Integration of differential GPS and inertial navigation using a complementary Kalman filter

Abstract

Precise navigation with high update rates is essential for automatic landing of an unmanned aircraft. Individual sensors currently available - INS, AHRS, GPS, LORAN, etc. - cannot meet both requirements. The most accurate navigation sensor available today is the Global Positioning System or GPS. However, GPS updates only come once per second. INS, being an on-board sensor, is available as often as necessary. Unfortunately, it is subject to the Schuler cycle, biases, noise floor, and cross-axis sensitivity. In order to design and verify a precise, high update rate navigation system, a working model of Differential GPS has been developed including all of the major GPS error sources - clock differences, atmospherics, selective availability and receiver noise. A standard INS system was also modeled, complete with the inaccuracies mentioned. The outputs of these two sensors - inertial acceleration and pseudoranges - can be optimally blended with a complementary Kalman filter for positioning. Eventually, in the discrete case, the high update rate and high precision required for autoland can be achieved.