Chalmers Open Digital Repository
Not a member yet
26247 research outputs found
Sort by
Prediktion av hårfärg och ögonfärg från genetiska markörer inom forensisk verksamhet
No full textJust nu pågår studier om nya tekniker inom forensisk verksamhet som ska möjliggöra använd ning av DNA för att förutsäga fenotypiska egenskaper, såsom ögon- och hårfärg, från biologiskt
material som hittats på brottsplatser. Dessa prediktioner kan vara särskilt värdefulla i utred ningar där traditionell DNA-profilering inte ger tillräcklig information. I denna rapport har
data från Rättsmedicinalverket använts, bestående av sex single-nucleotide polymorphisms
(SNPs) associerade med ögonfärg och 22 SNPs associerade med hårfärg, insamlade från 85
individer. Syftet med rapporten är att utveckla en statistisk prediktionsmodell som pålitligt
kan klassificera ögon- och hårfärg baserat på genetisk information med hjälp av Markov chain
Monte Carlo (McMC)-metoder. Det utvecklades flera modeller under projektets gång men i
huvudsak användes två modeller för ögonfärger och tre modeller för hårfärger. Prediktions modellerna för ögonfärg visade mycket god förmåga att särskilja personer med blå och bruna
ögon. Däremot uppstod svårigheter vid identifiering av individer med gröna ögon eller ögon färger som låg mellan blått och brunt. För hårfärg visade modellen en styrka i att identifiera
personer med brunt hår, men hade begränsad förmåga att korrekt klassificera övriga hårfär ger, exempelvis tenderade individer med blont, rött eller svart hår att felaktigt klassificeras
som brunhåriga. Dessa resultat understryker behovet av vidare forskning med större och mer
varierade datamängder för att förbättra modellens inlärningsförmåga och precision. Tekni ken har stor potential att bidra till effektivare brottsutredningar genom att avgränsa antalet
möjliga misstänkta, men det är också viktigt att beakta de osäkerheter som är förknippade
med fenotypisk prediktion. I rapporten diskuteras faktorer som kan påverka prediktionernas
tillförlitlighet, såsom tekniska begränsningar, kosmetiska förändringar, miljöfaktorer och trau man. Sammantaget indikerar resultaten att området är lovande, men att fortsatt forskning är
nödvändig för att stärka metodens praktiska användbarhet
Designing to Bridge the Gender Gap in Micromobility
No full textDespite the popularity of micromobility services, there is a prominent gender gap in users of micromobility, where men who ride e-scooters and e-bikes greatly outnumber women. This master thesis project investigates the barriers to women’s micromobility adoption that contribute to this gap, and subsequently proposes a design solution addressing these barriers. Adopting a user-centered approach, qualitative interviews were conducted with both user and non-user women to identify and understand their safety and accessibility needs. The primary barrier to women’s adoption of micromobility solutions was found to be an interconnected network of personal safety concerns, gender stereotypes, varying travel needs, and negative preconceived notions surrounding micromobility. Taking into account these findings, solutions were ideated upon and evaluated. A high-fidelity prototype was created to exemplify how the discovered barriers may be addressed through design solutions. This resulting prototype was dubbed Training Mode, a free browser-based e-scooter training platform that addresses the participants’ needs for more transparency and instruction on how to navigate the functions of an e-scooter. Initial evaluations suggest that Training Mode succeeds in enticing hesitant non-users to engage with e-scooters, prompting future work to investigate its potential to impact on female ridership
Motorisk funktionsbedömning med radarsensorteknik och maskininlärning - med fokus på fingertappning hos patienter med Parkinsons sjukdom
No full textDetta kandidatarbete undersöker hur radarsensorteknik kan användas för att analysera motoriska funktioner hos patienter med Parkinsons sjukdom, med särskilt fokus på fingertappningstestet. Eftersom projektet inte är en del av en klinisk studie har testerna därav inte utförts på verkliga patienter med Parkinsons sjukdom. Kandidatarbetet sker i samarbete med Sahlgrenska Universitetssjukhuset och dess fysioterapeuter. En metod baserad på maskininlärningsalgoritmerna 1D-CNN 3 layer och CNN-LSTM har utvecklats. Syftet är att klassificera fingertappningstestet i enlighet med MDS-UPDRS. Genom denna metod möjliggörs en objektiv och evidensbaserad bedömning av patientens motoriska funktion, vilket är av stor vikt både för diagnostik och för monitoreringen av sjukdomens progression. Resultatet med 1D-CNN 3 layer visar en klassificeringsnoggrannhet på 74,19 % utan dataaugmentering och 92,72 % med dataaugmentering. Noggrannheten förbättras avsevärt med CNN-LSTM-algoritmen, som utan dataaugmentering uppnår 98,38 %
noggrannhet. Modellen med augmentering uppvisade en något lägre noggrannhet, men en mer stabil prestanda. En 5-fold korsvalidering av algoritmen med augmenterad data visar stabil och konsekvent god prestanda med noggrannhet mellan 96,17 % och 99,28 %. Dessa resultat tyder på att den utvecklade metoden kan identifiera och klassificera de olika kategorierna, med endast mindre fel som beror på problemets
komplexitet
Differentiated SH-SY5Y cells as a model to study amyloid-β pathology in Alzheimer’s disease
No full textAlzheimer’s disease (AD) is the most prevalent form of dementia and has been linked to misfolding and aggregation of amyloid-β (Aβ) peptides in and around neurons in the brain. Current understanding of these mechanisms and their relationship to neurodegeneration is incomplete and requires more research for ultimately developing novel therapeutic agents.
This project has developed and optimized protocols for differentiating human neuroblastoma SH-SY5Y cells to exhibit neuron-like morphologies and functional presynapses, establishing a research tool to study Aβ pathology in AD. The differentiated cells were used to investigate the uptake and intracellular accumulation of Aβ, focusing on the pathological variant Aβ(1-42). Differentiation efficiency was assessed through morphological analysis. It was determined that the most vital parts of the differentiation procedure after RA pre-differentiation was BDNF induced differentiation, alone or in combination with NGF and/or VitD3 in serum-free conditions.
Confocal imaging and flow cytometry were used to study the uptake of fluorescently labelled Aβ monomers and fibrils in both undifferentiated and differentiated cells, using FM 1-43 as a complementary membrane probe. Morphological characterization of Aβ(1-42) fibrils was done through SDS-page and atomic force microscopy. A higher uptake of Aβ(1-42) was observed in differentiated cells and Aβ(1-42) seemed to increase endocytic activity in differentiated cells. The uptake rate of Aβ(1-42) was higher in differentiated cells, however, Aβ(1-42) monomers and fibrils did not
seem to colocalize with FM 1-43 stained vesicles. Undifferentiated cells seemed to internalize FM 1-43 stained vesicles to a higher extent, suggesting higher activity of endocytic pathways involving FM 1-43 stained vesicles. Differentiated cells showed minimal dextran uptake in comparison to undifferentiated cells, indicating alterations in endocytic pathways post-differentiation.
Differentiated cells were also used to study Acetylcholine release using an enzyme-based amperometric biosensor. The established differentiation protocol showed improved Acetylcholine release compared to previously tested protocols. For future experiments, this cell model could be used to investigate how Aβ(1-42) affects Acetylcholine release. This project resulted in the establishment of a neuron-like cell model useful in research related to AD, hopefully contributing to new understanding of the pathological mechanisms linked to the disease
Formalization of Opaque Definitions for a Dependent Type Theory
No full textDefinitions allow for the efficient reuse of code, but the process of unfolding complex nested definitions may cause usability issues for the programmer. Opaque definitions help mitigate this by restricting the unfolding of definitions at type-checking time, but their implementation demands some subtlety—for instance, subject reduction may be lost if an opaque type definition doesn’t unfold in exactly the right way. Despite this, there has been relatively little work on formally characterizing opaque definitions. We contribute here a formalization of opaque top-level definitions based on their implementation in the Agda programming language. The formalization is fully mechanized in Agda as an extension of the prior graded-type-theory project. We give typing and reduction rules for the definitions, then show, through a Kripke logical relation argument, that they enjoy many of the usual desirable type-theoretic properties: subject reduction, normalization, consistency, decidability of conversion, and so on
Frysseparation av slam och sediment: En studie av Effektivitet och Materialoptimering
No full textThis report investigates and compares three different methods for sludge drying: passive air drying, active air drying, and freeze separation using Siccum’s technology. The aim was to
evaluate each method based on energy consumption, drying efficiency, processing capacity, and potential environmental benefits. The evaluation was conducted through a comparative
analysis of each method’s technical and operational performance under controlled conditions.
The results show that passive air drying has the lowest energy demand but requires extensive space and time. Active air drying offers faster results at the cost of higher energy usage.
Siccum’s freeze separation technology demonstrated the highest energy consumption in the study, but also the greatest ability to process large material volumes through a closed and
automated system.
Although energy-intensive, freeze separation may be advantageous in large-scale operations where high throughput, controlled processing, and reduced waste volumes are prioritized.
The choice of method should be based on specific operational needs, considering factors such as energy costs, available space, processing time, and environmental impact
Embedded control firmware optimization for power electronics
No full textModern power converters face quicker input and load changes. With higher switching
frequency and smaller inductors or capacitors, there is less stored energy to
smooth disturbances. If the control reacts slowly, voltage or current overshoots or
undershoots will take longer to settle, resulting in energy waste and potential damage
to the device. Therefore, a faster response is needed in the power electronics system.
Embedded control firmware plays a key role in improving closed-loop speed and
system stability. In high-frequency DC-DC converters and automotive power electronics,
firmware execution performance directly affects control accuracy, energy efficiency,
and system robustness. In this thesis, we compare three automotive-grade
MCUs—TI F29H85x, TI AM263x, and Infineon AURIX TC4x—under a unified
closed-loop control framework. By dividing the control loop into stages such as ADC
sampling, PID calculation, and PWM output, and by analyzing differences in interrupt
systems, CPU architecture, peripheral interconnect, and compiler optimization,
we systematically show their impact on execution delay. Delay is measured using
GPIO toggling and interrupt timestamps, and platform-specific optimizations (such
as DMA acceleration, early interrupt mode, memory mapping, compiler tuning, and
CDSP/PPU offloading) are applied to explore the shortest possible execution time.
Results show that all three MCUs achieved significant improvements over their baselines,
with F29H85x reaching 710 ns, AM263x 793 ns, and TC4x 750 ns.
The contribution of this project is not only to compare real-time performance across
MCUs, but also to propose a unified cross-platform analysis method. By linking experimental
results with structural differences, we show how interrupt paths, CPU
pipelines, and peripheral interconnects determine real-time performance. This approach
goes beyond single-platform studies, providing a systematic framework for
analysis. It also offers practical guidance for MCU selection and firmware optimization
in industrial applications
4D Radar-Camera Fusion for Enhanced Point Cloud Clustering
No full textAbstract
Detecting and tracking objects in the environment is a central task in radar perception systems. However, due to the radar’s relatively low resolution, limited semantic information, and sensor-specific artifacts such as multipath reflections, a perception framework relying solely on radar data is likely to deliver limited performance. By fusing the radar data with information from a complementary sensor, such as a camera, the additional semantic information can mitigate these issues. Recent contributions to the field of radar-camera fusion focus mainly on deep learning methods, which perform well but require large amounts of annotated data. Collecting and labeling such data can be infeasible because of time or budget constraints.
This thesis instead explores an approach to 4D radar-camera fusion utilizing pretrained image models, with the goal of achieving better tracking performance. The suggested method projects a radar point cloud onto a corresponding image, associates radar points with objects detected in the image, and adds this information to the point cloud. The added information is used to distinguish points that are part of objects from those that are not. It is also used when grouping the point cloud, by suggesting which points likely belong to the same object. Evaluation indicates that the suggested method improves the tracking performance compared to using radar alone. Furthermore, the method shows potential for real-time deployment in
runtime evaluations
Iconic Architecture
No full textThe ‘STAR ARCHITECTURE’ is dead, Koolhaas stated in 2014. However,
our world is hooked on images, and formed by their re-production, spurred
on by a culture of CONSUMPTION. As consequence, the image now is
the accursed share, an inevitable waste of surplus in any culture of excess.
Since long — eventually tracing back to the renaissance and the invention of
perspective — architecture has become increasingly more image, to the point of
being it foremost. Hence, there is a case to state, that even if ‘star architecture’ is
rightfully claimed dead, it DOESN’T IMPLY the DEATH of the ICONIC.
Quite opposite. In this world, defined by its image creation and consumption,
the ‘iconic’ is of paramount cultural relevance. The icon can be argued being a
highly effective type of image, for instance featuring a high “SIGNAL-to-NOISE
ratio”, coupled with a quality of power. Thus, if the iconic is a key element for
architecture’s cultural capital and currency today, why not expand our critical
knowledge and projective practice to include it as a focus. Can we revisit the
ICONIC CONCEPT to find new relevance, suitable for the current day? With
the outset in architecture being image, design concepts has been tested, following
a method of continuous translation and glitching of images. Both as metaphors
and physical entities. Resulting in part superficial, part significant, translations
from image to “new” architecture, and in part disclosing an iconic nature.
Consequently breeding “new” or recycled imagery, and the iconic. The design
speculation is set within the frame of the partly real vision of a Gothenburg
metro line network.
The thesis started in a dissecting reading process combined with own textual
essay drafting, seeking to deconstruct ideas of “the iconic”. Architectural theory
was combined with media science, semiotics, visual studies and the multi
sensory. Sets of potential design principles were pinpointed. In a second phase
the theoretical framework was partially expressed, aiming to EVOKE discussion
and SENSATION. Including tailored scents as part of an expanded visuality of
the iconic. The designs are intended to be experienced as interfaces. They may
be read as a conceit and tautology of iconic architecture — as “architecturing
imagery”
Cirkulär näringsåtervinning: Kväve och fosfor från avloppsvatten till fiskfoder genom mikroalger och bakterier
No full textThis thesis examines the implementation of fish feed production through growth of
bacterial and micro algal biomass with nutrients from wastewater at a treatment
facility. In order to investigate this, a study with supporting calculations was con ducted to determine the energy and land usage necessary for the introduction of
such a production processes. This was achieved through analysing relevant articles
and studies. Furthermore the study examines the optimal point in a wastewater tre atment facility at which such a production process should be implemented. This was
done by experimentally determining which form of nitrogen bacteria and microalgae
prefer for optimal growth.
The results of this study show that the application of bacterial growth using nutrients
from wastewater is significantly more efficient in terms of land and energy use. The
experimental results show that this type of system should be implemented after the
nitrification step in the wastewater treatment process to ensure optimal bacterial
growth.
The findings of the study highlight the future potential of using bacterial biomass
for fish feed production. Its low energy and land requirements make implementation
of bacterial biomass production a relevant focus for future research in recycling of
wastewater nutrients