Masters Theses

Permanent URI for this collectionhttps://hdl.handle.net/2292/2972

Deposit Your Thesis instructions on ResearchHub
Copyright and Access information
Third Party Copyright information (School of Graduate Studies)
Contact Us

Browse

Now showing 1 - 20 of 2229

A systematic literature review of the environmental impacts of space activities: planetary boundaries for the New Space age
(ResearchSpace@Auckland, 2024) Foo, Darcy R.; Mankelow , Cody; Dowling, Thomas
Meeting the 1.5oC global warming goal of the Paris Agreement is becoming tenuous and six of the nine planetary boundaries, a framework outlining safe limits of anthropogenic pressure on the environment, have been transgressed. A seventh will likely being crossed in the coming years. The meteoric rise of the commercial space sector, dubbed New Space, is adding fuel to the fire and is forecast to continue to grow into a $1.8T USD industry by 2035. This poses significant environmental risk to the upper atmosphere and Earth’s orbits as space activities are the only direct source of pollution in these areas. Operating with limited environmental oversight, there is a tragedy of the commons unfolding which some researchers have called an uncontrolled geoengineering experiment. To understand the scope and scale of the environmental impacts of the New Space economy, an iterative systematic literature review following the ROSES protocol was undertaken. Assessing 4254 papers published over 60 years it was established that there are environmental impacts which can be assessed using the planetary boundaries framework. Specifically, the boundaries most at risk from New Space activity are climate change, stratospheric ozone depletion, and atmospheric aerosol loading. Using the planetary boundaries framework, the orbital environment was also considered due to its increasing congestion. 252 papers were given critical appraisal and 21 were included in a qualitative synthesis of the impacts against each discussed boundary, Earth’s orbits, and the regulatory environment governing space activities. The findings of this review indicate that space technology has progressed much faster than research into its environmental impacts. Indeed, knowledge of these impacts has declined over time as the nature and intensity of activity has evolved. The emissions from rocket launches and object re-entry into the atmosphere may have more of an impact than previously thought and the orbital environment is becoming dangerously crowded due to mega-constellations. Crucially, the planetary boundaries which are furthest from transgression, stratospheric ozone depletion and atmospheric aerosol loading, are under threat from New Space activity. International collaboration between industry and governments will be required. Regulatory overhaul and further research into the environmental and socio-cultural impacts of New Space activity is urgently needed.
Edge AI Application and Optimization: FPGA-Accelerated Railway Damage Detection and Progressive Quantization Framework
(ResearchSpace@Auckland, 2025) Yan, Dongwei; Ma, Sean Longyu
The increasing demand for real-time, low-power artificial intelligence (AI) applications has driven the deployment of deep neural networks on resource-constrained edge devices, such as Field-Programmable Gate Arrays (FPGAs). This thesis focuses on developing an efficient railway damage detection system using FPGA-based hardware and proposes a novel quantization framework to address the limitations posed by hardware constraints. Chapter 1 presents the design and implementation of a railway track detection system, which comprises four main components: an enhanced railway damage image dataset, an FPGA integrated with a convolutional neural network (CNN), a host computer for interaction and visualization, and an intelligent vehicle platform. The intelligent vehicle captures real-time images of railway tracks using a gimbal-mounted camera and transmits them to the FPGA for damage detection. The detection results, including track status and precise damage location, are wirelessly sent to the host computer for visualization. Although the system achieves real-time performance and low power consumption, aggressive model quantization is required to fit the FPGA’s resource limitations, which results in a reduction in accuracy. To mitigate this issue, Chapter 2 introduces a progressive quantization framework that incrementally reduces the bit-width of weights and activations while preserving model accuracy. The framework incorporates advanced techniques, including the Straight-Through Estimator (STE), Arctangent Soft Round (ASR), Minimize Discretization Error (MDE), and a filtering mechanism. It is evaluated on six neural network architectures—ResNet18, ResNet20, VGG7, VGG16, MobileNetV2, and ShuffleNetV2—using CIFAR-10 and Tiny ImageNet datasets. Experimental results demonstrate that the proposed approach significantly improves quantized model accuracy, achieving near-full-precision performance in most cases. Future research will focus on integrating the progressive quantization framework into the railway damage detection system, with the goal of balancing resource efficiency and high detection accuracy. This work provides a practical and scalable solution for deploying AI-based monitoring systems on edge platforms, contributing to real-time edge AI applications in critical infrastructure maintenance.
Along the Arahura
(ResearchSpace@Auckland, 2025) MacPherson, Bradley; Barton, Chris
Human-made climate change, population growth, and the expansion of primary industries have dramatically affected New Zealand's natural environment since the arrival of European settlers. Further to this climate change threatens human life across the planet putting pressure on countries to reduce emissions to prevent catastrophic rises in temperature. This thesis explores how future loss of land also threatens personal and collective identity. Identity; being inherently relational, relies on the external world to define itself, highlighting the existential threat to identity posed by destruction of land. The link between land and identity is integral within a Māori world view, land is seen as being intertwined with whakapapa(genealogy) and tribal history. Kai Tahu academic Hana O'Regan, in the book Ko Tahu, Ko Au explains: "On a spiritual level the land was carved with the histories of the iwi, thus hosting not only the living but also the past, the tupuna who had gone before the tribal atua that guarded the rohe." This link between identity and land is also present in 20th Century Modernist New Zealand Art, though rather than seeing land and individual as intertwined, artists created a fictional dialogue with landscape by applying ideas upon nature. Art historian Francis Pound critiques this lens, explaining that in viewing the landscape as a blank canvas, artists ignored existing life and culture already present within Aotearoa; Pound refers to this fiction as a "noisy silence." This research aims to engage with art and literature through a critical lens by focusing on, where identity is uncovered by viewing the land as sacred and sublime. These shared cultural links between land and identity are explored through model making and hand drawing, culminating in a covered bridge and chapel located on the Arahura River, retelling the story of Kai Tahu's discovery of Pounamu. The project aims to highlight the threat to identity from human-created environmental erasure.
Bacterial Extracellular Vesicles: a new target for a diagnostic device
(ResearchSpace@Auckland, 2024) Al-Mohammedawi, Maysaa; Swift, Simon; Dauros Singorenko, Priscila
Extracellular vesicles are produced by all the species across the domains of life, suggesting that vesiculation represents a fundamental principle of living matter, and over the years, the study of bacterial extracellular vesicles (bEVs) has gained more attention regarding their roles in bacterial survival, interactions with other cells and pathogenesis. The first aim of my study was to isolate and characterize bEVs from Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus). I found that using Size Exclusion Chromatography (SEC) technique, the bEVs at fractions 2-5 gave highest particle counts enriched with protein contents for E. coli and S. aureus during characterization with Nanosight Tracking Analysis (NTA) with size mean of 50-200 nm. Moreover, the structure of bEVs was confirmed with Transmission Electron Microscopical (TEM) showing regular spherical double membranes structures of a size range 50-100nm. Western blotting with Anti-LPS antibody detected LPS bands in both cell lysate and EVs of E. coli UPEC 536. Results of the dissociation constant (Kd) of different concentrations of single strand DNA aptamer (GN6-Cy5) aptamer achieved binding affinity towards E. coli DH5α, Nissle 1917, MG 1655 and UPEC 536 at 223.4, 325.5, 320.2 and 300.4 nM, respectively based on detecting the fluorescence intensity. This was confirmed by Stimulated Emission Depletion (STED) Microscopy during examining the GN6 aptamer and some red spots were detected around the surfaces of E. coli UPEC 536 bacteria. Recently a new dye was introduced to our Lab for labelling the bEVs called Acoerela (Aco), I applied the Aco-600 to dye the UPEC 536 EVs. Despite optimizing the ideal concentration of less to nil detected noise in the system background for theAco-600, UPEC 653 EVs and GN6-Cy5 aptamer, results of ZetaVeiw fluorescence analysis using 488nm laser didn’t detect any signal for labeling UPEC 536 EVs with the Aco-600, neither during incubating the UPEC 536 with GN6-Cy5 aptamer. The STED images for labelling UPEC 536-EVs with Aco-600 dye showed orange-stained spherical structures of EVs of a size around 100-150 nm. While no co-localization was detected from STED nor confocal microscopy images after incubation of the labelled UPEC 536 EVs Aco-600 mixture with the GN6-Cy5 aptamer.
Tailored to Me: Adapting Preference Assessments to Make Them More Acceptable and Effective for People Living with Dementia
(ResearchSpace@Auckland, 2024) Bigwood, Lily; Sharp, Rebecca
For individuals living with dementia, meaningful engagement with preferred activities is a vital component in maintaining quality of life. It is important to provide individuals with dementia effective methods to support their decision making, to ensure they have access to preferred activities and that their environment is tailored to them. Preference assessments have been found to be effective for people living with dementia. However, most evidence based behavioural assessments, such as preference assessments, have been predominantly validated for individuals with intellectual or developmental disabilities. Therefore, there are likely aspects of the assessment that need to be adapted to better suit the needs of adults living with dementia. The current study aimed to explicitly measure aspects of the assessment that need to be taken into consideration. We asked behavioural practitioners to identify practice issues in conducting preference assessments with people living with dementia, and then empirically evaluated the standard and the adapted versions in an alternating treatments design. Adaptations were also designed with input from our participants with dementia and their carers. Our measures included, choice and engagement, direct observation of indices of affect and verbal behaviour, and indirect social validity questionnaires. Our results varied across participants, however, based on common themes, we were able to develop some prompts for professionals conducting these assessments in the future. These themes varied from changing the language used, incorporating social attention, and creating a more natural environment, to align more with the social rules common amongst this population. Through incorporating these adaptations into the assessment, we observed increases in social validity measures of happiness, as well as a decrease in confusion levels amongst some participants. Our findings support that determining what is going to be more acceptable or effective for someone is completely variable, however, our study provides a helpful guide of social rules and environmental contingencies that may make preference assessments more valuable for people with dementia.
Creativity and Imagery: A Study of the Relationship between Creativity and Sensory Imagery
(ResearchSpace@Auckland, 2024) Maliekkal, Magna Christopher; Lambert, Anthony J
Theorists in psychology have long argued that mental imagery plays a pivotal role in the creative process. This study focused on sensory imagery - visual imagery, and two dimensions of auditory imagery: the mind’s ear and inner voice and its association with creativity. Considering the multifaceted concept of creativity, four different domains were assessed: Creative process, creative activity domains, creativity recognition, and divergent thinking. The instruments used for assessing sensory imagery were: modified Vividness of Visual Imagery Questionnaire (Marks, 1973), Auckland Scale of Auditory Imagery & Inner Speech (ASAIIS; Schelp, Quigley-Tump, Purdy, & Lambert, 2024). Creativity was assessed using self report measures and an objective task : Creative Process Assessment Scale (Gorlich, 2023), Divergent Association Task (Olson et al. 2021), and modified Creative Achievement Questionnaire (Carson et. al. 2005). The data were collected using a Qualtrics survey. The analysis consisted of 167 participants. Four multiple regression analyses were run using SPSS software to find: 1. Associations between dimensions of sensory imagery and creative process assessment, 2. Associations between dimensions of sensory imagery and creative activity domains, 3. Associations between dimensions of sensory imagery and creativity recognition and finally, 4. Associations between dimensions of sensory imagery and divergent thinking. The results of the analyses revealed interesting results. The first three analyses' results indicated associations between sensory imagery dimensions and: creative process, creative activities domains and creative recognition. Further observation on individual predictor variables showed no significant relationship between visual imagery and any of the domains. The central finding of the study is the significant relationship of inner voice with three domains of creativity: creative process, creative activities domain, and creative recognition. The study also found a significant relationship between both dimensions of auditory imagery i.e. mind’s ear and inner voice on creative process assessment even though there was no significant indication of the relationship between mind’s ear and creative activities domains and creativity achievements. There was no evidence of a significant association between sensory imagery and divergent association tasks. The results highlight the role of auditory imagery on self-reported creativity.
Pokémon GeO: Queer and Disabled Narratives in an Unintentional (P)GIS
(ResearchSpace@Auckland, 2025) Peek, Amber; Sila-Nowicka, Katarzyna; Martin, Michael
Pokémon GO has been a quintessential title in the location-based game (LBG) genre since its release in 2016. In the time since its release, there have been many changes to the game, however perhaps none more significant than the move to volunteered geographic information (VGI). Pokémon GO uses VGI through OpenStreetMap (OSM) environment data and the Niantic Wayfarer community suggestion/ratification platform for its points of interest, known as wayspots. The Wayfarer system allows Pokémon GO (and its predecessor, Ingress) players to define the most significant aspects of their real-world locales and bring them into the game world. As such, PokémonGOcan be understood to be an unintentional participatory GIS (PGIS) – built with the purpose of being an entertaining video game but creating VGI as a byproduct. The presence or absence of certain cultural experiences and narratives in the Pokémon GO wayspots presents players with an image of the area they find themselves in that is almost exclusively defined by players from that area. These narratives are investigated using Wellington, New Zealand as a case study. Despite having rich history in both areas, Pokémon GO is found to be lacking in wayspots corresponding to queerness and disability in Wellington. The broad absence of these themes is shown to be a combination of real-life biases making them less visible in the real world and part of a broader issue of diversity in video games in general.
The impact of storm sequences on coastal erosion: a Case Study on Tairua Beach, New Zealand
(ResearchSpace@Auckland, 2024) Wood, Sophie Ellen Rose; Coco, Giovanni
Under climate change, storms are projected to increase in frequency and magnitude, leading to a possible increase in the occurrence of sequences of storms. This thesis aims to investigate the impact of storms and storm sequences on shoreline change and to understand which storm characteristics “drive” shoreline change. Using Tairua Beach as a case study, hindcast wave characteristics were used to identify and analyse individual storms and storm sequences. Linear regressions were used to test the relationships between shoreline change and wave characteristics, revealing that storm power had a statistically significant relationship to shoreline change during individual storms. This relationship was also reflected in storm sequences; furthermore, a statistically significant relationship between initial shoreline position and the tidal range of the first storm and shoreline change was also seen for storm sequences. Additionally, a multiple linear regression showed that shoreline change during a storm sequence can be predicted based on the initial shoreline position and the sum of storm power. Findings also revealed temporal elements to the occurrence of storm sequences: seasonal variability and phases of the El Niño Southern Oscillation (ENSO). Results show that the most intense storm sequences are more likely to occur during the La Niña phase of ENSO. These insights provide a further understanding of how storm and storm sequences influence shoreline change and the drivers behind these changes. The findings highlight the importance of including the impacts of storm sequences in future coastal management strategies, especially considering current climate projections.
Understanding Household Climate Priorities
(ResearchSpace@Auckland, 2024-12-16) Austin, Hayden; Atkinson, Quentin
With the 2030 emission reduction deadline that represents a critical turning point in efforts to limit global temperature rise fast approaching, the urgency for climate change mitigation efforts is growing. Despite significant progress being made by the United Kingdom and other nations through decarbonisation policies, achieving the necessary reductions will require widespread behaviour change at the individual level. While public awareness of the risks associated with climate change, and intentions to reduce individual CO2 emissions have increased over the past decades, prior research has highlighted a barrier to optimised emission reduction. A perceived impact – actual impact gap exists in our understanding of common household emission reduction behaviours. The perceived CO2 reduction of low impact behaviours such as recycling, energy efficient lights, and reducing food waste are overestimated, while high impact behaviours such as reducing flights, energy efficient appliances, and switching a petrol car to an EV are underestimated. With the UK, and other nations falling behind their critical 2030 emission targets, this perceived impact – actual impact gap needs addressing. If efforts by governments and climate organisations are successful in motivating populations to take action to mitigate climate change, their emission reduction is likely to stray from what is optimum. The research in this thesis set out to retest the perceived impact – actual impact gap within an online survey of 1400 UK participants and administer 3 experimental messaging treatments in the form of knowledge boosts attempting to improve the perceived impact accuracy. The theoretical framework for the knowledge boost interventions were based on core components of Theory of Planned Behaviour, Technology Acceptance Model, and COM-B. This research successfully replicated the existence of a perceived impact – actual impact gap, however the 3 experimental conditions were not successful in reducing the perceived impact – actual impact gap. While this study failed in altering the perceptions of the participants, the case for further research in this area remains strong. Potential high return on investment behaviours suitable for future ‘nudge’ interventions based on the theoretical frameworks were highlighted. Future research to address the perceived – actual impact gap should consider strategies to overcome potential barriers such as prior held beliefs, misinformation, and careless responding. Given the limited time remaining to meet 2030 climate targets, bridging this gap is crucial for enabling effective and impactful climate action at scale.
Efficient Real-Time Railway Fault Detection Using FPGA-Based Edge AI System
(ResearchSpace@Auckland, 2025) Fu, Yulin ; Ma, Sean Longyu
Railway fault detection plays a crucial role in ensuring the safety and reliability of the railway infrastructure. In this study, we propose a real-time railway fault detection system implemented on Field-Programmable Gate Arrays (FPGAs), leveraging the efficiency of Binary Neural Networks (BNNs) for edge computing applications. BNNs, characterized by 1-bit weights and activations, offer significant advantages for low-latency, real-time monitoring tasks due to their efficient bitwise operations and minimal memory requirements. This makes them particularly well-suited for embedded systems, where low power consumption, high throughput, and real-time decision-making are essential. To address the challenge of resource utilization imbalance in FPGA-based systems, we introduce a novel DSP packing algorithm specifically designed to optimize Digital Signal Processor (DSP) usage while minimizing reliance on Lookup Table (LUT) resources, two primary resources in FPGA architecture. This approach effectively balances the usage of FPGA resources, enabling efficient deployment of the model on resource-constrained platforms. Experimental results show that our system achieves state-of-the-art detection accuracies of 90.9% and 93.4% with the RBPnet2 and RBPnet4 models, respectively. Furthermore, our FPGA-based solution outperforms conventional Central Processing Units (CPUs) and Graphics Processing Units (GPUs), achieving 82.77× and 9.64× improvements in energy efficiency, respectively. This work highlights the significant potential of FPGA-based BNNs for real-time railway fault detection, offering a robust and energy-efficient solution for edge computing applications. The high performance and low power consumption of the system demonstrate its suitability for deployment in consumer electronics and industrial IoT systems, where real-time monitoring and localized data processing are critical.
Cohesive Adaptability
(ResearchSpace@Auckland, 2024) Halliday, Imogen; Saad, Farah
The climate crisis has rapidly evolved into one of the most pressing challenges of our time. This thesis specifically addresses the growing vulnerability of housing to recurrent and devastating climate events, such as hurricanes and rising sea levels, which pose significant threats to both people and their homes. Consequently, the displacement of long-term residents is becoming an increasingly urgent issue, demanding immediate solutions. The core hypothesis revolves around the viability of “co-housing” as a socially and environmentally sustainable response to this crisis. Situated within the context of Aotearoa, New Zealand, the research aims to provide permanent cohousing solutions for individuals displaced from affected areas. The thesis commences by constructing a theoretical framework designed to elucidate the intricate relationship between cohousing, environmental considerations, and the needs of its occupants. Additionally, the research seeks to develop a flexible methodology capable of transcending geographical and architectural constraints, facilitating adaptation to diverse settings and occupant profiles. The primary objective is establishing three model cohousing sites across Aotearoa, New Zealand, each showcasing adaptable design solutions tailored to diverse topographies, climates, and communities. In a hypothetical scenario, these three sites will accommodate identical occupants to explore and compare the multifaceted interaction between cohousing, occupants and the environment in distinct ways. The design objectives encompass multiple facets. The thesis delves into environmentally sustainable construction methods, materials, and systems, aiming to minimise environmental impact and enhance climate resilience within cohousing contexts. It also explores architectural strategies that foster a symbiotic relationship between cohousing occupants and their natural surroundings. Furthermore, the study assesses ways to integrate environmentally sustainable living practices seamlessly within the cohousing framework, incorporating passive design principles and selfsufficient design approaches. Finally, the design strongly emphasises prioritising the comfort and well-being of cohousing occupants, achieved through communal areas, private retreats, and adaptable spaces that promote social interaction, collaboration, and individual privacy within the context of rehabilitation. Methodologically, this research adopts a designthrough- research approach, primarily relying on qualitative methods. Through this holistic approach, the thesis aims to offer valuable insights into the potential of cohousing as a socially and environmentally sustainable solution for climate rehabilitation in the unique context of Aotearoa, New Zealand.
Investigating the abundance of and information regarding exotic insects on native plants in New Zealand
(ResearchSpace@Auckland, 2024) McCullough, Madeleine; Ward, Darren
Exotic insects have ecological, economic, and human health impacts worldwide. In New Zealand, exotic insects have predominantly been prioritised based on economic impacts with less emphasis on ecological impacts. However, there is an ongoing need to understand potential future invading insects while also managing those that are already here. This research aims to increase understanding of the potential impacts of exotic insects on native New Zealand plants and to determine the usefulness of different record types for gathering data on exotic insects. Firstly, I sampled insects from select species of native plants at 18 Regional Parks around Auckland. I recorded environmental and disturbance variables and identified the insects collected to the lowest taxonomic level possible. Secondly, I downloaded over 86,000 records of exotic insect species in New Zealand from the Global Biodiversity Information Facility and analysed two record types: citizen science observation records (from iNaturalist) and specimen records (from digitised museum collections). I compared the differences between the two record types over taxonomic levels, temporal and geographic scales, and species body size. In the field study there was a significant increase in exotic insect abundance with increasing weed coverage; and some differences between plant species, with māhoe (Melicytus ramiflorus) having the highest total abundance and diversity of insects on average. The composition of insects was dominated by the exotic passionvine hopper (Scolypopa australis). In comparing record types, I found some differences between citizen science and specimen records, including the year of the record, the insect order, and the average body size of the species. Records from citizen scientists are focused on a smaller subset of exotic species, which is biased towards collecting larger bodied species and may indicate a lack of awareness of many other exotic species present in New Zealand. This research adds to the understanding of associations between native plants and exotic insects in New Zealand, and the potential impacts of these insects in native ecosystems. Additionally, this research provides insight into potential uses and biases of citizen science observation records and specimen records, specifically for exotic insects in New Zealand.
“It's a Whole School Approach; It's Not Just Me” Practitioner’s Views on Trauma-Informed Care Within Aotearoa New Zealand Primary and Intermediate Schools
(ResearchSpace@Auckland, 2024-03-03) Nazari, Flora; Connor, Helene
The experience of trauma can result in long-term adverse outcomes for children. It can present itself in youth as behavioural, social and cognitive difficulties. Trauma-informed care is a growing area of interest, given the benefits it offers to organisations and the clients they support. Within the education system, it has been recognised as allowing for better mental health and education outcomes for students who have experienced trauma. Despite the growing use of trauma-informed care within education systems internationally, this area is highly under-researched in the context of Aotearoa, New Zealand. This study looked at the experiences of Aotearoa intermediate and primary school counsellors and social workers regarding the impact of the implementation (or lack of implementation) of trauma-informed care approaches in their schools. Five participants from across Aotearoa were interviewed. Then, using Braun and Clarke’s (2006) thematic analysis, three themes were identified within a social constructionist epistemology: Trauma-informed care in practice, The practitioner within the school, and trauma-informed care effectiveness. Primarily, what stood out from the findings was that trauma-informed care frameworks were utilised within Aotearoa schools to differing degrees. Some schools are engaging in this effectively from the perspective of school counsellors and social workers. These practitioners felt well supported, as there was a shared language around trauma and school-wide application of trauma skills being utilised. However, there were also issues with some staff not applying trauma-informed care practice, as well as training not being consistent or maintained. This study highlighted how trauma-informed frameworks can influence the work of practitioners in schools, as well as their general well-being (i.e., vicarious trauma, increasing staff support). As part of the recommendations from this study, the importance of a trauma-informed care framework specific to Aotearoa being developed is reflected.
Wireless Power Supply for Low Power 'Smart Sediment Particle'
(ResearchSpace@Auckland, 2009) Kularatna-Abeywardana, Dulsha; Hu, AP
Studying the entrainment process in riverbeds has been an ongoing research at the Civil and Environmental Engineering department at the University of Auckland. The ‘smart sediment particle’ is a device developed over the years, to assist this study to characterise river sedimentation. It can monitor its own three dimensional motion with the aid of accelerometer and gyroscope sensors. The gathered data is processed in a microcontroller and stored for later retrieval. When needed that data is downloaded to a computer for mathematical processing to obtain the three dimensional trajectory of the device. Using this data forces acting upon it can be determined to comprehend the entrainment characteristics of the smart sediment particle. This project was to design a wireless power supply for this micro-miniaturised device while optimising its power performance. An inductive power transfer based wireless charging unit has been developed successfully to charge the smart sediment particle. A small secondary pickup coil embedded in the device, with additional circuitry, couples magnetically with the primary side to extract power from a primary coil loop and recharges the power supply module of the device. A rechargeable, hybrid power module has been designed to replace the standard alkaline battery previously used to power the smart particle. This consists of a supercapacitor array and a rechargeable lithium-ion battery as a backup source. The supercapacitors are the primary power sources and the battery will be utilised when the supercapacitors are discharged. The key feature of the system is that simple discrete components have been utilised as much as possible to simplify operation and minimise device dimensions. The power consumption of the new design is nearly half that of its predecessor. This has been achieved by changing the onboard microprocessor and sensors, and also by replacing functions of additional components with software techniques. Extensive testing on the power supply module proved that the novel hybrid power supply has been successful. However more investigations are required to optimise the supercapacitor array in terms of rated voltage, capacitance and dimensions. Further work is needed to fully optimise the design for device dimensions and better power transfer, by designing a three dimensional power transfer system, optimising the supercapacitors and redesigning the printed circuit board layout to optimise for space etc.
RISC-V customized instructions for AI-related operations
(ResearchSpace@Auckland, 2024) Liu, Louis (Dao, Lucky); Sham, Bruce
Deep-learning workloads, particularly convolutional neural networks (CNNs), rely on computationally expensive operations such as matrix multiplication and convolutions. Traditional general-purpose processors struggle to efficiently execute these workloads due to high instruction counts, memory access overhead, and suboptimal vector utilization. While dedicated AI accelerators (e.g., GPUs, TPUs, NPUs) offer superior performance, they lack flexibility and consume significant power, making them impractical for embedded AI inference and real-time applications. This thesis explores the design and evaluation of custom RISC-V instructions for AI acceleration, focusing on fused compute operations that optimize key AI tasks. By introducing custom fused instructions—such as load + multiply-accumulate (MAC), fused load-multiply, and output-stationary direct convolution operations—this work reduces instruction count and execution cycles in AI workloads while maintaining the flexibility of general-purpose computing. To evaluate the effectiveness of these custom instructions, the research employs cycle-accurate simulation using gem5 and benchmarks scalar, vectorized (RVV), and custom-instruction implementations across different CNN layers. The results show that: Custom fused instructions reduce instruction count by 10–30%, improving execution efficiency. Direct convolution custom instructions outperform naïve vectorized implementations, achieving 2×–5× speedups and Winograd convolution, which was already 1.2×–1.5× faster than im2col, saw an additional 10–20% improvement using fused operations. These findings demonstrate that custom instruction-level optimizations can bridge the performance gap between general-purpose RISC-V processors and dedicated AI accelerators, offering a scalable and power-efficient alternative for AI workloads. The extensibility of RISC-V allows for continuous AI-specific optimizations, making it an attractive platform for future low-power AI inference hardware. This research reinforces the potential of customizable, open-source hardware solutions in advancing efficient, domain-specific AI acceleration.
Impacts on Speech Features of Body Positions
(ResearchSpace@Auckland, 2024) Guan, Qing ; Watson, Catherine; Hui, Justine
Speech sounds are fundamental to human communication, and speech production allows people to express intentions. The speech can be produced in any position, such as lying, sitting, and standing. These positions have been suggested to affect the tissues and organs inside the human body, including the tissues used to produce speech, which might influence speech features, for instance, the vocal tract. The analysis of acoustic speech features in this thesis is based on recordings of speakers in three postures (lying supine, sitting, and standing) and will be carried out to investigate whether the body positions affect the speech features or not. Two databases examined in this research include the hVd isolated word database and the passage database, based on nine New Zealand English speakers and one British English speaker. The speakers are in different age ranges and genders to divide into groups, as age and gender are important factors that have been suggested to affect the speech features. The analysis focuses on the vowels, while formants are a critical element of the study. This thesis calculated and analyzed the first (F1) and second (F2) formant frequencies to address the research question. The fundamental (F0) frequency and vowel duration were also examined to explore the impact of body positions on speech features. The acoustic measurements examine the effects of the three postures. The analysis revealed that posture-related effects were observed in the speech features that have been focused on, with distinct impacts observed based on age, gender, and vowel type. Older speakers contained more variation in formant frequencies in the lying position than the sitting or standing. Significant differences in formant frequencies and duration occurred for distinct vowels across the three postures. In short, this thesis demonstrates the necessity of considering the position effect while analyzing speech and discusses the posture-related effects on the speech features in the results.
The Economic Impacts of Air Pollution and Greenhouse Gas Emissions: A Machine Learning Approach from a Global and Regional Perspective
(ResearchSpace@Auckland, 2025-03-15) Zhang, Mingzhe; Yi, Shan
This study uses machine learning methods to comprehensively analyze the multifaceted impacts of PM2.5 and greenhouse gas emissions on economic development from a global and regional perspective. Using data sets from 2008-2023 for 145 countries, we examine the nonlinear relationship and regional heterogeneity between environmental pollution and economic variables, illustrating how geographic and policy factors regulate the interactions between environmental pollutants and economic variables. By constructing global and regional models, this study highlights the unique economic impacts of pollution in different contexts, while addressing gaps in existing econometric research. These findings advance the theoretical framework of environmental economics and provide actionable, data-driven insights for designing pollution control and sustainable economic strategies that address regional differences. This interdisciplinary integration highlights the transformative potential of combining machine learning with environmental policy analysis to achieve the global Sustainable Development Goals.
Building the Plane as You Fly It: Researching Guidelines for Counselling in the Unique Context of Primary and Intermediate Schools
(ResearchSpace@Auckland, 2024) Strachan, Siân; May, Janet
This research project explored counsellors' experiences in primary and intermediate schools in Aotearoa. The intention was to share examples of best practices from the discipline of primary and intermediate school counselling that would be used as a foundation for both a proposed framework of school counselling in these specific contexts and a guide for how to do this. The perspectives of six counsellors from different settings with experience in the field were gathered via semi-structured interviews, and the data was analysed using reflexive thematic analysis. Four key themes emerged: (1) Participants share a desire to change the narrative of primary and intermediate school counselling, which has developed sporadically without any guidelines; (2) Participants sit at the crossroads of educational and health philosophies, and they possess a certainty and depth of vision for what a school counsellor’s role entails; (3) Insider knowledge revealed participants are already embedding ‘best’ practice through whanaungatanga, and they have a much deeper understanding of relationality because of their embracing of te ao Māori (holistic worldview) approaches; (4) At the core of any school counselling model to be developed for this young profession must sit an experienced and multi-relational practitioner who can work in a multidisciplinary way. The findings revealed that policies for school counselling in primary and intermediate schools need to be developed in partnership with key stakeholders and support the contention that the keys to partnership lie within te Tiriti o Waitangi. To guide a working party to develop such policies, which should include school counsellors, teachers, policymakers, whānau (families) and rangatahi (young people), a draft guide for school counselling practice is provided and other implications are discussed.
The Impact of Relaxing Music on Stress: An empirical approach highlighting the musical effects on objective and subjective stress parameters
(ResearchSpace@Auckland, 2024-12-01) Aleyas, Nettisha; Hautus, Michael; Shepherd, Daniel
The predominating influence of stress in modern times, has imposed devitalizing biological and emotional implications, recommending widely accepted and accessible coping solutions. Music, a creative medium has permeated into human culture offering various functions, including stress reduction. This study aimed to discover the relationship between music and its hold on stress through perceived and observed measures of Skin Conductance, Heart Rate, and BVP Amplitude. Also, the affective and hedonic responses to the music were recorded. Ratings from sixty-two participants were collected on exposure to six music conditions and a silence control condition. Repeated measures ANOVA and paired-sample t-tests were conducted to evaluate the effects of music conditions on both subjective and objective stress measures. The findings accentuate music as a promising technique to combat with stressful situations and implications, recommending effective and accessible management strategies. Music significantly reduces stress through its influence on both subjective perception and physiological responses. Statistical analysis revealed that participants showed increased heart rate and skin conductance after listening to Shape of You. Weightless and Silence portrayed the highest reduction in subjective stress while Bagel was assessed as the most pleasant song among the lot. This study further underscores the utility of music as a therapeutic and practical intervention, offering a foundation for future research and real-world applications in stress management.
Autonomous Systems Designed and Tested, for Aquatic Research
(ResearchSpace@Auckland, 2024-11-30) O'Brien, Reuben Alexander Patrick; Liarokapis, Minas
Monitoring aquatic environments is critical for preserving biodiversity, mitigating pollution, and addressing habitat degradation. Traditional methods, such as manned vessels and stationary monitoring stations, are often resource-intensive, costly, and unsuitable for remote or hazardous locations. The increasing demand for scalable, efficient, and autonomous solutions has driven advancements in robotic platforms tailored for environmental inspection in diverse aquatic settings. This thesis addresses these challenges by developing two innovative, autonomous robotic platforms optimized for hazardous and inaccessible environments. The first, AquaTri, is a waterjet-powered trimaran designed for bathymetry, water quality assessment, and environmental inspections across lakes, rivers, and coastal waters. AquaTri weighs under 5 kg, costs $600-$1,500 USD, and employs off-the-shelf components and 3D printing for affordability and reproducibility. Its waterjet propulsion ensures maneuverability and resilience in debris-laden shallow waters, with testing confirming stable operation at speeds up to 2 m/s and effective obstacle avoidance. The second platform, AeroBuoy, is a lightweight, drone-deployable buoy designed for environmental monitoring in narrow and difficult-to-access river systems. Weighing just 3 kg and costing $800 USD, AeroBuoy autonomously navigates downstream using river currents at speeds of 0.5 m/s while collecting environmental data such as water temperature. Its compact, 3D-printed design facilitates rapid deployment and retrieval via drones, minimizing human intervention in dangerous areas. Testing demonstrates its robust self-righting capabilities and adaptability in real-world scenarios, including validation trials at Orewa River, Auckland, New Zealand. Through a detailed comparison of these platforms, this thesis evaluates their complementary strengths, limitations, and suitability for diverse environmental monitoring applications. By advancing efficient, low-cost solutions for aquatic research, this work contributes to the broader adoption of autonomous systems for environmental preservation and paves the way for future improvements in durability, autonomy, and sensing capabilities.

Browse

Recent Submissions