Hybrid Positioning and Sensor Integration - Global Navigation Satellite Systems: Signal,Theory and Applications - page 344

imperfections such as accelerometer bias and gyroscope drift. Hybrid positioning can
mitigate the error by being updated periodically with external fixes, such as GPS, VMS,
images, radio aids, or Doppler radar. Hybrid positioning is for finding the location of a
mobile platform using or combining several different positioning technologies. The effect of
fixing positions is that it allows for the reset or the correction of the position errors of the
inertial system to the same level of accuracy inherent in the position fixing technology. The
inertial system error grows at a rate equal to the velocity error. Therefore, external data is
used not only for the position update but also the error correction of inertial components
such as attitude, heading, velocity, gyro bias, and accelerometer bias. Furthermore, the error
of the external data such as misalignment error, boresight error, and scale factor error is
corrected in the same manner. Typical hybrid strapdown navigation is shown in Figure. 2.

Rotation
Sensors

X-gyro
Y- g y ro
Z-gyro

System Output
- Position
- Velocity
- Acceleration
- Attitude
- Heading
- Angular Rate
- Application output

Δθ 's

Strapdown
Processing
and
Navigation

Translation
Sensors

Δ V's

X-accel.
Y- a c c e l .
Z-accel.

Kalman
filter

External aids
- GPS
- Images
- Speed meter
- etc.

Observation
Processing

Fig. 2. Typical hybrid strap down navigation configuration

4.1 GPS and IMU data are integrated by Kalman filter

The Kalman filter can be used to optimally estimate the system states. One of the distinct
advantages of the Kalman filter is that time varying coefficients are permitted in the model.
With this filter, the final estimation is based on a combination of prediction and actual
measurement. Figure 3 shows the pure navigation algorithm for deciding IMU attitude and
IMU velocity step by step (Kumagai et al., 2002). Inertial navigation starts to define the
initial attitude and heading based on the alignment of the system. It is processed and then it
changes to the navigation mode. Over the years, the quality of IMUs has risen, but they are
still affected by systemic errors. In this research, a GPS measurement is applied as an actual
measurement to aid the IMU by correcting this huge drift error. With Kalman filtering, the
sensor position and attitude are determined at 200 Hz.

Figure 4 shows the Kalman filter circulation diagram for the integration of the GPS and IMU
data (Kumagai et al., 2000). Individual measurement equations and transition equations are