In this master's thesis we explore various Machine Learning approach to Traffic Simulation. The goal was to integrate GNNs into SuMO (Simulation of Urban MObility), a German traffic simulation app and predict flows from given OD matrices and visa-a-versa i.e. the inverse problem. Novel concepts like traffic flux and traffic flow was introduced followed by intricate neural network architectures utilized to make predictions.

We present Deblur-SLAM, a robust RGB SLAM pipeline designed to recover sharp reconstructions from motion-blurred inputs. The proposed method bridges the strengths of both frame-to-frame and frame-to-model approaches to model sub-frame camera trajectories that lead to high-fidelity reconstructions in motion-blurred settings. Moreover, our pipeline incorporates techniques such as online loop closure and global bundle adjustment to achieve ...

Debris flows are highly dynamic natural hazards that pose significant risks in mountainous regions. Accurate detection and classification of these events are critical for early warning systems and hazard mitigation. This study investigates the potential of unsupervised learning, specifically Gaussian Mixture Models (GMMs), to cluster debris flow events using a range of seismic and acoustic features, including seismic envelopes, ...

While the rise of shared and free-floating e-scooters presents opportunities for sustainable urban transportation, little is known about the equitable integration into urban mobility ecosystems. This thesis investigates the social sustainability of e-scooters, focusing on spatial, social, and public space equity. In addition, it focuses on socially acceptable municipal policy strategies to effectively and efficiently improve the regulation ...