Método de estimativa de trajetória para seguimento de linhas de transmissão através de veículos multirrotores

Abstract

The inspection of power lines is one of several areas that have been benefited from the advent of Unmanned Aerial Vehicles (UAVs) that, due to characteristics such as mobility and flexibility, provide advantages over manual inspections and ground vehicles. One of the steps in this task is the determination of the trajectory to be performed by the UAV and thus, this work aims to introduce a method to power lines path following as a local trajectory planning. Initially, the images obtained by cameras attached to the UAV are processed and the lines are detected by a generic detector that does not distinguish power lines from other rectilinear elements. Then a curve fitting method is employed to mitigate the effect of spurious lines and determine the trajectory with the purpose of keeping it in the center of the lines. Tests were conducted and the best combination of techniques for line detection taking into account accuracy rates, precision and execution time was the Canny + Kernelized Hough Transform (KHT) method. For curve fitting the method that best approximated the trajectory to the expected straight line was the RANSAC algorithm. Practical tests involving the proposed methodology with an NMPC controller were performed in an indoor environment showing that the UAV was able to follow the lines according to the imposed constraints. Experiments involving the calculation of the path in real environments were performed with no controller integration and the results show that the estimation was consistent in most of the frames.