Resumen de Enhancing urban mobility and safety through virtual 3d scenarios constructed from multi-source point clouds
Urban mobility and safety are critical components of modern city planning, directly impacting the quality of life for citizens. Efficient transportation systems, including road network infrastructure, public transit networks, and pedestrian-friendly pathways, play a pivotal role in facilitating the movement of people and goods within urban environments.
However, ensuring the safety of these transportation modes is equally essential. Various factors, including weather conditions and obstructions in the road environment, can significantly impede the ability of drivers and pedestrians to accurately perceive their surroundings.
LiDAR technology has emerged as a tool to enhance urban mobility and safety within city environments. Various LiDAR sensors are available, each chosen based on specific needs and requirements. In the context of the presented doctoral dissertation, Mobile Laser Scanning (MLS), Airborne Laser Scanning (ALS), and Handheld Mobile Laser Scanning (HMLS) data serve as input data for the methodologies, each contributing distinct characteristics to the dataset. To ensure a comprehensive understanding, data must undergo semantic segmentation, typically through manual labelling. Recognizing the potential for human errors in this process, a methodology to quantify discrepancies is proposed. Subsequently, using labelled data, various Machine Learning (ML) and Deep Learning (DL) techniques are tested and compare. To generate a virtual 3D scenario, a dataset comprising MLS and HMLS urban point clouds are labelled. A methodology is proposed to categorize these point clouds into eight classes (road, sidewalk, curb, buildings, vehicles, trees and others). Then, using the labelled data, three PointNet++ models are trained, validated, and tested using MLS data, HMLS data, and a combination of both H&MLS.
A safety analysis is carried out using a virtual 3D scenario generated from MLS and ALS point clouds. A methodology has been developed to accurately calculate sun glare on drivers for any given time and day of the year. This approach considered the intersections of sun rays across both nearby and distant surroundings. This methodology is applicable to interurban roads and roundabouts, demonstrating its versatility across various road configurations and enhancing its utility in ensuring roadway safety.
Besides drivers, pedestrians are also users of the urban environment, emphasising the importance of conducting an analysis of their mobility. Public transport serves as a mode of transportation for pedestrians, facilitating movement between various locations within the city.
