Pose Estimation of Rotational Stationary Platform Using Monocular Camera and 6-Axis Inertial Measurement Unit

Abstract

In this research, we propose a pose estimation technique for a rotational stationary platform that fuses a monocular camera and 6-axis inertial measurement unit in environments where conventional GNSS-aided inertial navigation system or magnetometer-aided inertial navigation system is not feasible. The proposed estimation technique adopts an extended Kalman filter structure with loosely coupled approach, utilizing azimuth measurements obtained through feature point matching between images for state correction. As a result, the pose estimation error was significantly reduced compared to the case where only the inertial measurement unit was used. In addition, when comparing the results of the proposed method with those obtained from a commercial magnetometer-aided inertial navigation system, the azimuth estimation error was small, and the tilting error was similar.