Immersive Third Person View Automatic User Tracking with Live Video from an Unmanned Aerial Vehicle

Abstract

Third Person View (TPV) is a video game camera perspective where the game character is seen from behind. Due to decreased prices of Unmanned Aerial Vehicle (UAV) technology and modern Head Mounted Displays (HMD) it has become more available to create a system were a user sees themselves from behind. This thesis describes the process of designing a system for creating the TPV with low latency video streaming and short range tacking for automatic flight. A quadcopter is designed to track and follow the user for achieving automatic camera positioning. Location tracking is implemented with triangulation by ultrasonic sound. Following user rotation is implemented by comparing magnetometers on the user and the quadcopter. From a camera on the quadcopter live video is transmitted by Wi-Fi to a laptop connected to an Oculus Rift Development Kit 2 (DK2) which presents the video feed. The highest emphasis has been on achieving accurate tracking but also on rapid quadcopter regulation using PID-controllers for maintaining the view. The resulting system works for tracking and reacting to a moving user, however during fast user rotations or movements the tracking is lost, due to small angle of the ultrasonic receivers. For controlling and observing the system during flights and tests, a graphical tool was developed as well as software for tracking and regulation. These systems implemented shows promising results and is an area worth of future study.