Attitude estimation using fusion of monocular SLAM and inertial sensors

This paper presents a novel technique on attitude estimation based on fusion of orientation measurements obtained from monocular SLAM (Simultaneous Localization and Mapping) and inertial sensors, using an Extended Kalman filter as sequential estimator. The development of the Attitude and Heading Reference System (AHRS) is described in detail. Information obtained independently from the two systems is combined using two approaches for comparison purposes: an augmented observation vector, and a minimum quadratic mean estimator. The Kalman filter prediction procedure is carried out in a single block, improved by including the estimation of the fused state using a modified track to track approach. A comparison on system performance, before and after the described sensor fusion methods, is presented.