A Quadrotor Flight Mode For Tracking Based On Computer Vision Using ROS

Abstract

Unmanned Aerial Vehicles (UAVs) are particularly interesting for covering large areas or observing regions from a privileged angle of view. However, many applications still require direct human control. The addition of visual feedback can enable these vehicles to perform tasks more autonomously. In this context, this work proposes an autonomous flight module, based on computer vision, with embedded processing, to follow a moving target in external environments. Interesting results are obtained through real tests done with a quadcopter that uses Pixhawk for low-level control and a Raspberry Pi for high-level control.