Vision based target tracking using an unmanned aerial vehicle

Abstract

We present in this paper a backstepping controller for vision based target tracking with an Unmanned Aerial Vehicle. A down facing camera is used with a pose estimation algorithm to extract the position of the target (an Unmanned Ground Vehicle). The output is then fed into the developed controller to generate the necessary movements (pitch and roll) of the Unmanned Aerial Vehicle in order to keep the target in the coverage view of the camera (following it constantly). The developed scheme is used to help the Unmanned Ground Vehicle to navigate among obstacles, and the overall system is designed in order to help human operator to supervise the Aerial and Ground vehicles for area inspection or object transportation in industrial areas (when using multiple Unmanned Ground Vehicles).