Resumen de Use of mobile mapping technologies for automated inspection of road pavements
In today's world, road infrastructures play a fundamental role as part of transportation networks. Roads are essential in the daily lives of millions of people, enabling them to travel to their homes, workplaces and in general, connecting them. In addition, the transport of goods by road has an enormous weight in the world economy. Because of all these human and economic factors, the proper maintenance of roads is crucial to keep their performance and prevent the breakdown of the connections in this delicate network. Inspection and maintenance tasks must, as far as possible, anticipate the appearance of damages that could compromise the integrity of roads. For instance, pavement cracks entail a major concern, as they appear due to the stress roads are subjected to and they can rapidly develop unless action is taken. It is therefore necessary to develop technologies to facilitate, improve and increase the eficiency of preventive measures.
Considering this context, the methodologies proposed in this thesis aim to provide tools to facilitate the inspection of roads, focusing on the interpretation of point clouds acquired with mobile laser scanner systems and the detection of cracks on the pavement from this information. For this purpose, semantic segmentation techniques were developed to extract the most relevant structural assets of the road, applying both heuristic methods and classification neural networks, and using data acquired with different scanning systems.
Starting from the segmentation of the pavement in point clouds it was possible to analyze the pavement surface for the identitication of cracks. For this purpose, detection techniques were developed based on different approaches: on the one hand, analyzing the neighborhood of each point to obtain details of the pavement geometry at a local level. On the other hand, by analyzing the scanning profiles individually, which allowed to decompose them into different frequencies and thus focus on the highest ones, which can describe the texture and irregularities of the pavement. Both approaches allowed the detection of the points of the cloud belonging to cracks, thus being able to segment and measure them.
This thesis covers the development of these methodologies, the presentation of the results of the experiments that were carried out to test them, a discussion based on the objectives initially established and the conclusions drawn. It also includes an explanation of the technologies and data processing techniques used to carry out the work, as well as a review of the works related to the topics investigate
