Method and apparatus for performing three-dimensional...

Details Method and apparatus for performing three-dimensional... Method and apparatus for performing three-dimensional...

1998-08-18

2001-05-22

Sheikh, Ayaz (Department: 2155)

Data processing: generic control systems or specific application

Specific application, apparatus or process

C701S205000, C701S300000, C701S301000, C700S090000, C706S010000, C706S047000

Reexamination Certificate

active

06236899

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
In general, the present invention relates to the computer-implemented method and apparatus for performing three-dimensional alpha/beta tracking. In particular, the method and apparatus of the present invention utilize a change in acceleration in order to determine a direction of movement along an alpha curve and a beta curve in a probability versus error plane. The change in acceleration of a track is used to introduce or dampen oscillations around the track in order to ensure that a tracking error does not grow exponentially, resulting in unacceptable expansion of its prediction windows, which brings in clutter and undesirable tracks.
2. Description of the Prior Art
In conventional tracking systems, if the probability of detection (P
D
) per scan is high, if accurate target location measurements are made, if the target density is low, and if there are only a few false alarms, the design of the correlation logic (i.e., associating detections with tracks) and tracking filter (i.e., filter for smoothing and predicting track positions) is straightforward. However, in a realistic radar environment these assumptions are seldom valid, and the design of an automatic tracking system is complicated. In actual situations, one encounters target fades (changes in signal strength due to multipath propagation, blind speeds, and atmospheric conditions), false alarms (due to noise, clutter, interference, and jamming), and poor radar parameter estimates (due to noise, unstabilized antennas, unresolved targets, target splits, multipath propagation, and propagation effects). An accurate tracking system must deal with all these problems.
Conventional tracking systems include contact entry logic, coordinate systems, tracking filter, maneuver-following logic, track initiation, and correlation logic. The simplest tracking filter is the alpha-beta (&agr;-&bgr;) filter described by
x
s
(
k
)=
x
p
(
k
)+&agr;[
x
m
(
k
)−x
p
(
k
)]  (1)
V
s
(
k
)=
V
s
(
k
−1)+&bgr;[
x
m
(
k
)−
x
p
(
k
)]/
T
  (2)
x
p
(
k
+1)=
x
s
(
k
)+
V
s
(
k
)
T
  (3)
where x(k) is the smoothed position, V
s
(k) is the smoothed velocity, x
p
(k) is the predicted position, x
m
(k) is the measured position, T is the scanning period (time between detections), and &agr; and &bgr; are the system gains.
The minimal mean-square-error (MSE) filter for performing the tracking when the equation of motion is know as the Kalman filter. The Kalman filter is a popular filter for radar and is a recursive filter which minimizes the MSE. The state equation in xy coordinates for a constant-velocity target is
S
(
t
+1)=&thgr;(
t
)+&Ggr;(
t
)
A
(
t
)  (4)
where with X(t) being the state vector at time t, consisting of position and velocity components x(t) , x(t), y(t); t
Γ
⁡
(
t
)
=
&LeftBracketingBar;
T
2
/
2
O
T
O
O
T
2
/
2
O
T
&RightBracketingBar;
⁢
 
⁢
and
⁢
 
⁢
A
(
t
)
=
&LeftBracketingBar;
a
x
(
t
)
a
y
(
t
)
&RightBracketingBar;
⁢
 
+1 being the next observation time; T being the time between observations; and a
x
(t) and a
y
(t) being random accelerations with covariance matrix Q(t) The observation equation is
Y
(
t
)=
M
(
t
)+
V
(
t
)  (5)
where
Y
⁡
(
t
)
=
&LeftBracketingBar;
x
m
⁡
(
t
)
y
m
⁡
(
t
)
&RightBracketingBar;
⁢
 
⁢
M
⁡
(
t
)
=
&LeftBracketingBar;
1
0
0
1
0
0
0
0
&RightBracketingBar;
⁢
 
⁢
and
⁢
 
⁢
V
⁡
(
t
)
=
&LeftBracketingBar;
v
x
⁡
(
t
)
v
y
⁡
(
t
)
⁢
 
(
6
)
with Y(t) being the measurement at time t, consisting of positions x
m
(t) and y
m
(t), and V(t) being a zero-mean noise whose covariance matrix is R(t).
The problem is solved recursively by first assuming that the problem is solved at time t−1. Specifically, it is assumed that the best estimate {circumflex over (X)}(t−1|t−1) at time t-1 and its error covariance matrix P(t−1 |t−1) are known, where the caret in the expression of the form {circumflex over (x)}(t|s) signifies an estimate and the overall expression signifies that X(t) is being estimated with observations up to Y(s). The six steps involved in the recursive algorithm are:
1. Calculate the one-step prediction
{circumflex over (X)}
(
t|t−
1)=&phgr;(
t−
1)
{circumflex over (X)}
(
t−
1
|t−
1)  (7)
2. Calculate the covariance matrix for the one-step prediction
 
P
(
t|t
−1)=&phgr;(
t
−1)
P
(
t
−1
|t
−1)&phgr;
T
(
t
−1)+&Ggr;(
t−
1)
Q
(
t−
1)&Ggr;
T
(
t−
1)  (8)
3. Calculate the predicted observation
Ŷ
(
t|t−
1)=
M
(
t
)
{circumflex over (X)}
(
t|t−
1)  (9)
4. Calculate the filter gain matrix
&Dgr;(
t
)=
P
(
t|t−
1)
M
T
(
t
)[
M
(
t
)
P
(
t|t−
1)
M
T
(
t
)+
R
(
t
)]
31
1  (10)
5. Calculate the new smoothed estimate
{circumflex over (X)}
(
t|t
)=
{circumflex over (X)}
(
t|t−
1)+&Dgr;(
t
)[
Y
(
t
)−
Ŷ
(
t|t−
1)]  (11)
6. Calculate the new covariance matrix
P
(
t|t
)=[
I
−&Dgr;(
t
)
M
(
t
)]
P
(
t|t−
1)  (12)
In summary, starting with an estimate {circumflex over (X)}(t|t−1) and its covariance matrix P(t|t−1), after a new observation Y(t) has been received and the six quantities in the recursive algorithm have been calculated, a new estimate {circumflex over (X)}(t|t) and its covariance matrix P(t|t) are obtained.
It has been shown that, for a zero random acceleration Q(t)=O and a constant measurement covariance matrix R(t)=R, the alpha-beta (&agr;-&bgr;) filter can be made equivalent to the Kalman Filter by setting
α
=
2
⁢
(
2
⁢
k
-
1
)
k
⁡
(
k
+
1
)
(
13
)
and
β
=
6
k
⁡
(
k
+
1
)
(
14
)
on the kth scan. Thus as time passes, &agr; and &bgr; approach zero, applying heavy smoothing to the new samples. Usually it is worthwhile to bound &agr; and &bgr; from zero by assuming a random acceleration Q(t)≠O corresponding to approximately a 1-g maneuver.
The predicted position of velocities generated by an alpha/beta tracker can divergently oscillate about a track true position from scan to scan resulting in the loss of a track. These divergent oscillations can be caused by the uncertainties in a track position in velocity, or by linear or angular accelerations in time. Conventional alpha/beta trackers inherently do not fully account for acceleration, if they account for acceleration at all. When these divergent oscillations cyclically deviate from the predictions, the alpha/beta tracker will divergently over- and under-compensate for its errors from scan to scan. In this situation, a small error exponentially grows into a gross error, resulting in an unacceptable expansion of prediction windows, thereby bringing in clutter and undesirable tracks.
SUMMARY OF THE INVENTION
The present invention solves this problem by providing a method and apparatus in which a change in acceleration of a track is used to introduce or dampen oscillations around the track.
In the present invention, the alpha curve, which represents the position of the track, is applied to an error between the predicted and reported predictions to determine a track's current position. The beta curve, which represents the velocity of the track, is applied similarly to the error in velocity. The index into the alpha/beta curve is usually a constant or adjusted proportionally with the error. When the absolute values of the errors in the multiple scans begin to grow, the independent variable of the alpha/beta curves are finely adjusted in the opposite direction in its traversal along the curve. The amount of adjustment is further interpolated to r

Affiliated with

Also associated with

Rating

Method and apparatus for performing three-dimensional... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Method and apparatus for performing three-dimensional..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for performing three-dimensional... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2491013