6.1 Introduction

The computational flow diagram in Chapter 5 reveals three sets of differential equations that need to be integrated and initialized. These sets are the craft attitude (inherent in or ), the craft velocity, v, and the craft position, inherent in . Of these variables, only that does not depend explicitly on sensor data. Usually a conflict arises between high accuracy demands and limited computer resources when implementing integration algorithms. What we shall encounter is that inertial sensors supply data at high sampling rates, (typically at 1000 Hz or higher), while the variables that need to be integrated do not vary appreciably during these small time intervals. For example it suffices to compute the craft attitude at a rate of about 100 Hz and to compute the craft velocity and position at rate of about 10 Hz. To meet high accuracy requirements one might be tempted to integrate the navigation equations at the sensor high sampling rates. Doing so will inflicts a severe burden on computer resources. Conversely, integrating the equations at a slower rate, and not benefiting from the high sampling-rate data, will ultimately degrade the navigation solution.

Fortunately with little computational burden we can benefit from all the sensor data without overloading the navigation processor. We achieve that by preprocessing the sensor data at its high sampling rate and feeding the highly accurate solution to the navigation equations only at the attitude rate (100 Hz). The advantage of this preprocessing approach is that it...