Abstract :
Terrestrial inertial navigation is typically performed using an instrumented platform stabilized in a ¿local-level¿ configuration for convenient generation of geographic navigation information. The local-level geographic reference must be maintained by torquing the system gyros, a requirement which may be incompatible with high-precision inertial sensors currently under development. Gyro torquing in a gimballed navigation system can be avoided by employing a ¿space-stable¿ mechanization of the platform where an inertial, rather than geographic, reference is used for navigation calculations. The software design problems associated with this concept, especially those related to the application of Kalman filtering, are the principle focus of this paper. Although the space-stable configuration has been used extensively for spacecraft navigation and guidance, it has not been widely applied to terrestrial navigation, either for air or marine applications. The chief problem in this application is to perform navigation in local-level coordinates, using system outputs generated in an inertial reference frame. It can be demonstrated that, although the navigation system error dynamics are identical for the local-level and space-stable configurations, the dynamics of the sensor errors which drive these systems are quite different. These differences in sensor error propagation characteristics impose new requirements for the design of procedures to accomplish system calibration, alignment, and reset. This paper outlines a Kalman filtering approach which is applicable to all of the above procedures, and presents numerical results to demonstrate its effectiveness.
