Vision-based position estimation utilizing an extended Kalman filter

Abstract

The purpose of this project is to develop and integrate a prototype multicopter Unmanned Aerial Vehicle (UAV) with a vision-based algorithm to enable a relative position hold capability. The resulting solution will enable automatic operation of a UAV with respect to another visually detectable object without use of GPS receiver or when GPS signal is not available. Navigating a robot in a GPS-denied environment is a desired feature in many applications, including Maritime Interdiction Operations. While automatically maintaining its relative position with respect to a given target, the onboard system will also provide video coverage of blind spots and network relay between the boarding team and ship.