| Gyro stabilized inertial navigation systems are subject to drift and oscillation. Within hours navigational errors of several thousand meters can develop. This paper demonstrates, with the aid of highly accurate GPS data, the excellent error-tracking ability of a 15-dimension, time-varying Kalman filter. The three-axis inertial platform is modeled as a two-accelerometer local-level system having gyro drift and accelerometer bias errors. The filter, GPS receiver, and INS are configured in an open loop mechanization. The estimated gyro and accelerometer error parameters result in a state transition matrix able to accurately predict navigation errors during the initial three to four hours of run time. |
