Professional Documents
Culture Documents
net/publication/221077764
Conference Paper in Proceedings - IEEE International Conference on Robotics and Automation · January 1999
DOI: 10.1109/ROBOT.1999.774048 · Source: DBLP
CITATIONS READS
88 457
2 authors:
Some of the authors of this publication are also working on these related projects:
Navigation and control of a space robot capturing moving target View project
All content following this page was uploaded by J. Z. Sasiadek on 09 April 2015.
297 1
Computing the U posteriori covariance matrix is something wrong with the design and the filter
gives: is not performing optimally [4]. The Kalman
filters will diverge or coverage to a large bound.
In practice, it is dificult to know the exact value
:P = (I - K,H,)P,"- (14) for Q and R. In order to reduce computation, we
The initial condition is: have to ignore some errors, but sometimes those
unmodeled errors will become significant. These
p,"- =Po are the instrument bias errors of INS. Generally
the Wk does not always have zero mean. In those
cases, the residuals can be used to adapt the
In equation (lo), the term a-& is called filter. In fact, the residuals are the differences
residuals or innovations. It reflects the degree to between actual measurements and best
which the model fits the data. measurement predictions based on the filter's
internal model. A well-tuned filter is that where
the 95% of the autocorrelation function of
INS and GPS innovation series should fall within the 2 2a
boundary [7].If the filter diverges, the residuals
The inertial navigation system (INS) consists will not be zero mean and become larger.
of a sensor package, which includes There are very few papers on application of
accelerometers and gyros to measure fuzzy logic to adapt the Kalman filter. In (81,
accelerations and angular rates. By using these fuzzy logic is used to the on-line detection and
signals as input, the attitude angle and three- correction of divergence in a single state Kalman
dimensional vectors of velocity and position are filter. There were three inputs and two outputs to
computed [SI.The errors in the measurements of fuzzy logic controller (FLC), and 24 NI= were
force made by the accelerometers and the errors used. The purpose of our fuzzy logic adaptive
in the measurement of angular change in system (FLAS) is to detect the bias of
orientation with respect to inertial space made by measurements and prevent divergence of the
gyroscopes are two fundamental error sources, extended Kalman filter. It has been applied in
which affect the error behavior of an inertial three axes - East (x), North (y), and Altitude
system. The inertial system error response, (z). The covariance of the residuals and the mean
related to position, velocity, and orientation is of residuals are used as the inputs to FLAS for
divergent with time due to noise input [6]. There all three fuzzy inference engines. The
are biases associated with the accelerometers and exponential weighting a for three axes are the
gyros. In order to correct the errors of INS, the outputs. As an input to FLAS, the covariance of
GPS measurements are used to estimate the the residuals and mean values of residuals are
inertial system errors, subtract them from the used to decide the degree of divergence. The
INS outputs, and then obtain the corrected INS value of covariances relates to R. The equation
outputs. There is number of errors in GPS, such for covariance of the residual is:
as ephemeris errors, propagation errors, selective
availability, multi-path, and receiver noise, etc. P, = H,P;Hi + R
By using differential GPS (DGPS), most of the
errors can be corrected, but the multi-path and
receiver noise cannot be eliminated. When the Kaliilan filter is performing
optimally, the mean values of residuals are near
zero. Generally, when the covariance is
becoming large, and mean value is moving away
from zero, the K a h n filter is becoming
Fuzzy Logic Adaptive System unstable. In this case, a large a will be applied.
A large a means that process noises are added. It
It is assumed that both, the process noise wk can ensure that in the model all states are
and the measurement noise vk are zero-mean sufficiently excited by the process noise. When
white sequences with known covariance Q and R the covariance is extremely large, there are some
in the Kalman filter. If the Kalman filter is based problems with the GPS measurements, so the
on a complete and perfectly tuned model, the filter cannot depend on these measurements
residuals should be a zero-mean white noise anymore, and a small a will be used. The perfect
process. If the residuals are not white noise, there measurements are given more weighting. By
2972
selecting appropriate a,the FLAS will adapt the
Kalman filter optimally and try to keep the Table. 1. Rule Table for FLAS
innovation sequence acting as zero-mean white
noise.
I I Mean Value
I
small large
large
Simulation
M A W codes developed by authors has
been used to simulate and test the proposed
method.
Fig.3. Mean Value Membership Functions
The state variables used in simulation are:
$m small meciium large
2973
second simulation (Fig. 6), the same [2] Abidi, Mongi A. and Gonzalez, Rafael C. “
measurements as in the first simulation for INS Data Fusion in Robotics and Machine
were used. A white noise with a standard Intelligence ”, Academic Press, Inc., 1992.
deviation of 2 meter from 0 s to loo0 s and 1500
s to 2oooS was applied to GPS measurements. [3] Gelb, A., “ Applied Optimal Estimation ”,
From loo0 s to 1500 s, the standard deviation of The MIT Press, 1974.
6 meter with mean value of 6 meter was added to
GPS measurements. Although, the GPS [4] Lewis, F. L., “ Optimal Estimation with an
measurement noises features were changed, the Introduction to Stochastic Control Theory ”,
fuzzy adaptive EKF still worked well. Those John Wiley & Sons, Inc., 1986.
simulations also showed that the corrected errors
of EKF were proportional to the mean values of [5] Jochen, P., Meyer-Hilberg, W., and
INS measurements. In other word, the more ‘Ibomas, Jacob. “High Accuracy Navigation
errors are not modeled, the worse the EKF and Landing System using GPSAMU
performs. System Integration”, 1994 IEEE Position
Location and Navigation Symposium,
pp.298 - 305.
References
[ l ] Brown, R. G. and Hwang, P.Y.C. ‘‘
Introduction to Random Signals and Applied
Kalman Filtering ”, John Wiley & Sons,
Inc., 1992.
2974
Fuzzy adaptive E K F EKF
20 50
v I I I v I I I
C
0
._
G O
I I I 0 I I I
I I I
-c -50
0 500 1000 1500 2000
0.5 I I
I I I I I I
1 J
Fig.5. Simulation 1
Fuzzy adaptive E K F E KF
h h
c
L
2
-
0, s 50,
v
C
0
._
U
2 0
0 I I I 0 I I I
C I I I c
C I I I
I
-
-...
,>y---"*----
,
-
E- 0
zi 1
I
I
I
I
I
-10 I
I
, t
I I
I J "
0 500 1000 1500 2000 0 500 1000 1500 2000
Tim e (s) Tim e (s)
Fig.6. Simulation 2
2975