Any Object Tracking and Following by a Flying Drone

Abstract

AR.Drone is a quadcopter with onboard sensors including a frontal camera. The drone can be controlled from a computer via WiFi. This paper describes a method for autonomous tracking of a selected object by AR.Drone. We utilize a computer-vision approach called tracking-learning-detection (TLD) to track an arbitrary object selected by a user in the video-stream going from the front camera of the drone. Information about location of the tracked object is then used to guide the drone using the proportional-integral-derivative (PID) controller. The method was implemented in software FollowMe.