A robust GPS/INS positioning system for off-road autonomous navigation
Zhiyu Xiang · Journal of Circuits and Systems · 2008
To navigate efficiently on uneven off-road terrain, autonomous vehicles must have reliable 3D positioning capability. A total state space based Extend Kalman Filter (EKF) is proposed to fuse GPS/INS sensors. The observation equations are set up following the analysis of the error source of the sensors. The system state equations are obtained from discretizing the continuous state equation. Finally the system states are estimated through the classical EKF process. Comparing with the traditional error space based fusion algorithm, our method can relax the strict error modeling process of INS and is able to deal with short-time INS and GPS outages caused by the bumpy roads and obstructions respectively. A real GPS/INS positioning system is then built up and tested. The experimental results showed our success.