Proposta de um sistema de localização, posicionamento e movimentação para aplicação em VANTS autônomos de inspeção de tubulações aparentes não mapeadas

Abstract

The UAVs usage in the oil industry has been growing and helping to reduce costs, increase agility and safety. Operations such as cleaning and inspection are just two of the many tasks in wich the UAVs usage can be beneficial. This work aims to present an architecture that a UAV (Unmanned Aerial Vehicle), positioned at a point in the pipeline can recognize it, navigate through it completely, without traveling through the same pipeline twice and return to the starting point having constructed a full map of the pipeline, based on its own movement and distance in relation to the center of the pipeline. To carry out these tasks, some challenges need to be overcome, including: positioning; movement and mapping. To solve these three challenges, in the context off apparent pipelines transporting oil and gas derivates, the propoused architecture uses: (a) a positioning algorithm related to the interpretation of the image obtained from the onboard camera, (b) a global positioning algorithm, such as GPS, (c) a movement algorithm based on each type of positioning and (d) a heuristic to recognize the already traveled branches of the pipeline. The results obtained in simulated environments created to validate this architecture returned values of distance traveled less than twice the size of the pipeline. The execution time ir strongly related to the number of curves and branches of the pipeline, which turns out are related to the stabilization time of the UAV when decisions needs to be made and generate maps that can be used later for other applications, with less then 1 meter errors.