This paper presents the design, analysis, and experimental validation of a sensor-based globally asymptotically stable (GAS) filter for simultaneous localization and mapping (SLAM) with application to uninhabited aerial vehicles (UAVs). The SLAM problem is first formulated in a sensor-based framework, without any type of vehicle pose information, and modified in such a way that the underlying system structure can be regarded as linear time varying for observability, filter design, and convergence analysis purposes. Thus, a Kalman filter follows naturally with GAS error dynamics that estimates, in a robocentric coordinate frame, the positions of the landmarks, the velocity of the vehicle, and the bias of the angular velocity measurement. The online inertial map and trajectory estimation is detailed in a companion paper and follows from the estimation solution provided by the SLAM filter herein presented. The performance and consistency of the proposed method are successfully validated experimentally in a structured real world environment using a quadrotor instrumented platform.