System Development and Flight Experiment of Vision-Based Simultaneous Navigation and Tracking

Abstract

This paper outlines a UAV onboard vision-based system for simultaneous navigation and tracking of a moving ground target in a GPS-denied environment. The system consists of i) an image processor which detects the target and estimates the apparent motion of the groud surface, ii) a navigation filter which localizes both the own-ship UAV and the target by fusing onboard inertial sensor measurements and the image processing results, and iii) a guidance law which makes the UAV pursue the target by using the navigation results. The entire system including those algorithms is implemented as a single component of the Orocos robotic architecture, integrated with the auto-pilot system of the ONERA ReSSAC unmanned helicopter, and evaluated through closed-loop simulations and flight experiments.