Professional Documents
Culture Documents
INS/GPS Navigator
December, 2001
Purdue University
December,2001 1
Model and Parameters to Drive Simulation
Trajectory Input
Aircraft
Turbulence Input
Model
Errors
Processing Mode
Antennas GPS INS
Number, Location
Filter
December,2001 2
Outline
1. Overview
2. Structure of Simulation
3. Simulation Models
4. Kalman Filter
5. Initial Conditions
Error Source Specifications
6. Results
7. Conclusions
December,2001 3
Overview
(1)UAV Dynamics
Nominal Trajectory
(2) Navigation Equation
INS Output
(3) Tightly Coupled INS/GPS
INS/GPS Output
Covariance Data
(4) Covariance data is passed to
Imagery Analysis
December,2001 4
Structure of Simulation
Tightly Coupled
INS/GPS
INS Position
Velocity
UAV IMU Nav
Orientation
Position, Velocity, Orientation
and Covariance correction Covariance
Bias Correction -
Kalman
Filter
+
GPS Receiver
December,2001 5
Simplified IMU Model
~
x x x
where
x = Bias + White Noise
~x : Sensor Output
x : Sensor Input
Bias : Markov Process, tc=60s
for all
ax ,ay,az Accelerometer Outputs
x
x , y , z Rate Gyro Outputs
December,2001 6
GPS Receiver Model
Pseudorange
GPS (X x) 2 (Y y) 2 (Z z) 2 c t
Pseudorange Rate
d d dt d
c
dt GPS dt dt dt
where
X, Y, Z : Satellite Position
x , y, z : Platform Position
c t : Pseudorange equvalent
Clock Bias (Random Walk)
dt
c : Pseudorange rate equivalent
dt
Clock Drift (Random Walk)
: Normally Distributed Random Number
d
: Normally Distributed Random Number
dt
December,2001 7
Kalman Filter: Error Dynamics
d
x F x Gv
dt
December,2001 8
Kalman Filter: Output Equation
GPS X x (Y y) 2 ( Z z) 2 ct
2
d d dt d
c
dt GPS dt dt dt
INS X x INS (Y y INS ) 2 ( Z z INS ) 2 ct INS
2
d d dt
c
dt INS dt INS dt INS
Measurement: Z d Random Noise: d
dt dt
Output Equation:
ZGPS ZINS H x
where H
x
December,2001 9
Initial Error Condition
Initial Errors
[0 , 0 , 0 ] so that
[E10 , E 2 0 , E30 ] [0.001,0.001,0.002] rad
[v N , v E , v D ]0 [0.1,0.1,0.1] m/s
[, ]0 [1.57e 7,1.57e 7] rad
h 0 1 m
[ Bx , By , Bz ]0 [ B , B , B ] rad / s
[ Bax , Bay , Baz ]0 [ Ba , Ba , Ba ] g
[ b, d ]0 [ S b , Sd ] [ m, ( m / s)]
Initial Covariance Values
P0 (, , ) [02 , 02 , 02 )] rad 2
P0 (v N , v E , v D ) [v 2N 0 , v 2E 0 , v 2D 0 ] ( m / s) 2
i
2
P0 ( Bi ) ( rad / s) 2
ai
2
P0 ( Bai ) g2
P0 ( b, d ) [Sb , Sd ] [ m 2 , ( m / s) 2 ]
December,2001 10
Error Source Specifications
INS
Rate Gyros
Bias () B 0.003 0.35 deg/hr
White Noise (sqrt(PSD)) N 0.0015 0.07 deg/sqrt(h r)
(deg/hr/sqrt(Hz))
(good) (worse)
December,2001 11
Error Source Specifications
GPS
GPS Receiver Notation Receiver 1 Receiver 2 Units
Pseudorange () r 6.6 33.3 m
Pseudorange Rate () rr 0.05 0.5 m/s
ClockBias White Noise(PSD) Sb 0.009 0.009 m2
ClockDrift White Noise(PSD) Sd 0.0355 0.0355 ( m / s) 2
(good) (worse)
December,2001 12
Satellite Geometry during the Simulation
December,2001 13
Local Frame: x, y, z
Zecef
Nominal Trajectory
v x 61 (m / s)
(200 ft / s) x
h 0 6096 (m)
(20000 ft )
y
Yecef
z
Xecef x=Zecef
y=-Yecef
z=Xecef-6378137m
December,2001 14
Result 1:Comparisons between INS/GPS
and Unaided INS;(Good INS,Good GPS)
Local Frame Position Errors: (true) (estimated)
Local Frame Position Errors
1000
500
dx (m) dx (m)
0
-500
0 50 100 150 200 250 300 350 400
1000
INS/GPS Error
500
dy (m)
dy (m) 0
-500
0 50 100 150 200 250 300 350 400
30
20
dz (m)
dz (m) 10
-10
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
INS/GPS works very well
December,2001 15
Result 1:Comparisons between INS/GPS
and Unaided INS;(Good INS,Good GPS)
Local Frame Velocity Errors: (true) (estimated)
Local Frame Velocity Errors
4
dvx (m/s)
2
dv x (m / s) 0
-2
0 50 100 150 200 250 300 350 400
4
dvy (m/s)
2 INS/GPS Error
Unaided INS Error
dv y (m / s) 0
-2
0 50 100 150 200 250 300 350 400
0.3
0.2
dvz (m/s)
0.1
dv z (m / s) 0
-0.1
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
INS/GPS works very well
December,2001 16
Result 1:Comparisons between INS/GPS
and Unaided INS;(Good INS,Good GPS)
Local Frame Euler Angle Errors: (true) (estimated)
-3
x 10 Euler Angle Errors
1
dE1 (rad)
0
droll (rad) -1
-2
0 50 100 150 200 250 300 350 400
-4
x 10
5 INS/GPS Error
Unaided INS Error
0
dE2 (rad)
-5
-15
0 50 100 150 200 250 300 350 400
-3
x 10
0
dE3 (rad)
-1
dyaw (rad) -2
-3
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
Roll and Pitch errors are quickly corrected
Yaw error correction takes time
Effect on Geo Positioning?
December,2001 17
Result 2:Ensembles
(Good INS,Good GPS)
Local Frame Position Errors: (true) (estimated)
Local Frame Position Errors
0
-0.5
dx (m) dx (m)
-1
-1.5
-2
0 50 100 150 200 250 300 350 400
0
dy (m)
dy (m) -2
-4
0 50 100 150 200 250 300 350 400
1
dz (m)
dz (m) -1
-2
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
Position error is less than 3m LN-100G:10mCEP
Error value is not 0 mean locally
December,2001 18
Result 2:Ensembles
(Good INS,Good GPS)
Local Frame Velocity Errors: (true) (estimated)
Local Frame Velocity Errors
0.05
dvx (m/s)
0
dv x (m / s) -0.05
-0.1
0 50 100 150 200 250 300 350 400
0.05
dvy (m/s)
dv y (m / s) -0.05
-0.1
0 50 100 150 200 250 300 350 400
0.1
dvz (m/s)
0.05
dv z (m / s) 0
-0.05
-0.1
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
Velocity error is less than 0.05m/s
LN-100G:0.015m/s(rms)
December,2001 19
Result 2:Ensembles
(Good INS,Good GPS)
Local Frame Euler Angle Errors: (true) (estimated)
-3
x 10 Euler Angle Errors
1
dE1 (rad)
0
droll (rad) -1
-2
0 50 100 150 200 250 300 350 400
-3
x 10
1
dE2 (rad)
dpitch (rad) -1
-2
0 50 100 150 200 250 300 350 400
-3
x 10
1
0
dyaw (rad)
dE3 (rad)
-1
-2
-3
0 0 50 100 150
400 (sec)
200
time (s)
250 300 350 400
5
dx (m)
dx (m) 0
-5
0 50 gIgG 100 150 200 250 300 350 400
wIgG
10 wIwG
gIwG
5
dy (m)
dy (m) 0
-5
0 50 100 150 200 250 300 350 400
0
dz (m)
dz (m) -5
0 -10
0 50
400 (sec)
100 150 200
time (s)
250 300 350 400
0.2
dv x (m / s) dvx (m/s)
0.1
0
gIgG
wIgG
-0.1
0 50 100 150 200 250 300 wIwG
350 400
gIwG
0.2
0.1
dvy (m/s)
dv y (m / s) 0
-0.1
-0.2
0 50 100 150 200 250 300 350 400
0.3
0.2
dvz (m/s)
dv z (m / s) 0.1
-0.1
0 50 100 150 200 250 300 350 400
0 400 (sec)
time (s)
December,2001 22
Result 3: Comparisons between 4patterns
blue red (:) black (-.) green (--)
INS good worse worse good
GPS good good worse worse
Local Frame Euler Angle Errors: (true) (estimated)
-3
x 10 Euler Angle Errors
1
dE1 (rad)
0
droll (rad) -1
gIgG
wIgG
-2
0 50 100 150 200 250 300 wIwG
350 400
x 10
-4 gIwG
5
0
dE2 (rad)
-15
0 50 100 150 200 250 300 350 400
-3
x 10
0
dyaw (rad)
dE3 (rad)
-1
-2
-3
0 50 100 150 200 250 300 350 400
time (s)
0 400 (sec)
INS accuracy helps orientation accuracy
December,2001 23
Conclusions
December,2001 25
References
(INS)
[1] Titterton, D. H. and Weston, J. L. (1997). Strapdown Inertial
Navigation Technology. Peter Peregrinus Ltd.
[2] Rogers, R. M. (2000). Applied Mathematics In Integrated
Navigation Systems. AIAA Education Series.
[3] Chatfield, A. B. (1997). Fundamentals of High Accuracy
Inertial Navigation. Volume 174, Progress in Astronautics and
Aeronautics. AIAA.
[4] Britting, K. R. (1971). Inertial Navigation Systems Analysis.
Wiley Interscience.
(Kalman Filter)
[5] Brown, R. G. and Hwang, P. Y. C. (1985). Introduction to
Random Signals and Applied Kalman Filtering. John Wiley &
Sons.
[6] Gelb, A. (1974). Applied Optimal Estimation. M.I.T. Press.
December,2001 26
References (Cont.)
(Navigation Sensors)
[7] B. Stieler and H. Winter (1982). Gyroscopic Instruments and
Their Application to Flight Testing. AGARDograph No.160
Vol.15.
[8] Lawrence, A. (1992). Modern Inertial Technology. Springer-
Verlag.
[9] IEEE Standard Specification Format Guide and Test Procedure
for Single-Axis Laser Gyros. IEEE Std. 647-1995.
(GPS)
[10] Kaplan. E. D. (1996). Understanding GPS Principles and
Applications. Artech House.
(Others)
[11] Military Standard for Flying Qualities of Piloted Aircraft 1797A.
[12] Department of Defense World Geodetic System 1984, Its
Definition and Relationships with Local Geodetic Systems,
National Imagery And Mapping Agency Technical Report
December,2001 27
Kalman Filter:Output Equation
h1 h1 h1
x 1 x e
y z
y
1 1 0 e
h k h k h k
1 k4 z e
x y z b
k
k x
1 1 h1 h1 h1
1 e
y e
x y z
k GPS
k INS 0 z e
h k h k h k
k4 d
1 ~
x
x y z ~y
~
z
H1
k : Number of Visible Satellites
x e , y e , z e : Platform ECEF Coordinate s
h i Xi x
x (X i x ) 2 (Yi y) 2 ( Zi z) 2
December,2001 28
Kalman Filter:Output Equation
x x x
h
x e y y y
0 0
y e 0 0
z 0 0 h 0 0 0 0
e 43 43 z z z 43 43
b
x e h
1 0
x
y e 0 0 0
z 0 0
e NED 0
d 0 T ECEF 0 0 0 0
43 0 0
0 0 0
43 43 43 0 1
H2
x [, , , v N , v E , v D , , , h ,
Bx , By , Bz , Bax , Bay , Baz , b, d ]
December,2001 29
Simplified IMU Error Model
~
a a a
where 0
Sx M xy M xz a x B x D x
a M yx S y M yz a y B y D y
M zx M zy Sz a z B z D z
~
a : Sensor Output
a : Sensor Input
Bi : Bias (MarkovProc ess)
B Bias STD for ensembles
25 g for Accelerome ters
0.003 deg/ hr for Gyros
Di : White Noise
PSD D (5 g ) 2 for Accelerome ters
(0.0015 deg/ hr ) 2 for Gyros
December,2001 30
Clock Error Model
b d D b
d Dd
b ct : Clock Bias
d ct : Clock Drift
D b White Noise , PSD 0.009 m 2
D d White Noise , PSD 0.0355 (m / s) 2
Updating & Propagation in the Kalman Filter
b k b k
K k ( ZGPS Z INS )
d k d k
b k 1 1 T b k
0 1
d k 1 d k
December,2001 31