Adaptive Extended Kalman Filter for Integrated Navigation in a Satellite Launch Vehicle

Abstract

A flight navigation system usually integrates an Inertial Navigation System with a global positioning receiver by means of a Kalman Filter and using an appropriate model of the flight kinematics and the stochastic processes included on the inertial and external (global positioning by satellites) measurements. These processes are characterized mainly by their first and second order moments, the latter described by covariance matrices which can be time-varying. On a satellite launch vehicle the velocities are so high that the random delays of the external measurements generated by the receiver have a relevant effect on the navigation errors. This can be mitigated by an adaptive version of the Extended Kalman Filter, which is formulated in this work and verified by numerical simulations corresponding to an injection trajectory.