POS MV V5 Installation and Operation Guide
Theory
of Operation
Copyright © Applanix Corporation, 2017
B-30
also exhibits a strongly recognisable Schuler oscillation superimposed on the growing position
error.
•
The GNSS position error is noisy in the short term but exhibits a constant long-term error of only
a few metres (with DGNSS).
Schuler Oscillations
Schuler oscillations appear as regular oscillations about a mean
position in the output of a free inertial navigator. You can also see
them on the output from a gyrocompass. These oscillations have a
period of 84.4 minutes, which is the theoretical period of a
pendulum having a length equal to the radius of the Earth.
In a free inertial navigation system, the oscillations occur as the
platform (whether mechanical or mathematical) over-corrects for
small residual errors in platform levelling and alignment. These
residual errors may arise because of uncorrected bias in the
accelerometers or the gyros.
You can expect the best performance from the aided inertial navigation system when the advantages of
one navigation sensor compensate for the disadvantages of the other.
The following is a simplified example of how navigation data from an INS and a GNSS receiver can be
blended or integrated in a complementary fashion (refer to Figure 97). Note that POS MV uses a
significantly more complex process involving a Kalman Filter.
1. The position solution determined by the INS includes an error that varies with time. This error
comprises a drift, which increases at a linear rate with time, and the Schuler oscillations having a
period of 84.4 minutes. Depending on the quality of components used in the INS, the overall
combined effect of these two error sources can be considerable.
2. The DGNSS receiver delivers a position solution that is stable in the long term and includes an
error of only a few metres (depending on the quality of differential correction information
supplied). However, the DGNSS position solution includes short-term errors of several metres
amplitude that appear as ‘noise’ on the position solution.
3. Calculating the difference between the INS position and the DGNSS position, and then passing
this difference through a low-pass filter can determine a good estimate of the INS position error.
The filter smoothes the random noise from the DGNSS solution and passes the INS position error
and the small steady-state DGNSS position offset.
