A Design of Odometer-Aided Visual Inertial Integrated Navigation Algorithm Based on Multiple View Geometry Constraints

摘要

This paper is dedicated to studying robotic autonomous navigation technologies when the GNSS (Global Navigation Satellite System) is unavailable. To improve the robotic navigation accuracy, we propose an algorithm to fuse information from a single camera, an IMU (Inertial Measurement Unit) and a wheel odometer. At first, we present the inertial navigation equation and the error equation. Then we introduce the odometer-aided visual inertial integrated navigation algorithm based on Federal Kalman Filter. Information from wheel odometer and camera is taken as measurement by two different sub-filters, i.e., KF (Kalman Filter) and UKF (Unscented Kalman filter). KF is used to fuse the IMU/Odometer information while UKF is employed to optimize the motion estimation of carrier by using the output of KF and geometry constraints among different views of camera. We conduct a set of experiments on publicly available dataset. The results show the effectiveness, correctness and robustness of the proposed algorithm in real environment.