| Resumo: | In order to reduce the consumption of fossil fuels, investment in renewable energy sources, such as wind, has been increasing. The vast majority of wind farms are on land. However, the number of offshore wind farms is increasing. One of the technologies developed is the WindFloat, which is a semi-submersible structure with 3 floating columns that form a triangle, responsible for supporting the wind turbine. This accelerated maritime expansion requires the existence of adequate inspection and monitoring procedures. As WindFloat structures are located in a remote location on the sea and monitoring tasks are often repetitive, time consuming and potentially dangerous for human operators, these are ideal for being performed by autonomous robots, such as ASVs (Autonomous Surface Vehicle). ASVs are water vehicles that, as the name implies, travel on the surface, being equipped with sensors that give them the ability to locate themselves and to know the characteristics of the environment that surrounds them. To an ASV navigate in the vicinity of a WindFloat, a system for detecting and tracking its structure is required. this must operate in real time and be capable of responding safely. For this, it is required that the ASV has a system of relative location of high precision (centimeters) and that is based on the natural features of the surrounding environment. The use of GPS is not a viable option, not only because of its limited accuracy, but also because of the signal intermittence motivated by the proximity to the WindFloat metallic structure. In addition, it is also necessary to have a system for measuring and mapping danger areas in order to mitigate potential risks, both for the ASV itself and for the WindFloat. In order to fulfill the previously described objectives, a relative location algorithm was developed, in which each WindFloat column corresponds to a vertex of an equilateral triangle, used to obtain the location of the ASV by triangulation. These columns are detected by processing a point cloud collected by a LiDAR sensor. This processing consists of identifying the column and estimating the location of its center. This approach showed highly accuracy in the localization of the ASV. Its average euclidean error is of only 0.004 meters. The developed danger zone mapping algorithm took into account the current position and orientation of the ASV, as well as the influence of environmental factors, such as waves, sea currents and wind. In this way, it was possible to objectively classify the different danger areas, both for the ASV and for the WindFloat structure, in real time, reducing the risk of potential collisions between them.
|
|---|