An Empirical Evaluation of Four Off-the-Shelf Proprietary Visual-Inertial Odometry Systems

HIS article presents a benchmark comparison of off-theshelf proprietary visual-inertial odometry (VIO) systems in six challenging real-world environments, both indoors and used for autonomous navigation of robotic applications, which outdoors. Especially, we select the following four proprietary are the process of determining the position and orientation of VIO systems that are frequently used in autonomous driving a camera-inertial measurement unit (IMU)-rig in 3D space by robotic applications: analyzing the associated camera images and IMU data. As Apple ARKit [4] - Apple's augmented reality (AR) platform, the VIO research has reached a level of maturity, there exist which includes filtering-based VIO algorithms [8] several open published VIO methods such as MSCKF [1], to enable iOS devices to sense how they move in 3D OKVIS [2], VINS-Mono [3], and many commercial products space.