Closed-form solution for absolute scale velocity estimation using visual and inertial data with a sliding least-squares estimation

In this paper a method for the on-line absolute-scale velocity estimation of a system composed of a single camera and of an inertial measurement unit is presented. The proposed formulation makes use of spherical image measurements acquired from at least three camera positions and inertial measurements to estimate the system velocity by solving also the absolute scale problem. A new multi-rate formulation based on a sliding least-squares estimation formulation is proposed, which is capable of providing the velocity estimation also in cases of constant and zero velocity. The effectiveness of the proposed approach is shown through extensive simulations.