A method for passive GPS-free navigation of a small Unmanned Aerial Vehicle with a minimal sensor suite (limited to an inertial measurement unit and a monocular camera) is presented. The navigation task is cast as a Simultaneous Localization and Mapping (SLAM) problem. While SLAM has been the subject of a great deal of research, the highly non-linear system dynamics and limited sensor suite available in this application presents a unique set of challenges which have not previously been addressed. In this particular application solutions based on Extended Kalman Filters have been shown to diverge and alternate techniques are required. In this paper an Unscented Kalman Filter is applied to the navigation problem, which leads to a consistent estimate of vehicle and feature states. This paper presents: (a) simulation results showing mapping and navigation in three dimensions; (b) preliminary hardware test results showing navigation and mapping using an off-the-shelf inertial measurement unit and camera in a laboratory environment.