Visual Assisted EKF SLAM for Planetary Roving Vehicle

Abstract

Navigation of the planetary rover in the unforeseen planetary environment is an uphill task. In this paper, we presented a visual assisted inertial navigation system. System performance is evaluated by using real world KITTI data set. This work encompasses feature detection, feature matching, outlier rejection and motion estimation of the roving vehicle. This method is effective for the GPS-denied planetary environment. A tightly-coupled sensor blending method, for monocular camera image features and inertial measurement unit is presented for rover’s trajectory estimation. RANSAC is employed, for outlier rejection.