This graduation project presents the design and realisation of a kinetic light installation developed in collaboration with Lumus Instruments. The objective was to explore the experiential qualities of the studio's large-scale work and to gain a deeper understanding of its artistic language. By combining these insights with my own interests, inspirations, and design identity, the project aimed to develop a new direction for a kinetic light installation, one that relates to the work of Lumus Instruments while expressing a distinct personal perspective and creative approach.

This project presents a unique design challenge at the intersection of art and design. There is a clear context for the assignment, following from the studio Lumus Instruments and the gallery setting; however, the project did not have a predefined conceptual framework or guiding narrative. The direction of the installation remained intentionally open, following a practice-based design process. Guided by the principles of creative experimentation and knowing through making, the installation gradually developed through iterative cycles of sketching, prototyping, technical development, and reflection. Physical experimentation played a central role in shaping both the conceptual narrative and the final design. This approach is positioned in relation to reflective practice, practice-based design research, and STEAM-oriented thinking.

The resulting prototype is a kinetic light installation consisting of twenty repeating elements suspended from the ceiling. Each element contains two stepper motors, a custom-designed PCB, an Arduino Nano, and a double-sided LED module supported by super-elastic wires that allow the light source to move within a two-dimensional plane. A dark diffuse panel positioned behind the array creates two complementary readings of the installation: a front view in which individual light sources form slowly evolving wave-like patterns, and a rear projection surface that visualises their movement through light and shadow. The behaviour of the installation is controlled through custom-developed software that simulates the collective behaviour of the entire array and generates the control code for the physical prototype.

Conceptually, the installation draws inspiration from the fields of architecture, mathematics, and music composition and is based on elemental repetition, where complexity emerges from the interaction of many simple and constrained motions. The design is informed by Stephen Wolfram's theory of emergence through simple rules, the repetitive and continuously evolving structure of Simeon ten Holt's *Canto Ostinato*, and Buckminster Fuller's principles of structural logic. The final installation is further interpreted through Gestalt principles and aesthetic theory to explain how its visual organisation and experiential qualities are perceived.

The intended outcome is a hypnotic, autonomous object that evolves continuously. Validation focuses on whether these intended experiential qualities are perceived by viewers by comparing the design intention with the perceived effect.
... Read more

Accessibility reflects the ease with which different individuals can overcome travel impedance to reach spatially distributed opportunities. Since accessibility of public transport networks is jointly determined by network topology and service attributes, this study applies access graphs derived from time-weighted L-space and frequency-weighted P-space graphs as the standardized framework. For previously underresearched East Asia region, a new dataset is compiled from open-source data for 61 systems and access graphs are constructed over increasing generalized travel-time budgets, and reachability and equity indicators at critical time points are benchmarked against systems in other world regions. Finally, based on the temporal evolution of average degree, East Asian and metro networks worldwide are classified into four clusters using k-means clustering. The findings show that East Asian metro systems vary widely in sizes. Influenced by the accessibility growth pattern, most medium- to mega-sized networks follow logistic (S-shaped) curves of degree growth consistent with a core–branch structure, whereas smaller or systems with degraded service have irregular degree growth curves. In terms of performance, large systems show greater spatial disparity and less uniform service quality, resulting in lower equity and reachability at 30 minutes. Regardless of size, East Asian metros tend to underperform in the medium and late stages of accessibility growth due to early decay in degree-growth rate and limited reachability at time of maximum degree growth rate. When controlling for size, East Asian metros remain less reachable than European systems and experience earlier growth-rate decay than both European and North American counterparts. Based on the timing of the peak and subsequent decay in degree growth rate, four distinctive accessibility growth patterns are identified across the worldwide systems. Clustering analysis further reveals that large East Asian networks experience an earlier decline in accessibility growth, similar to networks lacking direct cross-line connections, whereas most medium and small networks sustain growth longer on par with those in other regions. The presence of networks of similar size in different clusters suggests that improving service attributes can enhance overall accessibility performance. ... Read more

Significant wave height (𝐻𝑠) is one of the sea-state parameters on which offshore workability decisions depend. Personnel transfers are commonly restricted around 𝐻𝑠 =1.5 m, motion-compensated lifting around 2.5 m, and many operations cease above 3 m. Near these limits, even a small error in the estimated sea state can change the operational decision. For 𝐻𝑠 estimates to be useful in this setting, they therefore need to be accurate to roughly 0.25 m RMSE across the workability range where such decisions are still being made.

Several methods are already used to estimate or describe offshore wave conditions, but each has shortcomings when considered against the needs of real-time vessel operations. Numerical wave models provide useful regional context, although their resolution is too coarse to capture the local conditions
around a single vessel. Satellite altimetry can support large-scale sea-state observation, but revisit times are too long for workability decisions made from hour to hour. Buoys provide direct measurements, but only at fixed locations, while onboard physics-based radar methods depend on processing assumptions
that may break down in the same conditions where reliable estimates are most needed. None of these sources fully provides a local, real-time, and sufficiently accurate estimate at vessel scale.

This thesis examines whether deep learning can narrow that gap by estimating 𝐻𝑠 directly from operational vessel data. The dataset combines X-band radar imagery, six-degree-of-freedom vessel motion measurements, and reference 𝐻𝑠 values from open-source wave buoys and ERA5 reanalysis, drawn from three operational vessels over several years. A Vision Transformer backbone is applied to preprocessed radar images, with optional vessel-motion fusion. In the sequence variant, the model is given a short series of consecutive radar images rather than a single image, allowing it to use temporal information in the sea surface pattern. Eight model variants are trained across a four-axis ablation covering preprocessing route, backbone initialisation, motion inclusion, and single-image versus sequence-based input.

On the development vessel, the best model comfortably reaches an RMSE lower than the set target under normal wind across the operationally relevant 𝐻𝑠 ≤3 m range, with little systematic bias. The strongest within-vessel configuration is the radar-only sequence model. Adding vessel motion does not consistently improve the estimate and, in the tested configuration, tends to introduce a positive offset at low sea states. For use on the same vessel, the radar-only sequence variant is therefore the preferred model.

Cross-vessel transfer remains the main unresolved part of the problem. Blindly applying successful models from one vessel to another does not meet the operational target, although the reasons are partly identifiable. In one direction, performance is mainly limited by the available sea-state coverage, while in the other it is dominated by a vessel-specific calibration offset. Low wind is the most consistent within-vessel failure mode, which is consistent with the weaker radar wave signature expected under reduced Bragg scattering. Overall, the thesis shows that deep learning can provide vessel-scale 𝐻𝑠 estimates from X-band radar imagery with accuracy well below the defined operational target, while also showing that reliable use across vessels requires either vessel-specific adaptation or broader multi-vessel training data. ... Read more

Two-photon lithography (2PL) is a high-resolution additive manufacturing technique achieving complex three-dimensional (3D) microstructures with sub-micrometer precision. This capability has driven applications across optics, microfluidics, bioelectronics, metamaterials, and biomedical engineering. Beyond geometry, the functionality of 2PL-fabricated structures critically depends on their mechanical properties, which are influenced by resin chemistry, printing parameters, and post-processing treatments. Understanding the mechanical behavior of 2PL-fabricated structures at both macro- and micro-scales is essential for the rational design and development of advanced devices and systems. Accordingly, this review provides the reader with a concise yet comprehensive overview of the current knowledge on the mechanical properties of materials usually employed in 2PL. Common photoresist classes (e.g., hydrogel-like, elastomeric polymer networks, rigid glassy polymers, and hybrid organic–inorganic networks) offer distinct trade-offs in stiffness, with Young's modulus spanning several orders of magnitude. Mechanical performance can be further tuned via laser settings, environmental conditions, or post-processing approaches such as UV curing, pyrolysis, or incorporation of responsive chemistries for dynamic “4D” behavior. We then examine the methodologies applied—or specifically developed—to characterize 2PL materials and microstructures in terms of stiffness, toughness, and viscoelastic response at the micro- and nanoscale. Finally, we discuss the applications where mechanical properties are critical to the functionality of 2PL structures, such as tissue engineering, microfluidics, and tunable metamaterials. Looking ahead, advances in material design, adaptive characterization, and predictive modeling will enable rational, data-driven workflows. Treating mechanical properties as fundamental design parameters will be key for developing reliable microdevices for next-generation technologies. ... Read more

Magnetic resonance imaging (MRI) produces high acoustic noise levels that can reduce patient comfort, make the scanning experience intimidating, raise hearing safety concerns, and interfere with auditory functional MRI experiments. Predictive noise cancellation (PNC) aims to reduce this noise by estimating the MRI gradient-to-acoustic transfer function and generating sequence-specific anti-noise. This thesis redeveloped an existing LabVIEW/MATLAB-based PNC pipeline into a Python-based framework and evaluated redesigned calibration strategies for phantom-based MRI acoustic noise reduction.

The Python framework rebuilt the main PNC workflow interface, digital signal processing, calibration, regular sequence processing, audio recording, and arbitrary function generator control. Validation against the original implementation showed that the rebuilt system preserved the main DSP and workflow behavior, with full-workflow errors below 5% for the evaluated blocks except for one flagged regular sequence alignment case. The Python implementation also reduced computation time by approximately 73% and improved workflow usability and stability.

Calibration redesign was evaluated using chirp-based gradients, transfer function stability metrics, live calibration-stage reduction, and phantom measurements using fast field echo (FFE) and echo planar imaging (EPI) sequences. Compared with the original sl014 calibration, the chirp-based gradients provided stronger excitation and generally improved transfer function stability and live reduction. Chirp 1 and chirp 4 were the most reliable broadband calibration candidates across calibration and FFE measurements. For EPI-focused testing, combined chirp+EPI-Y transfer functions improved EPI-band cancellation relative to chirp-only calibration and reduced residual amplitudes near the dominant EPI component.

The achieved cancellation remained limited by channel imbalance, frequency-dependent reduction, low-frequency playback constraints, and acoustic behavior that was not fully described by a fixed linear time-invariant model, especially for the more complex EPI sequence. In vivo evaluation was also not performed, so the results remain limited to phantom-based testing. Overall, this thesis validated a faster and more usable Python-based PNC framework and showed that redesigned broadband and EPI-targeted calibrations can improve sequence-specific MRI acoustic noise reduction. ... Read more