Planar-based visual inertial navigation

This paper investigates the problem of visual-inertial navigation. The proposed navigation system integrates inertial information from an inertial measurement unit (IMU) with visual data from a camera to provide relative pose estimation for a system which is navigating in an unknown structural environment. The main contribution of this paper is derivation of a novel measurement model based on inertial data and visual planar features. The proposed formulation is a solution to the 6-DoF motion estimation where the IMU-camera movement is not restricted over a desired navigation plane. Compared to previous works, which are restricted on using only horizontal plane features, the proposed model is generalized for arbitrary planar features. The theoretical finding of this study is extensively evaluated both with simulation and real world experiments. The presented experiments indicate the reliability of the proposed method to perform accurate 6-DoF pose estimation.