predictive model using the constructed clusters as an additional feature. This suggests that time series clustering can aid deep learning models better predict charging durations.

This thesis investigates which cybersecurity strategies should be implemented in control systems to enhance cybersecurity. The study addresses three central questions, carefully designed to guide the research through its various phases and fulfill its purpose. The first question examines which cybersecurity strategies should be implemented in control systems in accordance with current requirements and established best practices. To answer this, a literature review of previous research and new cybersecurity requirements was conducted, identifying best practices for cybersecurity in control systems. The second question explores which cybersecurity strategies are currently implemented by actors in their management of products within the control system architecture. This was investigated through semi-structured interviews with five experts who deliver products within this architecture. To address the third question, which focuses on how actors within the control system architecture can streamline cybersecurity measures in their products, an analysis and comparison between best practices and current practices were conducted. This discussion revealed that a risk assessment and network segmentation should be implemented to enhance cybersecurity. Additionally, increased cybersecurity competence and the introduction of logging and monitoring of systems and components can further improve security.

Surface defects in steel slabs during continuous casting pose significant challengesfor quality control and product integrity in the steel industry. Predicting and classifyingthese defects accurately is crucial for ensuring product quality and minimizing productionlosses. This thesis investigates the effectiveness of machine learning models in predictingsurface defects of varying severity levels (ordinal classes) during the primary coolingstage of continuous casting. The study evaluates four machine learning algorithms,namely, XGBoost (main and baseline models), Decision Tree, and One-vs.-Rest SupportVector Machine (O-SVM), all trained with imbalanced defect class data. Model evaluationis conducted using a set of performance metrics, including precision, recall, F1-score,accuracy, macro-averaged Mean Absolute Error (MAE), Receiver Operating Characteristic(ROC) curves, Weighted Kappa and Ordinal Classification Index (OCI). Results indicatethat the XGBoost main model demonstrates robust performance across most evaluationmetrics, with high accuracy, precision, recall, and F1-score. Furthermore, incorporatingtemperature data from the primary cooling process inside the mold significantly enhancesthe predictive capabilities of machine learning models for defect prediction in continuouscasting. Key process parameters associated with defect formation, such as tundish temperature,casting speed, stopper rod argon pressure, and submerged entry nozzle (SEN) argonflow, are identified as significant contributors to defect severity. Feature importance andSHAP (SHapley Additive exPlanations) analysis reveal insights into the relationship betweenprocess variables and defect formation. Challenges and trade-offs, including modelcomplexity, interpretability, and computational efficiency, are discussed. Future researchdirections include further optimization and refinement of machine learning models andcollaboration with industry stakeholders to develop tailored solutions for defect predictionand quality control in continuous casting processes.

I denna uppsats introduceras applikationen ”Min kropp, mitt val” med målet att informera våldtäktsoffer om rättsprocessen samt bidra till en ökad gemenskap bland offer. En kombination av kvalitativ och kvantitativ metod används för att uppnå detta mål. Resultatet visar att det finns ett behov av en applikation för att tillhandahålla information eftersom det finns en kunskapsbrist i hur man anmäler ett sexualbrott och hur rättsprocessen ser ut. Utifrån det behovet skapades en prototyp av en applikation baserad på den kunskap som samlades in från intervjuer. Information om att det är viktigt att anmäla brott så tidigt som möjligt för att öka chanserna för en fällande dom är ett exempel på vad applikationen innehåller. Applikationen är välbehövlig eftersom den har utformats utifrån det behov som finns och kan utvecklas för att bredda målgruppen ytterligare.

The investigation to understand the subsurface and its features has long been asubject of interest for various fields, including fields such as archaeology and infras-tructure projects. However, traditional excavation methods are often costly and time-consuming. In their place, alternative techniques such as borehole drilling, which isitself expensive, and ground-penetrating radar (GPR), which produces a good butdistorted image, have gained popularity. Nonetheless, the limitations of each methodimpede them from meeting the requirements of subsurface exploration. This Mas-ter’s thesis introduces an approach combining these two methods to overcome theirlimitations and enhance their accuracy to understand the subsurface.This thesis aims to demonstrate the feasibility and effectiveness of integratingborehole drilling and GPR for subsurface exploration. Specifically, the integrationof borehole with GPR-profiles will be examined to enhance their practicality andaccuracy, meaning that this thesis will investigate the utilization of borehole datato update and adjust GPR-profiles, thereby providing more precise and informativedata for further analysis.The findings of this work indicate that combining borehole drilling and GPR-profiling to improve and update the accuracy of the GPR-profiles is entirely feasibleand results in a substantially improved subsurface exploration capability. Further,the outcomes of this thesis suggest that the integrated approach can generate amore precise representation of the underground structure. Ultimately, the proposedintegration of borehole drilling and GPR-profiling presents a promising approach toenhance the accuracy and efficiency of subsurface exploration and has the potentialto be valuable in a wide range of fields.i