Robust cooperative UAV Visual SLAM

Abstract

This paper aims at proposing a framework for Airborne Cooperative Visual Simultaneous Localization and Mapping (C-VSLAM). The use of cooperative vehicles presents many advantages over single-vehicle architecture. We present a nonlinear H∞ filtering scheme adapted to multiple Unmanned Aerial Vehicle (UAV) VSLAM based on the extension of a robust single vehicle VSLAM solution. Loop closure concept, based on revisited features is described with feature uncertainty analysis. Comparisons between single and multiple UAV VSLAM are made using realistic simulation scenario.