Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

A Novel Initial Alignment Scheme for Low-Cost INS Aided by GPS

for Land Vehicle Applications

Songlai Han(1)(2) and Jinling Wang(2)


(1)
(National University of Defense Technology, China)
(2)
(The University of New South Wales, Australia)
(Email: songlai.han@student.unsw.edu.au; Jinling.Wang@unsw.edu.au)

This paper proposes a novel mechanism for the initial alignment of low-cost INS aided by GPS. For low-cost INS, the
initial alignment is still a challenging issue because of the high noises of the low-cost inertial sensors. In this paper, a
two-stage Kalman Filtering mechanism is proposed for the initial alignment of low-cost INS. The first stage is designed
for the coarse alignment. To solve the problems encountered by the general coarse alignment approach, an INS error
dynamics accounting for unknown initial heading error is developed, and the corresponding observation equation taking
into account the unknown heading error is also developed. The second stage is designed for the fine alignment, where
the classical INS error dynamics based on small attitude error is used. Experimental results have indicated that the
proposed alignment approach can complete the initial alignment more quickly and more accurately compared with the
conventional approach.

KEY WORDS
1. Low-Cost INS. 2. Initial Alignment. 3. GPS. 4. Land Vehicle.

1. INTRODUCTION. Inertial navigation system (INS) is a kind of dead-reckoning based system, so the initial
navigational information should be provided to perform the subsequent navigational computation. The initial position
and velocity can be easily obtained from GPS or other aiding sensors (Lee et al. 2002; Wang et al. 2003), while the
acquisition of the initial attitude is not that straightforward and requires specific initial alignment algorithms to deal
with.
The initial alignments of high-precision INS and low-cost INS should be treated using different methods, because
the noise levels of the inertial sensors are quite different for these two types of systems. For high-precision INS, which
sometimes can be further classified into strategic-grade INS and navigation-grade INS (Greenspan, 1995), the initial
alignment can be achieved through two stages: coarse alignment and fine alignment (Savage, 2000). The coarse
alignment aims at making the unknown initial attitude errors converge to small angle errors, and is generally carried out
analytically on a stationary base by comparing the measured gravity and Earth rate with the standard gravity and Earth
rate (Savage, 2000). Under the premise of small attitude errors, the fine alignment can be performed based on the INS
error equations (Benson, 1975). The fine alignment aims at eliminating the residual attitude errors, and can be carried
out on a stationary base using the gyrocompassing theory (Savage, 2000), or carried out on a dynamic moving base
using in-motion alignment techniques (Baziw, 1972a and 1972b; Savage, 2000; Shin, 2004; Groves 2008) with the
aiding of GPS or other sensors.
However, the above initial alignment approach for high-precision INS cannot be used in the initial alignment of low-
cost INS. Generally, the low-cost INS uses gyros with drift rates of 1-10deg/h or more, and accelerometers with biases
of 0.1-1.0 mg or more (Greenspan, 1995). As is known, the Earth rate is about 15deg/h, so the noise levels of the low-
cost gyros are near or higher than the Earth rate, which means that the low-cost gyros cannot measure the Earth rotation
properly. So the coarse alignment technique based on the measurements of gravity and Earth rate for high-precision INS
cannot be used in the initial alignment of the low-cost INS, and neither can the fine alignment technique based on the
gyrocompassing theory be used. A possible solution for this problem is using the in-motion alignment technique.
However, although the in-motion fine alignment methods have been intensively explored (Baziw, 1972a and 1972b;
Savage, 2000; Shin, 2004; Groves 2008), the in-motion coarse alignment is still a challenging issue.
Because the gravity is a relatively large quantity, even low-cost accelerometers can measure it properly. So the
coarse leveling to determine the pitch and roll of the low-cost INS can be carried out using the gravity measurements in
stationary base prior to the in-motion coarse alignment (Savage, 2000). The remaining task is to deal with unknown
initial heading error using the in-motion coarse alignment techniques. Various approaches have been studied about this
issue. Dmitriyev et al. (1997) and Kong et al. (1999) constructed a nonlinear INS error dynamics model to account for
the unknown initial heading error. The benefit of the nonlinear method is that the in-motion coarse alignment is not
necessary and a single Kalman Filter can complete the initial alignment, but that method suffers from the complexity of
nonlinear operations. Rogers (1997), Kelley et al. (1990), and Pham (1992) developed a linear INS error dynamics
model in the wander azimuth frame for the in-motion coarse alignment to deal with the unknown initial heading error.
That method transferred the heading error into wander azimuth angle error, and modeled the sine and cosine errors of
the wander azimuth angle into the INS error equations. But the observabilities of the errors of sine and cosine of the
wander azimuth angle are very weak, because the coupling coefficients between these trigonometric function errors and
the position errors as well as the velocity errors are all small quantities. A quantitative analysis about this can be found
in Section 2. Discussions on the observability concept can be found in e.g., Bar-Itzhack and Porat (1980), and Han and
Wang (2008).
Inspired by the work of Rogers (1997), Kelley et al. (1990), and Pham (1992), a new linear INS error dynamics
model in the north pointing frame is developed in this research for the in-motion coarse alignment to deal with the
unknown initial heading error. Instead of the sine and cosine errors of the wander azimuth angle, the sine and cosine of
the platform heading misalignment are modeled into the INS error equations. In this new model, the observabilities of
the sine and cosine of the platform heading misalignment are much better, because the coupling coefficients between
these trigonometric functions of the platform heading misalignment and the velocity errors are much larger. The
coefficients in the velocity dynamics are the elements of the specific force, which is generally a large quantity because,
on the one hand, gravity is measured in the specific force, and, on the other hand, some intentional maneuvers should be
made during initial alignment of INS for land vehicle applications. Experimental tests show that this proposed model
results in a faster and more accurate alignment.
The subject matter of this paper is presented as follows. Section 2 reviews the conventional approach for the initial
alignment of low-cost INS and analyzes its drawbacks. To account for the unknown initial heading error, Section 3
develops a new error dynamics model and observation model for the in-motion coarse alignment of low-cost INS.
Section 4 is devoted to the in-motion fine alignment, the error dynamics model and observation model of which can be
easily derived as a special case of the in-motion coarse alignment under the condition of small heading error. Section 5
discusses the Kalman Filtering correction scheme for both the in-motion coarse alignment and fine alignment. Section 6
illustrates the configuration and results of the experimental tests, and Section 7 presents the concluding remarks.

2. CONVENTIONAL APPROACH FOR THE INITIAL ALIGNMENT OF LOW-COST INS: REVIEW AND
ANALYSIS. In Rogers (1997), Kelley et al. (1990), and Pham (1992), a two-stage initial alignment scheme for INS
was proposed. In the first stage, a linear INS error dynamics model accounting for unknown initial heading error was
developed in the wander azimuth frame to perform the in-motion coarse alignment, and in the second stage, a linear INS
error dynamics model accounting for small initial heading error was developed in the wander azimuth frame to carry
out the in-motion fine alignment. Although it was not specifically proposed for the initial alignment of the low-cost INS,
it did provide an alternative solution to this problem. However, this conventional approach has two drawbacks: weak
observability of the heading error and correlated driving noises of the velocity error dynamics.
The weak observability of the heading error arises from the fact that the coupling coefficients between the heading
error and the position error as well as the velocity error in the INS error dynamics are small quantities. The heading
error propagates as the velocity error and, as a result, the position error through the coupling coefficients. So, if these
coefficients are not large enough, the velocity error and position error propagated from the heading error may be
‘submerged’by the system noises, especially, for low-cost systems, and under such circumstances, the heading error
will become unobservable or weakly observable. The weak observability will affect the accuracy and increase the time
of alignment.

Table 1. The coupling coefficients between the heading error and system error states in the in-motion coarse alignment
using the conventional approach.
Coupling Coefficients Coupling Coefficients Coupling Coefficients
with Position Error with Velocity Error with Attitude Error

 x  y  h  v xn  v ny  v zn x y  s  c

 s 1  e n 0 2v z  n 0  2v x  n 0  n 0 0

 c  n  e 0 0 2v z  n  2v y  n n 0 0 0
Notes: 1. All the symbols are kept the same as equation (19) in Rogers (1997) for the convenience of comparison.

Table 2. The coupling coefficients between the heading error and system error states in the in-motion fine alignment
using the conventional approach.
Coupling Coefficients Coupling Coefficients Coupling Coefficients
with Position Error with Velocity Error with Attitude Error

 x  y  h  v xn  v ny  v zn x y  

2( v x  x
  1 y  x 0  2v z  x  2v z  y  y  n 0
 vy y )
Notes: 1. All the symbols are kept the same as equation (22) in Rogers (1997) for the convenience of comparison.

The coupling coefficients for the in-motion coarse alignment and fine alignment using the conventional approach
can be obtained from the equations (19) and (22), respectively, in Rogers (1997). The coupling coefficients are time-
varying, and their expressions are listed in Table 1 for coarse alignment and Table 2 for fine alignment, respectively.
From Table 1 and Table 2, it can be noted that the coupling coefficients between the heading error and the position error

are determined by,  en , the rotation rate of the navigation frame with respect to the Earth centered Earth fixed (ECEF)

frame, and the coupling coefficients between the heading error and the velocity error are determined by both ven , the
 
velocity of the navigation frame with respect to the ECEF frame, and  , the Earth rate. However,  en , ven , and 

are all small quantities in the initial alignment of INS for land vehicle applications. As aforementioned,  is about
 
15deg/h, which is a small quantity.  en , ven cannot be very large during the initial alignment, because the land vehicle
should perform some special maneuvers to accelerate the initial alignment process, and in our tests, which will be
discussed in detail in Section 6, two kinds of maneuvers were carried out: small ‘eight-like’movement, small circular

movement. These maneuvers should be carried out in restricted areas, which will limit the maximum value of ven and,

as a result,  en . Although the velocity cannot be very large during these maneuvers, the acceleration can, and based on
this analysis, this paper develops a new initial alignment scheme.
Another problem of the conventional approach is that the INS velocity error dynamics is driven by correlated noises.
According to equation (19) in Rogers (1997), the horizontal velocity error dynamics for coarse alignment can be
rewritten as follows:
vxvz vz
v xn  (v y  y  2v z  z )   x  v y  x   y  2
 h 
 v xn  2 z  v ny  (  y  2 y )  v zn
R R (1a)
 v y y   x  (v y x  f z )   y  2v z  n   s  f x  g x  v y  z

v y vz vz
v ny  v x  y   x  (v x  x  2v z  z )   y  2
 h  2 z  v xn 
 v ny  (  x  2 x )  v zn
R R (1b)
 (v x y  f z )   x  v x x   y  2v z  n   c  f y  g y  v x z

From the above equations, it can be noted that the driving noise of equation (1a) consists of f x , g x , and  v y  z ,

and the driving noise of equation (1b) consists of f y , g y , and  v x z . Obviously,  v y  z and  v x z are
correlated with each other. A similar situation can also be observed in the velocity error dynamics for the fine alignment.
According to Kalman Filtering theory, the correlated driving noises will make the filter suboptimal (Brown and Hwang
1997). This suboptimality of the filter will also affect the accuracy of the alignment.
To avoid the aforementioned two drawbacks, Sections 3 to 5 propose the new initial alignment scheme, in which the
coupling coefficients between the heading error and velocity error are enhanced and the problem of the correlated
driving noises is removed. Experimental tests in Section 6 show impressive improvement in both the accuracy and the
speed of the alignment by using this new scheme.

3. NEW APPROACH FOR IN-MOTION COARSE ALIGNMENT. Coarse alignment aims to estimate the initial
heading angle and to make the heading misalignment converge to a small angle. For strategic-grade and navigation-
grade INS, the coarse alignment can be achieved analytically on a stationary base by comparing the measured gravity
and Earth rate with the standard gravity and Earth rate (Savage, 2000). But for low-cost systems, the gyros used are
generally too noisy to measure the earth rotation, so the analytical approach cannot be used to obtain the initial heading
angle. Under the assumptions of small tilt misalignment and unknown heading misalignment, this paper develops a new
INS error dynamics model, which is used in the Kalman Filtering to estimate the initial heading angle for low-cost INS.
3.1 The Development of INS Error Dynamics Model for Unknown Heading Error. INS error dynamics with small
platform misalignments was discussed in Benson (1975). The position error dynamics with an unknown heading
misalignment remains the same as that with a small heading misalignment, because the position error dynamics does
not rely on small misalignment assumptions. The velocity error dynamics and platform misalignment dynamics are
redeveloped in this paper.
The detailed derivation of the velocity error dynamics and the platform misalignment dynamics with unknown
heading misalignment can be found in Appendix A. This development proceeds with the factorization of the direction
cosine matrix C cp :

C cp  C pp C cp (2)
where c represents the computed geographic frame, and p represents the platform frame which is a mathematically
virtual platform for strapdown INS, and p  represents a medium frame which is a local level frame as the c frame and
has the same heading misalignment as the p frame. The geometric relationship between the c frame and the p frame is
shown in Figure 1.

Figure 1. Geometric Relationship between the c frame and the p  frame ( y z is the unknown heading misalignment).

So the direction cosine matrix C cp  can be computed as follows:


 cos y z sin y z 0
 
C cp   sin y z 0

cos y z
  (3)
 1 
 0 0

Considering the tilt misalignment are small angles, define the tilt misalignment vector of the p frame with respect to the
p  frame as:

y xy  [ y x yy 0]T (4)

Because the p frame and the p  frame have the same heading misalignment, so the direction cosine matrix C p  can be
p

computed as:
 1 0 y y 
 
 I  y xy    0 y x 

C pp 1 (5)
y y x 1 
 y 

where y xy  represents the skew symmetric matrix of vector yxy . According to equations (2), (3), and (5), the

direction cosine matrix C cp can be computed as:


 cos yz sin yz yy 
 
Ccp 
 sin yz cos yz yx  (6)
 y cos y  y sin y yy sin yz  yx cos yz 1 
 y z x z 
When the heading misalignment is also a small angle, the following relationship holds:
C cp  I  y  (7)

where y is the small angle misalignment vector with elements y x , y y and y z , and (y) is the skew symmetric matrix

of vector y . This correspondence will be used to derive the observation equation in Subsection 3.2. The resulting INS
error dynamics with an unknown heading misalignment is summarized in equation (8).
 0 w ecz
c
 w ecy
c 
 1 0 0 0 0 0 0 
 
w ecz
c
0 w ecx
c
0 1 0 0 0 0 0 
 
 wc w ecx 0 
  ecy
c
0 0 0 1 0 0 0
 dr c   d rxc 
    
(2 w iez
x
   c  
c
 d ryc   0 0 0 0 w ecy
c
0  f zc fyc
f x   d ry c

    w ecz
c
  
 dr c   )
  d rzc 
    
   (2 w iez (2 w iex 
z c c
 d v xc   0 0 0 0 f zc 0  f xc f yc   d v xc 
    w ecz
c
 w ecx
c
  
   ) )
  dv y 
 dv y
d c c
  
dt    (2 w iex
c

 d v zc   0 0 0 w ecy
c
0  f yc f xc 0 0   d v c 
    w ecx
c
  z 
 yx   )
 yx 
   c  
 yy   ( w iez
c
 ( w iex   yy 
   0 0 0 0 0 0 0 w ecyc
  
  )   sin y 
 sin y z    w ecz  w ecx
c c
)
    z

( w iez ( w iex 
 cos y z  1  0 c   cos y z  1 (8)
c c

 0 0 0 0 0 0 w ecy 
  w ecz
c
)  w ecx
c
) 
 
 0 0 0 0 0 0 0 0 0 0 
 
 0 0 
 0 0 0 0 0 0 0 0
 0   0  0
     
 0   0  0
     
 0   0  0
     
  d g   0 
 x  x  
  d g   0 
 y  y  
   
 z  d g z   0 
     
 0   0  ex 
     
 0   0  e y 
     
 0   0  0
     
 0   0  0
     

3.2 Observations for Coarse Alignment. To estimate the initial heading with a Kalman Filter, external observations
should be provided. Herein, DGPS positions are used as the observations. However, the observation equation in this
development is different from that used in conventional GPS/INS integration because of the unknown heading
misalignment. This unknown heading misalignment will lead to different projection error from the body frame to the
navigation frame of lever arm..
The positions observed from GPS and computed from INS have the following relationship:
c
rGPS  rINS
c
 C bc l b (9)

where l b is the lever arm between the center of IMU and the antenna phase center of GPS. Impose a perturbation in

both sides of equation (9), and then the following relationship can be obtained:
d rGPS
c
 d rINS
c
 d C bc  l b  C bc  d l b (10)

where d rINS
c is the same as
d r c in equation (8); and dl b is the error of the lever arm which can be estimated during

Kalman Filtering; and dCbc is the perturbation of the attitude matrix, and its expression will be derived as follows.

The attitude matrix Cbc fulfills the following equation:

C bc  C cp C bp (11)

Supposing the error of Cbc is introduced only by the platform misalignment, so the perturbation of Cbc can be obtained
as follows:
d C bc  d C cp  C bp (12)

For small heading misalignment, the perturbation of C cp can be calculated as follows (Goshen-Meskin and Bar-Itzhack,
1992):
d C cp  C cp ( y) (13)

Insert equation (13) into equation (12) and expand Cbp as the multiplication of Ccp and Cbc , and then the following

relationship holds:
d C bc  C cp ( y) C cp C bc (14)

Considering the correspondence indicated by equation (7), the perturbation of the attitude matrix, dCbc , for unknown
heading misalignment is:
d C bc  C cp ( I  C cp )C cp C bc  ( I  C cp ) C bc (15)

Insert equation (15) into equation (10), and the observation equation for unknown heading misalignment can be
formulated as follows:
d rGPS
c
 d rINS
c
 ( I  C cp )C bc l b  C bc  d l b (16)
Ignore the two nonlinear items in equation (6), and insert it into equation (16), the matrix form of the observation
equation can be obtained as:
 drc 
 x 
 
 d ryc 
 
 drc 
 
 
z

d r c  1 l zc l yc c
l x   d v x
c  d l c   
 GPSx   0 0 0 0 0 0   x   wx 
 c     d   c  
0 c c
d rGPS
c
  d rGPSy  1 0 0 0 0 l zc 0 l xc l y  v   d l y    w y  (17)
 
y
      
d r c     dvc   c  w 
 GPSz  0 0 1 0 0 0 l yc l xc 0 0    d l z   z 
 
z
yx 
 
 y 
 y 
 
 sin y z 
 
 cos y z  1

where w is the measurement noise vector.

4. IN-MOTION FINE ALIGNMENT. Coarse alignment discussed in Section 3 will make the heading misalignment
converge to a small angle. Under the condition of small platform misalignment, including small tilt misalignment and
small heading misalignment, the fine alignment can proceed.
The following two approximations hold under the assumption of the small heading misalignment:
sin y z  y z (18a)
cos y z  1 (18b)

The error dynamics of small heading misalignment, yz , can be derived from equation (A.6) in Appendix A by using
equations (18a) and (18b) and by ignoring the high order small terms:
y z   y y w icx
c
 y x w icy
c
 ez (19)

Considering equations (18a), (18b) and (19), the INS error dynamics for small platform misalignment and the
corresponding observation equation can be obtained from equations (8) and (17), respectively, as follows:
 w ecz
c
 w ecy
c 
 0 1 0 0 0 0 0 
 
  w ecz
c
0 w ecx
c
0 1 0 0 0 0 
 
 wc  w ecx 0 

c
0 0 0 1 0 0
 
ecy
d r c   (2 w iez  c
c   d rx 
c
 x   w ecy
c
 f zc
 c  0 0 0 0 0 fy   
 d ry    w ecz
c
)   d ryc 
    
d r c    (2 w iez (2 w iex  c
c   d rz 
c c
 z  f zc  fx  
 c  0 0 0 0 0 
d v x    w ecz
c
)  w ecx
c
)   d v xc 
d     
d v c    0  (2 w iex
c  d v c 
dt    w ecy
c
 f yc f xc   
 c  0 
y y
0 0 0 0 
d v z    w ecx  dv c 
c
)
    z
y   ( w iez  
c   yx 
c
 x   w ecy
y   0 0 0 0 0 0 0 y 
 y   w ecz
c
)  y 
    
 y z    ( w iez
c
( w iex   y z 
c (20)
 0 0 0 0 0 0 0 
  w ecz
c
)  w ecx
c
)
 
  ( w iex
c 
 w ecy 0 

c
0 0 0 0 0 0
  w ecx
c
) 
 0   0  0
     
 0   0  0
     
 0   0  0
     
  d g   0 
 x  x  
     
  y   d g y    0 
     
 z  d g z   0 
     
 0   0  ex 
     
 0   0  e y 
     
 0   0  e 
     z

d r c 
 x
 c
 d ry 
 
d r c 
 z
d r c  1 l zc l yc   c  d l xc   w 
 GPSx   0 0 0 0 0 0
 
 c    d vx     x 
d rGPS
c
  d rGPSy  0 1 0 0 0 0 l zc 0 l xc   d v c   d l yc    w y  (21)
    y    
d r c  0 l xc 0   c  d l zc   w z 
 GPSz   0 1 0 0 0 l cy
  d vz   
y 
 x
y 
 y
 
 y z 

5. KALMAN FILTERING CORRECTIONS OF ALIGNMENT. Based on the INS error dynamics and the
observations discussed in Section 3 and Section 4, Kalman Filter is used to achieve the coarse and fine alignments.
After the INS errors are estimated by using the Kalman Filter, they should be used to correct the navigation results. The
correction approaches of position and velocity are the same for coarse and fine alignments, while the attitude correction
should be treated separately by using different methods for coarse and fine alignments.
The position correction can be achieved by the following steps: (a) Change the position representation from
longitude, latitude and height to the ECEF frame, and change the position error representation from the c frame to the
ECEF frame; (b) Correct the position error in the ECEF frame by using the following equation:
  
r ecef  rˆecef  d r ecef (22)
where rˆecef and r ecef are the position vectors before and after corrections in the ECEF frame, and d r ecef is the

position error vector resolved in the ECEF frame; (c) Change the position representation from ECEF frame to longitude,
latitude and height.
The velocity correction is more straightforward and can be achieved as follows:
  
v c  vˆc  d v c (23)
 
where vˆc and v c are the velocity vectors before and after the corrections in the c frame, and dv c is the velocity error

vector in the c frame.


The attitude correction of coarse alignment can be achieved by using the following approach. From equation (8), it

can be found that yx , y y , sin yz and cos yz 1 can be estimated during the coarse alignment Kalman Filtering, so the

direction cosine matrix from the c frame to the p frame, Ccp , can be computed according to equation (6). Then, the

attitude matrix, Cbc , can be computed as:

C bc  C cp Cˆbc (24)

where Cˆbc and Cbc are the attitude matrix before and after corrections.
For the fine alignment, yx , y y , and yz can be estimated from the alignment Kalman Filter. Because equation (7)

holds under the condition of small platform misalignment, during fine alignment C cp can be computed as:

C cp  I  ( y) (25)

Finally, the attitude correction equation of fine alignment can be obtained by inserting equation (25) into equation (24):
C bc  ( I  ( y)) Cˆbc (26)

6. EXPERIMENTAL TESTS. A field test was carried out to verify the performance of the proposed initial alignment
approach. In the experiment, CMIGITS II IMU (Martin and Detterich, 1997) was used to collect inertial data, and the
noise parameters of the inertial sensors are listed in Table 3. Leica Gx1230 receiver with Leica Leiat502 antenna was
used to record rover GPS data, and the base station receiver is a Leica Mc500 receiver with Leica Leiat504 Leis antenna,
which is located on the top of the Electrical Engineering building, UNSW.

Table 3. CMIGITS II IMU Noise Parameters.


Gyro Noise Accelerometer Noise
Parameters Parameters
Bias 30 deg/ h 4 mg
Bias Instability 5 deg/ h 500 m g
Random Walk 0.035 deg/ h 60 m g / hz

The field test was carried out in open places without GPS signal blockages, which is about 20km away from the
GPS base station. Two kinds of maneuvers were carried out to excite the errors of the inertial sensors and to accelerate
the alignment process: small ‘eight-like’movement, small circular movement.
In the remaining part, the experimental results from the proposed algorithm and the conventional algorithm
proposed in Rogers (1997), Kelley et al. (1990), and Pham (1992) are compared. The test trajectory is shown in Figure
2, where the computed trajectories from the conventional alignment algorithm, the proposed alignment algorithm, and
the DGPS reference are compared.
Figure 2. Trajectory Comparison by Using Different Approaches: DGPS Reference, the Conventional Algorithm, and
the Proposed Algorithm.

Figure 3 shows the position estimation errors during the alignment process. It can be noted that the position errors
converged to less than 0.5m in about 75 seconds for the proposed alignment algorithm, while it took 250 seconds for the
conventional algorithm.
Figure 4 shows the heading estimation errors during the alignment process. From the zero value initialization, the
estimated heading converged to the true heading. For the proposed algorithm, it took about 150 seconds to converge to
less than 20 arc minutes; while for the conventional algorithm, it took about 560 seconds to converge to the same error
level.
Figure 5 illustrates the tilt angles estimation errors. The tilt angles, namely the roll and pitch angles, are initialized
before the Kalman Filtering by using the gravity measurements in stationary base. Then the remaining errors of the tilt
angles are further eliminated by the coarse alignment and fine alignment. From Figure 5, it can be noted that the
convergences of the tilt angle errors by using the proposed algorithm are also much faster than that by using the
conventional algorithm.
Table 4 compares the alignment accuracy between the proposed alignment approach and the conventional alignment
approach. The alignment performance of the proposed approach has obvious improvement comparing with that of the
conventional approach.
Figure 3. North position errors (top), and east position errors (bottom) comparison between the proposed and the
conventional algorithms during the initial alignment.

Figure 4. Heading errors comparison between the proposed and the conventional algorithms during the initial
alignment.
Figure 5. Roll errors (top), and pitch errors (bottom) comparison between the proposed and the conventional
algorithms during the initial alignment.

Table 4. Alignment Accuracy of the Proposed and Conventional Algorithms.


Alignment Accuracy Alignment Accuracy Alignment Accuracy
of Heading of Pitch of Roll
Proposed Algorithm
5.6 5.3 3.8
(arcminutes)
Conventional Algorithm
15.5 14.6 19.2
(arcminutes)

7. CONCLUDING REMARKS. In this paper, a new two-stage Kalman Filtering mechanism has been developed for
the initial alignment of low-cost INS with the aiding of GPS. Based on this framework, the following novel work has
been completed:
 An INS error dynamics model accounting for unknown initial heading error has been developed for coarse alignment.
 The corresponding observation equation accounting for unknown initial heading error has been developed for coarse
alignment.
 An attitude error correction method of Kalman Filtering for coarse alignment has been proposed.
Experimental test results have demonstrated that this novel alignment mechanism can accelerate the alignment process
and improve the alignment accuracy.
ACKNOWLEDGEMENTS
The first author is sponsored by the Chinese Scholarship Council for his studies in Australia.

REFERENCES
Baziw, J.,and Leondes, C.T. (1972a). In-Flight Alignment and Calibration of Inertial Measurement Units-Part I: General
Formulation, IEEE Transactions on AES, 8(4), 439-449.
Baziw, J.,and Leondes, C.T. (1972b). In-Flight Alignment and Calibration of Inertial Measurement Units-Part II:
Experimental Results, IEEE Transactions on AES, 8(4), 450-465.
Bar-Itzhack, I.Y., and Porat, B. (1980). Azimuth Observability Enhancement During Inertial Navigation System In-
Flight Alignment,Journal of Guidance and Control, 3(4), 337-344.
Benson, D.O., (1975). A Comparison of Two Approaches to Pure-Inertial and Doppler-Inertial Error Analysis, IEEE
Transactions on AES, 11(4), 447-455.
Brown, R.G. and Hwang, P.Y.C. (1997). Introduction to Random Signals and Applied Kalman Filtering (3rd Edition).
John Wiley & Sons, Inc.
Dmitriyev, S.P., Stepanov,O.A., and Shepel, S.V., (1997). Nonlinear Filtering Methods Application in INS Alignment,
IEEE Transactions on AES, 33(1), 260-272.
Goshen-Meskin, D., and Bar-Itzhack, I.Y. (1992) Unified approach to inertial navigation system error modeling.
Journal of Guidance, Control, and Dynamics, 15(3): 648-653.
Greenspan, R.L., (1995). Inertial Navigation Technology from 1970-1995, Navigation: Journal of The Institute of
Navigation, 42(1), 165-185.
Groves, P.D., (2008). Principles of GNSS, Inertial, and Multisensor Integrated Navigation Systems. Artech House.
Han, S. and Wang J. (2008) Monitoring degree of observability in GPS/INS integration. Int. Symp. on GPS/GNSS,
Yokohama, Japan, 25-28 November, 414-421.
Kelley, R., Katz, I., and Bedoya, C.,(1990). Design, Development and Evaluation of an ADA Coded INS/GPS Open
Loop Kalman Filter, Proceedings of the IEEE 1990 National Aerospace and Electronics Conference, Vol.1, 382-388.
Kong, X., Nebot, E.M., and Durrant-Whyte, H., (1999). Development of a non-linear psi-angle model for large
misalingment errors and its application in INS alignment and calibration, Proceedings of the 1999 IEEE International
Conference on Robotics and Automation, Detroit, Michigan,May, 1430-1435.
Lee, H.Y., Wang J., Rizos, C. Grejner-Brzezinska, D. and Toth C. (2002) GPS/pseudolite/INS: Concept and first tests.
GPS Solutions, 6(1-2), 34-46
Shin, E., and El-Sheimy, N., (2004). An unscented Kalman filter for in-motion alignment of low-cost IMUs, Position
Location and Navigation Symposium, Monterey, CA, Aprial, 273–279.
Martin,M.K., and Detterich,B.C.,(1997). C-MIGITSTM II Design and Performance.Proceedings of the Satellite
Division of the Institute of Navigation 10th InternationalTechnical Meeting: ION GPS-97, September 16-19.
Pham, T.M. (1992). Kalman Filter Mechanization for INS Airstart, IEEE AES Systems Magazine, 7(1), 3-11.
Rogers, R.M., (1997). IMU In-Motion Alignment without Benefit of Attitude Initialization, Navigation: Journal of The
Institute of Navigation, 44(3), 301-311.
Savage, P.G., (2000). Strapdown Analytics. Strapdown Associates, Inc.
Wang, J., Lee, H.K., Hewitson S. and Hyung-Keun Lee (2003) Influence of Dynamics and Trajectory on Integrated
GPS/INS Navigation Performance, Journal of Global Positioning Systems, 2(2), 109-116,
(http://www.cpgps.org/journal.php)
Appendix A. INS ERROR DYNAMICS WITH UNKNOWN HEADING ERROR.
A.1. Attitude Error Dynamics. According to Benson (1975), the torquing rate of the mathematical platform of a
strapdown INS is:
w ipp  w icc  e (A.1)

where e is the drift vector of gyros.


The rotation rate of the platform frame p relative to the computed frame c can be expressed as follows:

w cp
p
 w ipp  w icp (A.2)

Insert equation (A.1) into equation (A.2), and then obtain:


w cp
p
 w ic
c
 w icp  e (A.3)

where wicp can be expressed as:

w icp  C cp w ic
c
(A.4)
Considering equation (A.4), equation (A.3) can be transformed into:
w cp
p
 ( I  C cp ) w ic
c
e (A.5)

According to equation (6), equation (A.5) can be expanded as:


w p 
 cpx   1  cos y z  sin y z y y   w icx   e x 
c
 p     
 c 
 w cpy    sin y z 1  cos y z y x  w icy   e y  (A.6)
   y sin y  y cos y  
 p  y x cos y z  y y sin y z 0   w c   e z 
 w cpz 
   x z y z   icz   

Because the tilt misalignment angles are small angles, so the following relationship holds (Benson, 1975):
y x  w cpx
p
 (1  cos y z ) w icx
c
 sin y z w icy
c
 y y w icz
c
 ex (A.7)

y y  w cpy
p
 sin y z w icx
c
 (1  cos y z ) w icy
c
 y x w icz
c
 ey (A.8)

Because yz is an unknown angle, sin yz and cos yz 1 may be large initially. But the change rate of sin yz and

cos yz 1 are small, so the following relationships will hold:


d sin yz
0 (A.9)
dt
d (cos yz 1)
0 (A.10)
dt
Combining equations (A.7) ~ (A.10), the INS attitude error dynamics with unknown heading error can be obtained as
follows:
   0 c 
yx w icz  w icy  w icx   yx  ex 
c c
     
   c 
  e y 
d  yy  w icz w icx w icy  yy
  
c c
0
 
dt  sin y z 
    (A.11)
 0 0   sin y z   0 
   0 0
 cos y z  1     
 0 0 0 0  cos y z  1  0 

From equation (A.11), it can be noted that sin yz and cos yz 1 are modeled as random constants during the coarse
alignment.
A.2. Velocity Error Dynamics. According to Benson (1975), the INS velocity error dynamics coordinatized in the
c frame is as follows:

d v c  f p  f c  d f p
 d g c  ( w iec  w icc )  d v c (A.12)
where f p can be expressed as:

f p
 Ccp f c (A.13)

Considering equation (A.13), equation (A.12) can be transformed into:


d v c   ( I  C cp ) f c  ( w iec  w ic
c
) dv c    d g c (A.14)

where  represents the error vector of accelerometers and is equal to d f p , and Ccp is computed in equation (6).

Equation (A.14) is the INS velocity error dynamics with unknown heading error. When substituting y
for ( I  Ccp ) , equation (A.14) is exactly the same as the Psi velocity error dynamics (Benson, 1975) with small heading

error.
According to equation (6), equation (A.14) can be expanded as follows:
dv c   cos y z  1 sin y z  y y   f x 
c
 x    
d   
d v y c     sin y z cos y z  1 y x   f yc 
dt    y sin y  y cos y  
dv c   y x cos y z  y y sin y z 0   f c 
 z   x z y z   z 
  (A.15)
(2 w iez  w ecz 2 w iey  w ecy c
  d v x    x   d g x 
c c c c
 0 )
   
  2 w iez
c
 w ecz
c
0 (2 w iex
c
 w ecx
c
)   d v y c     y    d g y 
  
 (2 w c  w c ) 2 w iex  w ecx   c      d g 
   d v z   z   z 
c c
0
 
iey ecy

Although the equation for dv z c includes a nonlinear item, considering the facts that both yx and y y are small quantities,
it can be approximated as:
d v z c  f xc ( y x sin y z  y y cos y z )  f yc (y x cos y z  y y sin y z )
 f xc ( y x sin y z  y y (cos y z  1))  f yc ( y x (cos y z  1)  y y sin y z )  f xc y y  f yc y x (A.16)
 f xc y y  f yc y x

Rearrange (A.15) by using (A.16), and then the INS velocity error dynamics with unknown heading error can be
obtained as follows:
 dvc 
 x 
 
 d v cy 
d vc   c 
(2 w iez
c
 w ecz
c
2 w iey
c
 w ecy
c
 f zc f yc fx    
 x  d v zc    x  d g x 
0 ) 0
d  c    
 d v y   2w iez
c
 w ecz
c
(2 w iex
c
 w ecx
c
f zc  f xc c     d g y  (A.17)

 yx   y
0 ) 0 fy
dt       d g 
 d v c  (2w c  w c ) 
 z   2 w iex
c
 w ecx
c
0  f yc f xc 0 0   yy   z z 
iey ecy  
 sin y z 
 
 cos y  1
 z 

You might also like