Autonomous visual navigation of Unmanned Aerial Vehicle for wind turbine inspection

An autonomous machine vision module for Unmanned Aerial Vehicle (UAV) navigation for inspection of wind turbines is presented. The system estimates the relative position and distance between the UAV and the wind turbine, as well as the position of its blades, in order to support the initial phase of autonomous inspection before the UAV start to move along the blades to acquire pictures. The key algorithms used are Hough transform for detection of the wind turbine tower, hub and blades, as well as the Kalman filter for tracking. Experimental data acquired by a UAV at a wind park is used to evaluate the accuracy and robustness of recognition and navigation. It is found that under the tested conditions, the method gives sufficient accuracy for the task and can execute in real time on a single board computer in the UAV.