APPLICATION OF QUAD-COPTER TARGET TRACKING USING MASK BASED EDGE DETECTION FOR FEASIBILITY OF WIND TURBINE BLADE INSPECTION DURING UNINTERRUPTED OPERATION

Abstract

In this thesis we present a feasibility study on the use of UAVs & edge detection on an existing drone platform to allow for inspection of wind turbines without the need for priming the blade or stopping the wind turbine. We focus our efforts on implementing two codes: The first is suitable for a controlled lab environment using python OpenCV while the later use of machine learning mask-based algorithm with a stationary camera to test blade tracking feasibility in noisy environment. Both codes were able to track wind blade position visually up to 10rpm under stationary drone conditions. But hardware limitation hindered the drone tracking portion to about 1-3rpm. Future works to improve this methodology involve the use of more robust sensing and the implementation of control dynamics to further explore different control schemes and tracking paths.