Vision based ground object tracking using AR.Drone quadrotor

This paper presents an approach to the ground object tracking problem of the low cost AR.Drone quadrotor via its integrated vision systems. It consists of three major components: (1) on-board vision based object detection and localization, (2) comprehensive empirical models of X-Y coordinate motion of the quad rotor and (3) closed-loop PD controllers on the ground station to generate steering commands via Wifi connection. The developed platform has been extensively validated in various testing environments, demonstrating its effectiveness.