Multi-Sensor Fusion for Navigation of Ground Vehicles

In this thesis the navigation solution of a ground vehicle is computed by fusing navigation data from Inertial Navigation System (INS), Visual Odometry (VO), and Global Positioning System (GPS) using a Dual Extended Kalman Filter (DEKF) algorithm. The research in this thesis is conducted in three phases. The first phase presents the VO navigation system. In this phase an improvement to traditional Stereo Visual Odometry (SVO) methodology is proposed and the Modified Stereo Visual Odometry (ModSVO) algorithm is presented. The second phase presents the development of INS/VO and INS/GPS navigation solutions using EKF. The results showed improved accuracy compared single sensors. However, in case of VO failure or GPS failure the accuracy is shown to deteriorate. The third phase presents the development of INS/VO/GPS system using DEKF algorithm. It is shown that the INS/VO/GPS system outperforms INS/VO and INS/GPS systems in cases VO or GPS sensor failure.