Publikationer från Linköpings universitet
Not a member yet
58266 research outputs found
Sort by
Design and Evaluation of a Microstrip Resonator-Based Sensor for Dielectric Sensing
No full textResonators are simple but powerful circuits that can be used in various applications. One such application is a microstrip resonator sensor, which can be used to detect and compare substances, including liquids, gels, and ice, and can be further scaled down to detect many more materials. The initial phase of this thesis includes a literature review on the microstrip resonator sensors to gain knowledge about the recent developments related to the topic. The second phase involves designing a microstrip-based resonator sensor, which has two split-ring rectangular resonators that resonate at 1 GHz. The designed sensor is simulated using Momentum Microwave and EMPro, and the impact of varying certain parameters is studied. Once the sensor is fine-tuned, it is fabricated on a PCB. The third phase of the thesis is the evaluation of the sensor. As liquids are used to evaluate, a few hollow cubes are 3D printed for better sensing. A few liquid samples are used to evaluate the performance of the sensor, where they are compared with an empty cube, a cube filled with the same liquid, and a cube filled with another liquid. Volume-based tests are also performed to evaluate the sensors’ performance when the volume of the sample varies. This will expand the scope for further development in the field of sensing and resonator-based sensors
Solution-down procedural content generation via reinforcement learning for hyper-casual puzzle games using data-driven difficulty estimation
No full textProcedural content generation via reinforcement learning (PCGRL) has shown promising results in generating solvable levels for puzzle games. Still, little work has been done in exploring its ability to create engaging levels. In this work, we investigate the use of PCGRLinconjunction with difficulty estimation to create engaging levels by making them more challenging. We also propose a novel process for generating levels, called solutiondown generation, that guarantees solvability, allowing the model to focus on learning diff iculty rather than solvability. To test and compare these methods, we implemented four generators: one that uses random selection and a puzzle-up process, another that also uses random selection but our solution-down process, a third that is based on the original Turtle PCGRL method, and one that uses our proposed solution-down PCGRL method (SDPCGRL). All the generators were developed to generate levels for the hyper-casual puzzle game Longcat, created by Fancade. In this game, the player attempts to fill different boxes with an elongating cat. Its hyper-casual nature allowed it to have a manageable number of states and actions, and its popularity provided us with a substantial amount of data. Wefirst implemented a difficulty predictor to evaluate and train the models for generating engaging and challenging puzzles. This predictor is based on a previous method for estimating difficulty in casual puzzle games and was trained on player data collected from the Longcat application. We conducted an analysis phase before developing the generator to use this data properly. After the generators were developed and/or trained, we tested them by having each generate 20,000 levels. We then compared the solvability, number of duplicates, and difficulty of these levels. We found that both puzzle-up approaches struggled significantly with generating solvable levels, with the Turtle PCGRL generator unable to generate any solvable levels. We found that the solution-down random generator had fewer duplicates than the SD-PCGRL method; however, the vast majority of these levels were of trivial difficulty. In conclusion, the SD-PCGRL generator showed the best result in creating solvable, unique, and challenging levels for Longcat
Scripts of Credibility: Discursive Framings and Stereotypes in Queer Asylum Adjudication in Greece : A Critical Reading through Intersectionality, Temporality, and Orientalism Theory
No full textThis study investigates the role of stereotypes in the refugee status determination procedure for LGBTQI+ applicants with SOGI-based asylum claims before the Greek Asylum Service. Employing a qualitative methodology grounded in Critical Discourse Analysis, it examines two first-instance asylum decisions to explore how reasoning is constructed in the adjudication of such claims. This approach enables a close examination of assumptions, language and power relations depicted in legal reasoning. The analysis is informed by queer theory, intersectionality, temporality and orientalism. These theoretical frameworks provide a critical lens for investigating how normative assumptions around gender, sexuality, and cultural background shape credibility assessment. The findings reveal that normative assumptions and discursive techniques are applied to construct the “ideal” LGBTQI+ asylum seeker, systematically excluding non-conforming performances of sexuality and gender. In light of these findings, the study proposes systemic reforms for the purpose of a fairer asylum process, such as the revision of national guidelines, the adoption of a trauma-informed assessment model, and the incorporation of COI reports that account for diverse SOGI experiences
AI integration in Product Development : Development of AI integrated workflows through an automotive case-study
No full textThis master thesis explores how Artificial Intelligence (AI) can be meaningfully integrated into the product development process at Volvo Group Advanced Design. As AI tools become increasingly available and powerful, their potential to enhance efficiency, creativity, and decision-making in design workflows grows. However, AI adoption remains limited by a lack of structured, designer-centered workflows tailored to real-world industrial needs. The objective of this study is to investigate current design practices at Advanced Design, identify pain points and opportunities, benchmark emerging AI tools, and propose AI-augmented workflows that are both effective and engaging for designers. The research employs a combination of a literature review, semi-structured interviews with designers at Advanced Design, and tool benchmarking based on defined evaluation criteria. From this foundation, five AI workflow concepts were developed. Two workflows; User Insights Miner and User Interaction Visualizer, were further implemented and evaluated through a future-oriented autonomous mobility case-study. Evaluation methods included tool performance testing and reflective design journaling to assess their practical impact. Findings show that AI offers most value in the diverging stages of the Double Diamond process, particularly in user research synthesis, data-driven ideation, and concept visualization, without undermining the creative and intuitive aspects that designers value. The proposed workflows improved both efficiency and creativity, while emphasizing the necessity of human oversight and expertise. This thesis contributes a validated framework for aligning AI capabilities with industrial product development needs and offers strategic insights for fostering sustainability and stimulating human-AI collaboration in design.Detta examensarbete undersöker hur artificiell intelligens (AI) meningsfullt kan integreras i produktutvecklingsprocessen vid Volvo Group Advanced Design. I takt med att AI-verktyg blir allt mer tillgängliga och kraftfulla, ökar deras potential att förbättra effektivitet, kreativitet och beslutsfattande i designarbete. Trots detta förblir AI-användningen begränsad på grund av bristen på strukturerade, designercentrerade arbetsflöden anpassade till verkliga industriella behov. Syftet med studien är att kartlägga nuvarande designprocess på Advanced Design, identifiera behov och möjligheter, utvärdera framväxande AI-verktyg, samt föreslå AI-stödda arbetsflöden som är både effektiva och engagerande för designers. Arbetet baseras på en kombination av litteraturstudie, semistrukturerade intervjuer med designers på Advanced Design, samt benchmarking av AI-verktyg enligt fastställda utvärderingskriterier. Utifrån detta utvecklades fem koncept för AI-baserade arbetsflöden. Två av dessa; User Insights Miner och User Interaction Visualizer, implementerades och utvärderades genom ett framtidsinriktat case med autonom mobilitet. Utvärderingen genomfördes via prestandatester av verktygen samt en designjournal för reflekterande analys. Resultaten visar att AI har mest värde i det divergenta faserna av Double Diamond-processen, särskilt vid syntes av användarinsikter, datadriven idégenerering och konceptvisualisering, utan att kompromissa med de kreativa och intuitiva aspekter som designers värdesätter. De föreslagna arbetsflödena förbättrade både effektivitet och kreativt stöd, samtidigt som de underströk vikten av mänsklig expertis och översyn. Studien bidrar med ett validerat ramverk för att anpassa AI:s kapabiliteter till industriell produktutveckling och erbjuder strategiska insikter för att främja en hållbar och stimulerande samverkan mellan människa och AI inom design
Aesthetic Ageing and Its Impact on User Experience : Exploring the Role of Aesthetic Ageing in User Experience. A Study of a Novel PLA/SCG Biocomposite
No full textThis thesis explores how aesthetic ageing affects user perception and experience of materials, with a particular focus on a novel biocomposite made of polylactic acid (PLA) and 30 wt.% spent coffee grounds (SCG). Through a material-driven design (MDD) methodology, the study investigates how environmental factors, specifically moisture, UV exposure, and elevated temperatures, influence the visual and tactile qualities of PLA and PLA/SCG 30 wt.% materials over time. A combination of quantitative measurements (profilometry, balance, and colour analysis) and qualitative insights from a user-centred workshop was used to assess how these changes are perceived by users. Results show that PLA remains relatively stable under environmental stress, while PLA/SCG 30 wt.% is more responsive, exhibiting greater surface variation and perceptual shifts. Notably, the PLA/SCG 30 wt.% samples were often described as “cheap” or “ordinary,” which, rather than being negative, suggests a potential for discreet integration into low-profile plastic applications without disrupting user expectations. The study concludes with a material vision that frames PLA/SCG 30 wt.% as a subtle but effective alternative to disposable plastic in everyday contexts, particularly where sustainability is desired without drawing overt attention to itself. In this thesis, generative AI has been utilised to give suggestions regarding sentence structure as well as general thesis structure. AI has also been used to summarise large bodies of text, particularly in discussion chapter. The model used was ChatGPT, GPT-4.5-turbo. All text influenced by AI has been carefully scrutinised to minimise discrepancies. Results were not affected by this, since the AI was not allowed to draw any conclusions or provide any sources without being scrutinised by the authours
Spectral Computed Tomography of the Injured Knee
No full textComputed tomography (CT) is widely used for fracture detection and postsurgical evaluation. Recent advancements have opened new possibilities, particularly in spectral CT, which utilises the energy-dependant attenuation properties of tissues to enable tissue differentiation and enhance diagnostic capabilities. The two primary techniques employed in spectral CT are Dual-energy CT (DECT) and, more recently, photon-counting detector CT (PCD-CT). The overall aim of this thesis is to investigate the use of spectral CT for improved imaging of the injured knee. In Study I, the diagnostic accuracy of DECT for detecting bone marrow lesions (BMLs) in patients with suspected anterior cruciate ligament (ACL) injury was assessed. BMLs, commonly observed on MRI as a result of bone contusions, provide insights into trauma mechanisms and associated injuries. In standard CT, the attenuation differences corresponding to these lesions are obscured by overlying bone. DECT overcomes this, generating virtual non-calcium (VNCa) images as colour overlays on the CT images. DECT knee images from 48 patients were evaluated for the presence and size of BMLs by two readers. Consensus MRI served as the reference standard. The diagnostic accuracy was moderate with 70.1 % (95% CI 63.3-76.2) and 69.1% (95% CI 61.7-75.9) overall sensitivity and specificity, moderate to substantial intrareader agreement and fair to moderate interreader agreement. Lesion volumes were comparable in the femur but underestimated in the tibia. Study II evaluated the diagnostic accuracy of DECT and MRI for detecting ACL injuries. The ACL, being an important stabilising structure of the knee, is typically imaged using MRI. The diagnostic performance of DECT and MRI was evaluated, with arthroscopy serving as the reference standard. From the cohort in Study I, arthroscopy data was available for 21 patients who were included in the analysis. The combined sensitivity across all four readings for DECT was 76.3 % (95% CI 66.8-85.9), which was lower than that of MRI at 86.8% (95% CI 71.9-95.6). Both methods demonstrated high positive predictive values: 93.6% (95% CI 84.3-98.2) for DECT and 91.7% (95% CI 77.5-98.3) for MRI. In Study III, a PCD-CT prototype was evaluated using a bovine knee specimen with an induced fracture and metallic osteosynthesis material. Virtual monoenergetic images (VMI) generated by the PCD-CT at 120 kV and 140 kV tube potentials were compared to conventional energy-integrating detector CT (EID-CT) images acquired with dual- and single-energy, both with and without iterative metal artefact reduction algorithm (iMAR). Four readers assessed image quality with respect to metal artefacts. PCD-CT 140 kV VMI at 150 keV received the highest scores and was significantly superior to single-energy Sn150 kV images with iMAR. While DECT iMAR images scored lower than PCD-CT images, the difference was not statistically significant. Image quality scores were also significantly lower for 120 kV tube potential and lower VMI values at 72 keV. Study IV explored the metal artefact reduction capabilities of a commercially available PCD-CT in twelve patients with surgically treated tibial plateau fractures. PCD-CT 140 kV ultra-high resolution (UHR) scans were reconstructed as VMI at 70, 110 and 150 keV, using smooth and sharp kernels, with and without iMAR, as well as polychromatic images with an extra-sharp kernel. EID-CT Sn150 kV scans were reconstructed with smooth and sharp kernels, with and without iMAR. Five readers assessed the image quality. EID-CT iMAR images scored significantly higher for reducing metal artefact streaks, whereas bone-metal interface visualisation was comparable between PCD-CT high VMI and EID-CT with IMAR. PCD-CT with an extra sharp kernel was superior for bone visualisation.Datortomografi (DT) en allmänt använd metod för frakturdiagnostik och för postoperativ utvärdering. De senaste framstegen inom DT-tekniken har öppnat nya möjligheter, särskilt inom spektral DT, där vävnadernas energiberoende attenueringsegenskaper nyttjas för att förbättrad vävnadsurskiljning och skapa nya diagnostiska möjligheter. Spektral DT kan baseras på två huvudsakliga tekniker: dubbelenergi-DT (dual-energy computed tomography (DECT)) och den nyare fotonräknande datortomografen (photon-counting detector computed tomography (PCD-CT)). Det övergripande syftet med denna avhandling var att undersökta användningen av spektral DT för förbättrad bilddiagnostik av knäskador. I Studie I utvärderades den diagnostiska noggrannheten för DECT vid detektion av benmärgslesioner (BML) hos patienter med misstänkt främre korsbandsskada. BML, ett vanligt fynd vid MRT efter benkontusioner, kan ge information om traumamekanism och associerade skador. På konventionella DT-bilder döljs ofta de små attenueringsskillnaderna, associerade med BML, av de mellanliggande bentrabeklerna, men med DECT kan virtuella bilder utan calcium skapas och förbättra synligheten av BML. DECT-bilder av knän från 48 patienter utvärderades av två bedömare avseende förekomst av och storlek på BML. Konsensusbedömningen av MRT bilderna utgjorde referensmetod. Den diagnostisk tillförlitligheten var måttlig med övergripande sensitivitet och specificitet på 70,1% (95% (KI) 63,3–76,2) respektive 69,1% (95% KI 61,7–75,9), måttlig till betydande intrabedömaröverensstämmelse och skaplig till måttlig överensstämmelse mellan bedömare. Den uppmätta volymen för lesioner i femur skilde sig marginellt åt mellan DECT och MRT, medan volymen underskattades med DECT för lesioner i tibia. Studie II handlade diagnostisk tillförlitlighet avseende korsbandsskador. Det främre korsbandet, en viktig stabiliserande struktur i knäleden, avbildas vanligen med MRT. I Studie II användes artroskopi som referensmetod för att utvärdera den diagnostiska prestandan hos DECT och MRT avseende detektion av främre korsbandsskador. Från kohorten i Studie I, erhölls artroskopidata för 21 patienter, vilka kunde inkluderas i analysen. Den kombinerade sensitiviteten från alla fyra granskningarna av DECT-bilderna var 76,3 % (95% KI 66,8–85,9), vilket var lägre än för MRT, där sensitiviteten var 86,8% (95% KI 71.9–95.6). Båda metoderna uppvisade höga positiva prediktiva värden: 93,6% (95% KI 84,3–98,2) för DECT och 91,7% (95% KI 77,5–98,3) för MRT. I Studie III utvärderades en PCD-CT-prototyp med hjälp av ett knäpreparat från ko med en fraktur och metalliskt osteosyntesmaterial. Virtuellt monoenergetiska (VMI) genererade med 120 kV respektive 140 kV rörspänning jämfördes mot bilder från en konventionell DT med energiintegrerande detektor (energy-integrating detector computed tomography (EID-CT)), erhållna med dubbelenergi (DECT) respektive enkelenergi, både utan och med metallartefaktreducerande algoritm (iMAR). Fyra bedömare utvärderade bildkvaliteten avseende metallartefakter. PCD-CT 140 kV VMI på 150 keV skattades högst och skillnaden mot skattningen av enkelenergibilderna erhållna med 150 kV och tennfiltrering samt iMAR var statistiskt signifikant. DECT bilder med iMAR skattades lägre än nämnda PCD-CT-bilder, men skillnaden var inte signifikant. Även skattningen av PCD-CT-bilder erhållna med 120 kV och de som rekonstruerats som VMI 72 keV var signifikant lägre. I Studie IV undersöktes de metallartefaktreducerande förmågorna hos en kommersiellt tillgänglig PCD-CT för bilder av tolv patienter med operativt åtgärdad tibiaplatåfraktur. PCD-CT-bilder skapade med 140 kV rörspänning och ultrahög upplösning (ultra-high resolution, UHR) rekonstruerades som VMI på 70, 110 och 150 keV, med både mjuk och hård kernel, samt utan och med iMAR. Fem granskare bedömde bildkvaliteten. EID-CT med iMAR erhöll signifikant högre skattning för reduktion av metallartefaktsstråk, medan benmetall- gränsen visualiserades likartat mellan PCD-CT med hög VMI och EIDCT med iMAR. PCD-CT extra hård kernel uppvisade bäst kvalitet för visualisering av ben
Interviewing people on sensitive topics: challenges and strategies
No full textInterviewing participants on sensitive topics is challenging and requires carefully planned and executed strategies throughout the research process. Sensitive topics have the potential to cause distress to participants or researchers. In this methods paper, strategies for conducting interviews on sensitive topics are described, including choosing the appropriate interview approach, creating a safe and comfortable environment, building rapport, utilizing sensitive and open questioning, reflexivity of the researcher, and awareness of the potential therapeutic effect of the interviews on participants. The strategies are illustrated with experiences from research on sensitive topics
The construction of subjectivities and enemies in global warming: searching for the political climate issue in Swedish Election Campaigns, 1988-2014
No full textIt has been argued that the issue of global warming has become incorporated into a postpolitical condition that has deprived it of a proper political subject. In this article, the depoliticizing process is examined through a historical analysis of Swedish election campaigns, 1988 to 2014, a domain that traditionally features political language and politicizations. The analysis shows how the rhetoric of political parties was characterized by an increasingly universalizing language which made it problematic to name a political subject, and how the enemy of global warming was constructed as an outsider that threatened the current order. The article argues for the need to re-politicize the climate issue by understanding climate crises as social crises; that is, crises that can be properly politicized.Funding Agencies|Stiftelsen Riksbankens jubileumsfond (the Swedish Foundation for Humanities and Social Sciences) [P18-0121:1]</p