Integrated INS/GPS System for an Autonomous Mobile Vehicle

This paper presents the design of an extended Kalman filter (EKF) to fuse GPS and MEMS-based inertial measurement unit measurements (IMU) for real-time state estimation of a high performance vehicle called CENTAUR. To fully enable its autonomous control capability, the navigation system requires estimates of the vehicle position, velocity and heading angle. Also, to control stability of the vehicle during more difficult maneuvers, estimation of bank angle as well as pitch angle should be provided. The estimation problem is complicated by the large random-walk error, along with vehicle´s wide-band vibration. By combining the IMU data with GPS position measurements, the position estimation is adequate, and by incorporating Centaur´s kinematics model (and using its stepper motors control input signals), the velocity and orientation estimation effectively improves. The filter represents quaternion rather than Euler angles which eliminates the problem of singularities and relatively high required computation time associated with attitude estimation. Both computer simulation results and real-time experimental results show that integrated system has better long-term precision than a low-cost INS alone, and also better than a stand GPS in terms of availability and continuity.