Development of target detecting and landing function in persistent UAV system

Abstract

In this study, vision-based object tracking function for persistent UAV system is developed and implemented. Vision-based object tracking system is the necessary part in the implementation of UAV systems that return to the charging station after the mission. Mixed integer nonlinear programming (MINLP) was used to optimize trace route without the information about the location of the target. Through the MILNP, a tracking path has been developed that maximizes the given detection probability and minimizes the travel time within the region. In addition, the flight of a real UAV tracking a target along a given line through ROS is implemented. The software was developed on the basis of a Parrot's AR Drones, and the tag provided with the AR Drone was used for target detection.