Over recent years, subjective well-being (SWB) has become a primary goal in urban planning, with research showing that the built environment can significantly influence residents’ well-being. This study focuses on the role of the subjective nature of aesthetic quality, which traditional segmentation-based computer vision approaches often fail to capture. To address this, we evaluate the Computer Vision-enriched Discrete Choice Model (CV-DCM) developed by Van Cranenburgh and Garrido-Valenzuela (2025), which uses a vision transformer and classifier to extract holistic visual features from Google Street View images and estimate continuous utility scores that reflect perceived visual quality, trained on stated trade-offs that people make between visual environments. We link these scores to life satisfaction and hedonic well-being measures from the Netherlands Mobility Panel (SWLS, 2020–2022; MHI-5, 2020) and analyze their relationships using Structural Equation Modeling (SEM), controlling for socio-demographic and built environment variables. Results show that PC5-level utility aligns more closely with life satisfaction than PC6, indicating that broader neighborhood context matters more than immediate street conditions. When non-linear age effects are modeled, a small but significant direct path from utility to life satisfaction emerges, whereas no significant association is found for hedonic well-being. Overall, the current explanatory power for SWB is modest and appears mainly driven by who lives where. Nevertheless, a perception-based computer vision model provides a scalable way that can quantify subjective visual quality, which could gain relevance when improved model fit is achieved by reducing variance in data collection or retraining the model on SWB-specific objectives.

Reliable autonomous recovery of Unmanned Aerial Vehicles (UAVs) on moving maritime platforms remains a critical challenge, primarily due to complex, stochastic deck motion, particularly vertical heave, and unpredictable environmental disturbances. Existing Reinforcement Learning (RL) approaches often simplify this environment, limiting their real-world applicability. This thesis investigates the robustness trade-offs of RL-based guidance controllers under realistic, high-dynamicity maritime conditions. We benchmarked a classical Proportional-IntegralDerivative (PID) controller against two RL architectures trained using Soft Actor-Critic (SAC) in a high-fidelity PyBullet simulation: a Full RL 3D controller and a novel Hybrid RL 1D controller, which strategically applies RL only to the critical, stochastic vertical (heave) axis. The results demonstrate that the Hybrid RL 1D architecture (86.6% success rate) achieved superior overall robustness and efficiency. Notably, the RL controllers dramatically reduced average landing time (RL_1D: 3.31 s vs. Baseline: 11.51 s), though the classical PID baseline maintained higher horizontal precision (Err𝑋𝑌 of 0.17 ± 0.17 m ). The Hybrid RL 1D maintained a superior success rate up to 89% in high sea states (SS7) and exhibited greater resilience to sensor noise. However, a critical limitation was identified: both RL-based policies experienced a pronounced performance collapse under strong, untrained wind disturbances, a regime where the non-adaptive classical PID baseline proved unexpectedly stable. These findings confirm the benefits of hybrid control for maximizing robustness and highlight that the system’s ability to handle wind disturbance rejection remains a significant, unresolved shortcoming for current RL guidance systems.