A Stereo-Based Visual-Inertial Odometry for SLAM

Abstract

It is becoming more and more popular to combine vision and inertial units in mobile robots, because these two kinds of sensors can complement each other and achieve good results in calculating incremental motion of sensors. SLAM algorithm based on pure vision, do not apply to the carrier fast movement, not a good deal with environmental degradation problem, so the current research focus is to do the camera and innovative marketing combination to SLAM, there are quite a good monocular visual inertial navigation system, but a lack of direct monocular vision inertial navigation of distance measurement, monocular vision is difficult to directly with inertial measurement data fusion structure, need very strict initialization process.Therefore, this paper proposes an improved binocular vision inertial navigation system for SLAM.