gains.  In this case, the barometric altimeter error is modeled in the Kalman filter

state vector.  The barometric altimeter measurement will only be processed by the

Kalman filter when GPS satellite coverage is incomplete.  If not processed as a

measurement to the Kalman filter, the barometric altitude can be differenced with

inertial altitude to create an error signal that is filtered and fed back to inertial

altitude, giving a blended baro/inertial altitude.

9.4.5  Kalman Filtering and GPS/AHRS

An AHRS is a strapdown system which uses lower quality gyros and accelerometers

than a strapdown INS.  Used as a stand  alone system, an AHRS is similar to an

INS, except that no position is available, only attitude (pitch, roll and heading),

attitude rates and acceleration.

GPS can provide three dimensional bounded position and velocity aiding

information to the AHRS to improve its outputs.  An AHRS, in its turn, can be used

for short term fill in of velocity information if the receiver outputs are lost.

Potentially, AHRS velocity could also aid the GPS receiver tracking loops during

short periods of jamming or high dynamics. The separate Kalman filter, in case of

an aided GPS/AHRS, does not include the vertical channel. A typical Kalman filter

model has 14 error states: horizontal position (2), hori zontal velocity (2), rotations

(2), gyro drift rates (3), gyro scale factor (1), wander  azimuth angle (1) and

accelerometer bias (3).  The combined Kalman filter for a tightly coupled, integrated

GPS/AHRS adds four error states: vertical position, vertical velocity, clock phase

error and clock frequency error.

9 17