Computer Vision Controlling an Autonomous Unmanned Aerial Vehicle Flight over a Railway

In this work, an algorithm to control the autonomous quadcopter flight along the railway, without an operator and without the use of satellite navigation systems has been developed. The quadcopter camera image processing determines the rails position and direction and the software generates signals to control flight characteristics, such as pitch, yaw, and roll, to fly over the railway at a given height. The algorithm does not require significant computing resources, therefore its hardware does not affect the quadrocopter weight and size parameters and power consumption, and therefore does not affect the range and duration of its flight. Quadrocopter tests have confirmed the algorithm feasibility and reliability.