Gyro Drift modeled as random walk and

exponentially correlated processes

AE 645
Integrated INS/GPS Navigation
April 2, 2009
 Gyro drift – an example of random walk

random walk process

0.04
random walk: drift (deg)

0.02

-0.02

-0.04

-0.06
0 1 2 3 4 5 6 7 8 9 10
simulation time - sec

Sample mean and standard deviation, and analytical SD

0.04
mean
rt(var)
0.03
rt(svar)

0.02

0.01

-0.01
0 1 2 3 4 5 6 7 8 9 10
simulation time - sec
Gyro Drift modeled as first-order exponentially correlated process

Gyro drift: exponentially correlated (Gauss-Markov)

-1

-2

-3
0 1 2 3 4 5 6 7 8 9 10
simulation time - sec

Gyro drift sample mean, SD; analytical SD (deg)

1.5
mean
rt(var)
1 rt(svar)

0.5

-0.5
0 1 2 3 4 5 6 7 8 9 10
simulation time - sec
AE 457 – 641: Navigation and Guidance: Quiz 1
(February 13, 2012, Monday)
Time: 1 hr should be enough, but no more than 2 hrs.

Que.1 Air Traffic Controller, Sunanda, granted Captain Rahul permission to take off from
Mumbai Airport. So, before taking off, Rahul initialized the airplane’s navigation platform by
using the earth’s gravity and the spin vector. Suppose the platform has both north and east
tilt, and some azimuth error. [The answers below are expected to be analytical, not
numerical.]

•What are the ideal measurements of the 3-axis accelerometer and gyros in the geographic
frame while the plane is still on the ground?

•What would be the measurement of the three-axis accelerometer in the platform frame?
•What would be the measurement of the east gyro on the platform?

•What are the tilt angles that the platform should be tilted by in order to align the platform
vertical with the geographic vertical?

•What is the azimuth angle by which the platform should be rotated so as to align the
platform north and east with the geographic north and east?
Que. 2 Some platform misalignment with the geographic frame
remained after the above process, however.

•Suppose the remaining north tilt is 3 millirad. How much would be the east velocity
error (in ft/s) in 3 minutes after takeoff?

•Suppose Rahul gets a Doppler radar measurement of this east velocity then, and the
measurement is accurate to 2 ft/s. Could he use this measurement to reduce the tilt
error? How much would be the revised tilt error?

•If the angle of the Doppler measurement were 45 degrees from the east, what would
be the revised tilt error of the platform?

•Now suppose the tilt error is taken care of by the Doppler measurements, but there
was an initial azimuth tilt error of 0.5 degree, and the flight is southbound to Goa, with
a speed of 600 ft/s. How much would be the east velocity error (in ft/s) in 20 minutes
from takeoff just due to the azimuth tilt error? Ignore the earlier north tilt.

•With a 2 ft/s accurate Doppler measurement again, in the direction of flight, how
much this azimuth error would be reduced to?