JP3432737B2 - Tracking device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a target tracking device using a radar sensor, and more particularly to a tracking device which improves tracking accuracy without causing a false correlation.

[0002]

2. Description of the Related Art Conventionally, in order to simultaneously realize "securing a good followability with respect to a turning target" and "high-precision tracking with respect to a straight ahead target", a tracking filter defined by N motion models is constructed to obtain a smooth state. A tracking device was integrated to integrate the two in accordance with the probability.

As such a conventional technique, the following FIG.
7 and FIG. 18. Figure 17
A. Houles and Y. Bar-Shalo
m, "Multisensor tracking o
f a maneuvering target in
clutter ", IEEE Aerosp. & E
electron. Syst. FIG. 18 is a block diagram showing the configuration of the tracking device described in 1989, and FIG. 18 is a flowchart showing the operation of the tracking device shown in FIG.

First, the structure of the conventional tracking device will be described with reference to FIG. For the purpose of explanation, the target observation value vector at the sampling time tk is Ck, and the tracking target state variable vector defined for each motion model is Bk, a.
(A = 1, 2, ..., N), the state variable vector of the tracking target obtained by integrating Bk, a is set as Bk.
Note that Ck and Bk are expressed in the north reference Cartesian coordinate system.

In FIG. 17, 1 is an observing means for outputting an observation value vector of a tracking target, and 21 to 22 are prediction vectors Bk, a (-) as prediction states of N targets based on N motion models. Prediction means for outputting the error covariance Pk, a (-), 31 to 32 input N prediction states,
Gate calculation means for calculating N kinds of gates and gate volumes, 41 is gate selection means for inputting N gate volumes and selecting the largest volume as a gate, 51-
52 is an observation value vector output from the observing means, and a smoothing value vector Bk, a as a smoothing state for each of N prediction states.
(+) And smoothing error covariance Pk, a (+) are output as smoothing means, 6 is smoothing value integrating means for integrating N kinds of smoothing states, and 71 to 72 are delay elements.

The above means will be described in more detail below. First, the observation means 1 outputs the target observation value vector Ck at the sampling time tk. Where C
k is composed of position information of the tracking target. The predicting means 21 to 22 input the smoothed state for each of the N motion models which is the output of the smoothed value integrating means 6, and predict the predicted value vector Bk, a (-) and the predictive error covariance as the predicted state of a unit time ahead. Output Pk, a (-).

The gate calculating means 31 to 32 are the predicting means 2
The predicted value vector Bk, a (-) and the prediction error covariance Pk, a (-), which are the outputs of 1 to 22, are input, and the gate centers Ck, a (-) and the error covariance matrix Sk, for each motion model. a
(-) And the gate volume Vm, a are output.

The gate selecting means 41 is a gate calculating means 3
The N gate specifications obtained for each motion model, which are the outputs of 1 to 32, are input, and the gate having the largest gate volume is adopted and output.

The smoothing means 51 to 52 input the observed value vector Ck of the tracking target at the sampling time tk and N prediction states for each motion model which is the output of the prediction means,
N smoothed value vectors Bk, a (+) for each motion model
And a smooth error covariance matrix Pk, a (+) is output.

The smoothing value integrating means 6 inputs N smoothing states for each motion model, which is the output of the smoothing means, and weights them by the probability, thereby smoothing value vector Bk (+) and smoothing error covariance matrix Pk ( +) Is output.

Next, the operation of the conventional tracking device shown in FIG. 17 will be described with reference to FIGS. 17 and 18, particularly the flowchart of FIG. First, the operation is started, and step ST
1, the prediction means 21 to 22 are smoothing means 5 described later.
Input the N smoothing states for each motion model, which are the outputs of 1 to 52, and use the predicted value vector B as the predicted state at a unit time ahead.
k, a (-) and prediction error covariance Pk, a (-) are output for each motion model.

In step ST12, the gate calculation means 31 to 32 input N prediction states for each motion model which are the outputs of the prediction means 21 to 22, and the gate centers Ck, a.
(−), Error covariance matrix Sk, a (−), and gate volume Vm, a are calculated.

Next, in step ST31, the gate volume Vm calculated by the gate selecting means 41 for each motion model,
a is compared, and the one with the largest volume is selected as the tracking gate and output.

In the next step ST4, the observation means 1
The target is observed by and the observation value vector Ck is output.
The observation value vector Ck is input to the smoothing means 51 to 52.

Further, in step ST5, the smoothing means 5
1 to 52 input the observed value vector Ck from the observation means 1 and the prediction state (Bk, a (−), Pk, a (−) above) calculated for each motion model by the prediction means 21 to 22, A smooth value vector B for each motion model as the target smooth state
k, a (+) and smoothing error covariance Pk, a (+) are output.

Finally, in step ST6, the smoothing value integrating means 6 performs the smoothing state (Bk, a (+), Pk, a above) for each motion model obtained by the smoothing means 51-52.
(+)), The probability that each motion model calculated before the unit time is correct, and the transition probability Pab between motion models are input, and the smoothed states of N motion models are integrated and processed. Ends. If the processing is not completed, the process returns to step ST1.

[0017]

However, the above-described conventional tracking device tries to realize both "good tracking performance for a turning target" and "high tracking accuracy for a straight ahead target" at the same time by the above-mentioned configuration. . However, in this configuration, the prediction states are not integrated, and the tracking gate cannot be uniquely determined. Therefore, since the gate having the largest volume is selected, there is a problem in that there is a high possibility that the gate is made larger than necessary to cause a cross correlation.

The present invention has been made in order to solve the above-mentioned conventional problems, and integrates the predicted states calculated based on the respective motion models according to the goodness of fit of the motion models, thereby causing no cross correlation. It is an object of the present invention to provide a tracking device that simultaneously improves the tracking performance for a turning target and the tracking accuracy for a straight ahead target.

[0019]

A tracking device according to a first aspect of the present invention is a tracking device for a turning or straight ahead target.
In the above , the prediction means for outputting the respective prediction states based on the plurality of motion models and the plurality of prediction states are integrated.
And output the predicted value integration means, the gate calculation means for inputting the integrated prediction state and outputting the area where the next observed value vector is obtained, and the tracking target in the area is observed and the observed value vector is output. Observing means, a smoothing means for inputting an observed value vector from the observing means and a plurality of prediction states from the predicting means, and outputting a plurality of target smooth states based on each motion model; The smooth states
Includes a smoothing value integrating means for integrating and outputting, a plurality
When combining the prediction states of
It is characterized in that each variance matrix is weighted and integrated according to the fitness of each motion model.

In the tracking device according to the invention of claim 2, the number of the motion models is set to 1 in accordance with the sampling interval.
It is provided with a motion model control means for switching to one or more .

In the tracking device according to the invention as defined in claim 3, the number of the motion models is one or, depending on the tracking time.
A second motion model control means for switching to a plurality of units is provided.

According to a fourth aspect of the present invention, a tracking device is provided with a constant input setting means for defining a plurality of target motion models and setting N kinds of independent same-dimensional constant input vectors. is there.

A tracking device according to a fifth aspect of the present invention is provided with an observation matrix calculating means for processing the observation information in consideration of the arrangement positions of a plurality of radars having different arrangement positions.

A tracking device according to a sixth aspect of the invention is a hypothesis setting means for performing a correlation process of which signal is a target signal when a plurality of observation signals are generated due to the disappearance of unnecessary signals. The correlation means is provided.

The tracking device according to the invention of claim 7 is a second hypothesis setting means for performing correlation processing of which observation signal is a signal from an already tracked target in a multi-target environment in which a plurality of targets exist. And a second correlation means.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION The accompanying drawings, FIGS.
Based on this, an embodiment of the present invention will be described in detail. Figure 1
2 is a block diagram showing the configuration of the tracking device according to the first embodiment of the present invention, FIG. 2 is a flowchart showing the operation of the tracking device shown in FIG. 1, and FIG. 3 is a configuration of the tracking device according to the second embodiment of the present invention. 4 is a block diagram showing the operation of the tracking device shown in FIG. 3, FIG. 5 is a block diagram showing the configuration of the tracking device according to Embodiment 3 of the present invention, and FIG. 6 is the tracking shown in FIG. FIG. 7 is a flowchart showing the operation of the apparatus, FIG. 7 is a block diagram showing the configuration of the tracking apparatus according to Embodiment 4 of the present invention, FIG. 8 is a flowchart showing the operation of the tracking apparatus shown in FIG. 7, and FIG. FIG. 10 is a diagram showing a motion coordinate system used in the fourth embodiment, and FIG. 10 is a fourth embodiment.
FIG. 11 is a block diagram showing the configuration of a tracking device according to the fifth embodiment of the present invention, FIG.
12 is a flowchart showing the operation of the tracking device shown in FIG. 11, FIG. 13 is a block diagram showing the configuration of the tracking device according to the sixth embodiment of the present invention, and FIG. 14 shows the operation of the tracking device shown in FIG. FIG. 15 is a flowchart, FIG. 15 is a block diagram showing a configuration of a tracking device according to Embodiment 7 of the present invention, and FIG. 16 is a flowchart showing an operation of the tracking device shown in FIG.

First, the reference numerals for describing the embodiment will be described. The target observation value vector at the sampling time tk is Ck, and the state variable vector based on the motion model a is Bk, a (a = 1, 2 ,. , N), the state variable vector of the tracking target obtained by integrating the state variable vectors Bk, a is set as Bk . Ck and Bk are expressed in the north reference Cartesian coordinate system.

Embodiment 1. First, the configuration of the tracking device according to the first embodiment of the present invention will be described with reference to FIG. In FIG. 1, reference numeral 1 is composed of a sensor such as a radar for monitoring a target, and observation means 51 to 52 for outputting a distance r to a tracking target and an azimuth angle u1 and an elevation angle u2 are connected to the observation means 1. Smoothing means for inputting the observed value vector Ck of the target and the output of the predicting means and outputting the smoothing state of the tracking target based on each motion model (note that there are a plurality of equivalent means between these marks (N marks)).
, 71 and 72 are delay elements,
Numerals 21 to 22 are input with the smoothed value vector Bk (+) and the smoothed error covariance matrix Pk (+), which are the outputs of the smoothed value integrating means 6, and the predicted value vector Bk for a single time interval based on each motion model. , A (−) and the prediction error covariance matrix Pk, a
It is a prediction means that outputs (-).

Further, 6 is a smoothing state Bk, a (+), Pk, for each motion model obtained by the smoothing means 51-52.
a (+), likelihood Λk, a, and transition probability P between motion models
The smooth value integrating means 21 to 22 which inputs ab and calculates the goodness of fit βk, a of each motion model is a smooth value vector Bk (+) output from the smooth value integrating means 6 and the smooth error covariance matrix Pk ( +) (Which will be described later) is input to predict the predicted value vector Bk, a for a unit time ahead based on each motion model.
It is a prediction means that outputs (-) and the prediction error covariance matrix Pk, a (-).

Further, 3 is a prediction value vector Bk, a (-) for each motion model which is an output of the prediction means 21 to 22, a smooth error covariance matrix Pk, a (-), and a fitness degree β of the motion model.
k, a and the transition probability Pab are input, the prediction states based on the respective motion models are integrated, and the integrated prediction state, that is, the prediction value vector Bk (−) and the prediction error covariance matrix Pk (−) are output. The predictive value unifying unit 4 inputs the predictive value vector Bk (−) and the predictive error covariance Pk (−), which are the outputs of the predictive value unifying unit 3, and inputs the predictive observed value Ck (−) and the predictive observed value error. It is a gate calculation means for calculating the covariance matrix Sk (-).

Next, the configuration of each component shown in FIG. 1 will be described in more detail. The tracking device according to the first embodiment includes an observation unit 1 that outputs an observation value vector Ck of a tracking target, and the observation unit 1 is configured by a sensor or the like included in a radar or the like for monitoring the target. Distance r
And the azimuth angle u1 and the elevation angle u2 are output.

The observation means 1 has a target observation value vector C.
The smoothing means 51 to 52 for inputting k and outputting the smoothing state of the tracking target based on each motion model are connected. The smoothing means 51 to 52 are equal to the number of motion models (N in the present embodiment).
), And outputs a smooth state based on each motion model (motion models 1 to N) in parallel. Here, the motion model is a system equation that describes a temporal change of the target state variable vector Bk.

For example, a state variable vector assuming a target motion as a uniform linear motion is Bk, 1, a state variable vector assuming a uniform acceleration motion is Bk, 2, ..., A state variable vector assuming a turning motion. Is Bk, n, the prediction units 21 to 22 and the smoothing units 51 to 52, which will be described later, perform the processes based on these motion models in parallel.

The smoothing means 51 to 52 input the predicted state of the target (described later) based on the motion models 1 to N which are the outputs of the prediction means 21 to 22 described later and the observed value vector from the observation means 1, Smooth value vector B based on motion model
It outputs k, a (+), smoothing error covariance matrix Pk, a (+) and likelihood Λk, a. These are input to the smoothed value integrating means 6 connected to the subsequent stage.

The smoothing value integrating means 6 includes smoothing means 51-52.
Smoothing state Bk, a for each motion model obtained in
(+), Pk, a (+) and likelihood Λk, a and the transition probability Pab between each motion model are input, and the fitness βk, a of each motion model is calculated. Furthermore, the goodness of fit β of the motion model
The smoothed states of the respective motion models are integrated according to k and a , and the integrated smoothed state, that is, the smoothed value vector Bk (+)
And a smooth error covariance matrix Pk (+) is calculated. The signals obtained by delaying these by a unit time by the delay element 71 are used for the predicting means 21 to 22 and the predictive value integrating means 3 described later.

The predicting means 21 to 22 are smooth value integrating means 6
The smoothed value vector Bk (+) and the smoothed error covariance matrix Pk (+), which are the outputs of the
The prediction value vector Bk, a (-) for the unit time ahead and the prediction error covariance matrix Pk, a (-) are output. These are inputs to the predicted value integration means 3.

The predictive value unifying means 3 comprises predicting means 21-22.
The prediction value vector Bk, a for each motion model, which is the output of
(-), Smoothing error covariance matrix Pk, a (-), goodness of fit βk, a of the motion model and transition probability Pab are input, and the prediction state based on each motion model is integrated to obtain a prediction value vector B.
It outputs k (−) and the prediction error covariance matrix Pk (−).

The gate calculating means 4 inputs the prediction value vector Bk (-) and the prediction error covariance Pk (-), which are the outputs of the prediction value integrating means 3, and inputs the prediction observation value Ck (-) and the prediction observation value Ck (-). The error covariance matrix Sk (−) is calculated, and the region in which the next observation value vector is likely to be obtained is set.

Next, with reference to FIGS. 1 and 2, particularly in accordance with the flowchart of FIG. 2, an operation procedure of the tracking device in the first embodiment of the present invention will be described. First, step ST
1 (represented by ST1 in the figure, the same applies hereinafter), the predicting means 21 to 22 input the smoothed state which is the output of the smoothing value integrating means 6, and the target of a unit time ahead based on each motion model. The prediction value vector Bk, a (-) and the prediction error covariance matrix Pk, a (-) are output as the prediction state.

In step ST2, the predictive value unifying unit 3 outputs the predictive value vectors Bk, a (-) for each motion model, which are the outputs of the predicting units 21 to 22, and the predictive error covariance matrix P.
k, a (−), the fitness βk, a of the motion model, and the transition probability Pab are input, and the prediction states based on the respective motion models are integrated to integrate the prediction value vector Bk (−) and the prediction error covariance matrix Pk ( -) Is output.

In step ST3, the gate calculation means 4 inputs the target prediction state which is the output of the prediction value integration means 3, and calculates the prediction observed value vector Ck (-) and the predicted observed value error covariance matrix Sk (-). It is calculated and the region where the next observation value vector Ck is obtained is output.

In step ST4, the observation means 1 observes the target and outputs the observation value vector Ck. This becomes an input to the smoothing means 51 to 52.

At step ST5, the smoothing means 51-
52 is the observation value vector Ck, and the prediction state (prediction value vector B) for each motion model obtained by the prediction means 21 to 22.
k, a (-) and the prediction error covariance matrix Pk, a (-),
The same applies hereinafter), and the smoothed value vector Bk, a (+) based on each motion model and the smoothed error covariance matrix Pk, a
(+) And likelihood Λk, a are output.

In step ST6, the smoothing value integrating means 6 obtains smoothing states (smoothing value vectors Bk, a (+) and smoothing error covariance matrices Pk, a (+) for each motion model) obtained by the smoothing means 51 to 52. , The same hereinafter), likelihood Λk, a
Then, the transition probability Pab between the motion models is input, and the goodness of fit βk, a of each motion model is calculated. Further, the smoothed states of the motion models are integrated according to the fitness βk, a of the motion model, and the smoothed value vector Bk (+) and the smoothed error covariance matrix Pk (+) are calculated.

When the processing is not completed, Bk (+), Pk
(+) And Λk, a are delayed by unit time in the delay elements 71 and 72, and the process returns to step ST1 to repeat the above processing.

The operation of the tracking device according to the first embodiment will be described in more detail below in accordance with steps ST1 to ST6 based on the equations (1) to (19) below.

[0047]

[Equation 1]

[Equation 2]

[Equation 3]

Equation (1) represents a target motion model, Bk, a is a target state variable vector based on the motion model a, and Φk−1, a is a target state transition based on the motion model a. Is a matrix, and wk-1, a is a motion model a
Represents the target driving noise vector based on.

Expression (4) represents the target observation model, and is a formulation of the process of obtaining the observation value vector Ck from the target state variable vector Bk, a. Hk,
a is an observation matrix based on the motion model a, and vk is an observation noise vector.

The smoothed state obtained in step ST6 is input to the prediction means 21 to 22 and the predicted value vector Bk, a (-) based on each motion model ahead of the unit time is output according to the equation (6). Also, the prediction error covariance matrix Pk, a (−)
Is output according to equation (7) (step ST1).

The prediction state obtained in step ST1 is input to the prediction value integrating means 3, the prediction states are integrated, the target prediction value vector Bk (-) is output according to the equation (8), and the prediction error covariance is obtained. The matrix Pk (-) is output according to the equation (9) (step ST2).

The prediction state obtained in step ST2 is input to the gate calculating means 4, the prediction observation value vector Ck (-) is calculated by the equation (11), and the prediction observation value error covariance matrix Sk is calculated by the equation (12). Calculate (-). With these values, a region in which the next observation value vector is likely to be obtained is set according to the equation (10). Where d is 3 degrees of freedom
It is determined in advance from the χ2 distribution table of.

In the smoothing means 51 to 52, the predicting means 2
Prediction value vector B for each motion model, which is the output of 1 to 22
k, a (−), the prediction error covariance matrix Pk, a (−), and the observation value vector Ck are input, and the likelihood Λk, a is calculated by the equation (1).
3), and the fitness degree βk, a of the motion model is calculated according to equation (14). Furthermore, the gain matrix Kk, a is calculated according to Expression (15), the target smoothed value vector Bk, a (+) based on the motion model is calculated according to Expression (16), and the expression (1
The smoothing error covariance matrix Pk, a (+) is calculated by 7).

In the smoothing value integrating means 6, the smoothing means 51
˜52, the smooth state Bk for each motion model,
a (+), Pk, a (+) and fitness β of the motion model
By inputting k and a, the smooth state of each motion model can be expressed by the formula (1
8) and (19) are integrated to obtain a smoothed value vector Bk
(+) And the smooth error covariance matrix Pk (+) are calculated.

According to the tracking device according to the first embodiment of the present invention described above, the predictive value integrating means and the gate calculating means for weighting and integrating the predictive state calculated for each motion model according to the fitness of the motion model. Since the gate size corresponding to the movement of the tracking target can be obtained by providing the above, it is possible to simultaneously improve the tracking performance for the turning target and the tracking accuracy for the straight-ahead target .

Embodiment 2. Next, a tracking device according to the second embodiment of the present invention will be described with reference to FIGS. 3 and 4. First, the configuration of the tracking device in the present embodiment will be described with reference to FIG. In addition, in FIG.
Constituent parts similar to those are denoted by the same reference numerals and will not be described again.

In FIG. 3, reference numeral 8 is a motion model control means for controlling the value of the fitness βk, a of the motion model according to the sampling interval. The movement model control means 8 calculates the sampling interval Δt from the observation time given to the target observation value vector which is the output of the observation means 1, and when the sampling interval is less than or equal to a predetermined value, one movement model is set. Switch and set only constant speed linear motion.

The motion model control means 8 of the first embodiment is used as a motion model number switching means for switching the number of motion models to a plurality (N in the present embodiment) or one according to such a sampling interval. By adding to the tracking device according to the present invention, when the sampling interval is equal to or greater than a predetermined value, N motion models are implemented as in the first embodiment.

Next, the operation of the tracking device according to the second embodiment of the present invention will be described with reference to FIGS. 3 and 4, and particularly according to the flowchart of FIG. Note that the operation of the tracking device in each step from step ST1 to step ST6 is the same as that in the above-described first embodiment, and therefore a repetitive description will be omitted.

Therefore, from step ST4 to step ST
7, the motion model control means 8 calculates the sampling interval Δt from the observation time given to the target observation value vector which is the output of the observation means 1. If the sampling interval is less than or equal to the predetermined value, Only set for constant speed linear motion. If the sampling interval is equal to or greater than the predetermined value, N motion models are executed as in the first embodiment.

According to the tracking apparatus according to the second embodiment of the present invention described above, the motion model control means (motion model number switching means) for switching the number of motion models to one or a plurality (N) according to the sampling interval. ) to be equipped with the
Therefore , in addition to the effects of the first embodiment, the number of motion models can be increased only when the sampling interval is wide.
It

Embodiment 3. Next, a tracking device according to the third embodiment of the present invention will be described with reference to FIGS. First, the configuration of the tracking device according to the present embodiment will be described with reference to FIG. In addition, in FIG.
Constituent parts similar to those are denoted by the same reference numerals and will not be described again.

In FIG. 5, reference numeral 9 is a second motion model control means for controlling the number of motion models according to the tracking time.
The second motion model control means 9 determines the motion model to be a constant-velocity linear motion only when the tracking time is less than a predetermined time from the observation time given to the target observation value vector output from the observation means 1. Set to. On the other hand, when the tracking time is equal to or longer than the predetermined time, the N motion models are executed as the motion model as in the first embodiment.

Next, the operation of the tracking device according to the third embodiment of the present invention will be described with reference to FIGS. 5 and 6, and particularly according to the flowchart of FIG. Note that the operation of the tracking device in each step from step ST1 to step ST6 is the same as that in the above-described first embodiment, and therefore a repetitive description will be omitted.

Therefore, from step ST4 to step ST
8, the second movement model control means 9 looks at the observation time given to the target observation value vector which is the output of the observation means 1, and when the tracking time is less than a predetermined time, the second movement model control means 9 Set only for straight forward movement. For it,
When the tracking time is equal to or longer than the predetermined time, the N motion models are executed as the motion model as in the first embodiment.

According to the tracking device according to the third embodiment of the present invention described above, the second motion model control means (motion model) for switching the number of motion models to one or a plurality (N) according to the tracking time. to be provided with a number switching means)
In addition to the effects of the first embodiment , the motion model can be only a straight-line straight motion at the start of tracking, and the number of motion models can be plural when the tracking time is a predetermined time or more.
it can.

Fourth Embodiment Next, a tracking device according to the fourth embodiment of the present invention will be described with reference to FIGS. 7 and 10. First, the configuration of the tracking device according to the present embodiment will be described with reference to FIG. 7. Note that, in FIG. 7, the same components as those in FIG.

In FIG. 7, reference numeral 10 is a constant input setting means for defining a target motion model by the independent same-dimensional input vectors u _k, a (a = 1, ..., N). A motion model in which the target motion direction is different is defined by setting the elements of the input vector u _k, a.

Referring now to FIGS. 7-10, particularly FIG.
The operation of the tracking device according to the fourth embodiment of the present invention will be described with reference to the flowchart of FIG. It should be noted that the description of the operation of the tracking device in each step from step ST1 to step ST6, which is similar to that in the first embodiment, will not be repeated.

The operation is started. First, in step ST9, the constant input setting means 10 sets the constant input vector u _k, a for changing the target movement direction to N in consideration of the movement direction.
Set according to the street. Hereinafter, based on the following formulas (20) to (42),
The operation of the tracking device in steps ST9 to ST6 will be described.

[0071]

[Equation 4]

[Equation 5]

[Equation 6]

Equations (20) and (21) of [Equation 2] represent a target motion model, Bk, a are target state variable vectors based on the motion model, and Φk−1 is a target state transition matrix. , Wk−1 represents the target driving noise vector.
u _k, a is a constant input vector based on the motion model, Γ _k ,
a is a transformation matrix of u _k, a. The difference from the first to third embodiments described above is that the order of the target state variable vector, the state transition matrix, and the driving noise vector are unified among the motion models.

The constant input vector u _k, a is an independent same-dimensional constant input vector, and each element represents each component of the motion coordinate system o-uvw as shown in FIG. At this time, there is a relationship of equations (21) to (24) between the motion coordinate system and the north reference orthogonal coordinate system. By setting each element of the constant input vector u _k, a _, a motion model having different motion directions as shown in FIG. 10 can be defined.

Equations (26) and (27) represent the target observation model, and the target state variable vector Bk, a
This is a formulation of the process of obtaining the observation value vector Ck from Hk represents an observation matrix, and vk represents an observation noise vector.

The smoothed state obtained in step ST6 is input to the predicting means 21 to 22, and the predicted value vector Bk, a (-) based on each motion model ahead of the unit time is output according to the equation (28). Further, the prediction error covariance matrix Pk, a (-) is output according to the equation (29) (step ST1).

The prediction state obtained in step ST1 is input to the prediction value integrating means 3, the prediction states are integrated, and the target prediction value vector Bk (-) is output according to the equations (30) and (31), and the prediction is performed. The error covariance matrix Pk (-) is output according to equations (31) and (32) (step ST10).

The prediction state obtained in step ST10 is input to the gate calculation means 4, and the prediction observation value vector Ck (-) is calculated by the equation (34), and the prediction observation value error covariance matrix Sk (-) is calculated by the equation (35). calculate. From these values, the region in which the next observation value vector is likely to be obtained is given by the formula (3
It is set according to 3). Here, d is determined in advance from the χ2 distribution table with three degrees of freedom.

At step ST5, the smoothing means 51-
Reference numeral 52 inputs the prediction value vector Bk, a (-), prediction error covariance matrix Pk, a (-), and observation value vector Ck for each motion model, which are the outputs of the prediction means 21 to 22, and formula (36) The likelihood Λk, a is calculated in accordance with Equation (37), and the fitness βk, a of the motion model is calculated according to Equation (37). Furthermore, the gain matrix Kk, a is calculated according to the equation (38), and the target smoothed value vector Bk, a based on the motion model is calculated according to the equation (39).
(+) Is given by equation (40) as the smooth error covariance matrix Pk,
Calculate a (+).

In step ST6, the smoothing value integrating means 6 has the smoothing states Bk, a (+), Pk, a (+), and the likelihood Λk, a obtained by the smoothing means 51 to 52 for each motion model.
And the goodness of fit βk, a of the motion model are input, and the smooth states of the motion models are integrated by the equations (41) and (42) to obtain a smoothed value vector Bk (+) and a smoothed error covariance matrix Pk (+). To calculate.

The tracking device according to the fourth embodiment of the present invention described above is provided with the constant input setting means for defining the target motion model.
In addition, a plurality of motion models can be defined with the same dimensional state variable vector and the same state transition matrix .

Embodiment 5. Next, a tracking device according to the fifth embodiment of the present invention will be described with reference to FIGS. 11 and 12. First, the configuration of the tracking device according to the present embodiment will be described with reference to FIG. Note that, in FIG. 11, the same components as those in FIG.

In FIG. 11, reference numeral 2 denotes an observation matrix calculation means for calculating an observation matrix in consideration of the radar position from the radar position information when using the observation value vectors from a plurality of radars having different arrangement positions. At this time, it is assumed that, in addition to which target position information, the observation value vector is provided with information regarding from which radar it was obtained. Further, it is assumed that the installation position of the radar is self-explanatory. The observation matrix calculation means 2 inputs the target observation value vector, which is the output of the observation means 1, and outputs the observation matrix in consideration of the installation position of the radar which is the information source.

Next, referring to FIGS. 11 and 12, and in particular according to the flowchart of FIG. 12, the fifth embodiment of the present invention will be described.
The operation of the tracking device in FIG. Incidentally, step ST
The description of the same parts of the operation of the tracking device in each step from 1 to step ST6 as those in the first embodiment will be omitted.

The operation is started, and the process proceeds from step ST2 to step ST11. The observation matrix calculating means 2 inputs the target observation value vector output from the observing means 1, and the radar installation position based on the position information of the radar which is the information source. Calculate the observation matrix considering. Hereinafter, the operation of the tracking device in steps ST1 to ST6 will be described based on the following equations (43) to (64).

[0085]

[Equation 7]

[Equation 8]

[Equation 9]

Since the target motion model is the same as that of the first embodiment of the invention, the repetitive description will be omitted. Expression (4
3) and (44) represent the target observation model,
This is a formulation of the process of obtaining the target observation value vector Ck, l from the radar of the observation means 1 from the target state variable vector Bk, a. Here, Hk, a is an observation matrix in the north reference Cartesian coordinate system in which the target state variable vector is defined, and h (•) shown in Expression (46) is the north reference in which the target state variable vector is defined. It is a transformation matrix from the Cartesian coordinate system to the North reference Cartesian coordinate system in which the observation value vector is obtained, and is a function of the radar installation latitude / longitude as shown in Expression (45). Further, vk, l represents an observation noise vector.
Further, the observation matrices H ¹ k, H ¹ k, a are calculated by the equations (47) and (48).

In step ST1, step ST6
The smoothed state obtained in step S21 is input to the prediction means 21 to 22, and the predicted value vector Bk, a based on each motion model ahead of the unit time
(-) Is output according to equation (49), and the prediction error covariance matrix Pk, a (-) is output according to equation (50).

In step ST2, step ST1
The prediction state obtained in step (3) is input to the prediction value integrating means 3, and the prediction states are integrated to calculate the target prediction value vector Bk (-) by the formula (5).
1) and outputs the prediction error covariance matrix Pk (−) according to equation (52).

From step ST2 to step ST11, the prediction state obtained in step ST2 is input to the observation matrix calculating means 2, and the observation matrix considering the installation position of the radar is calculated by the equations (47) and (48).

From step ST11 to step ST3, the predicted state obtained in step ST2 and step ST1
The observation matrix obtained in 1 is input to the gate calculation means 4, the prediction observation value vector Ck (−) is calculated using Equation (54), and the prediction observation value error covariance matrix Sk is calculated using Equation (55).
Calculate (-). With these values, a region in which the next observation value vector is likely to be obtained is set according to the equation (53). Here, d is determined in advance from the χ2 distribution table with three degrees of freedom.

At step ST4, the smoothing means 51-
52 inputs the prediction value vector Bk, a (-), the prediction error covariance matrix Pk, a (-), and the observation value vector Ck for each motion model which are the outputs of the prediction means 21 to 22, and formula (56) Then, the likelihood Λk, a is calculated according to Equation (59) and the fitness of the motion model is calculated according to Equation (59). Further, in step ST5, the gain matrix Kk, a is changed to the expression (6) according to the expression (60).
1) The target smoothed value vector B based on the motion model
For k, a (+), the smooth error covariance matrix Pk, a (+) is calculated by the equation (62).

In step ST6, the smoothing value integrating means 6 has the smoothing states Bk, a (+), Pk, a (+) for each motion model obtained in the smoothing means 51 to 52 and the fitness βk, of the motion model. a is input, the smooth states of the respective motion models are integrated by the equations (63) and (64), and the smooth value vector Bk (+) and the smooth error covariance matrix Pk (+)
To calculate.

The tracking device according to the fifth embodiment of the present invention described above includes the observation matrix calculating means for calculating the observation matrix in consideration of the arrangement position of the radar .
In addition to the effects of the first embodiment, it is possible to accurately use the observation information of a plurality of radars having different arrangement positions .

Sixth Embodiment Next, a tracking device according to the sixth embodiment of the present invention will be described with reference to FIGS. 13 and 14. First, the configuration of the tracking device according to the present embodiment will be described with reference to FIG. Note that, in FIG. 13, the same components as those in FIG.

In FIG. 13, reference numeral 12 is a hypothesis setting means for setting a hypothesis regarding which signal is the signal from the tracking target when there are a plurality of signals in the gate due to remaining unnecessary signals, and 131 to 132 are , A correlation means for inputting the prediction value vector and prediction error covariance matrix for each motion model, which is the output of the prediction means, and outputting the reliability βk, i, a of the hypothesis.

Reference numerals 141 to 142 indicate a plurality of observation signals from the observation means 1, the hypothesis reliability βk, i, a from the correlation means 131 to 132, and the motion models from the prediction means 21 to 22. The predicted state is input, and the smoothed value vector Bk, a is set as the target smoothed state based on each motion model.
(+) And a smoothing means for outputting the smoothing error covariance matrix Pk, a (+), 15 is a smooth state based on the motion model from the smoothing means, a predictive state based on the motion model from the predicting means 21 to 22, and a correlation It is smoothed value integrating means for inputting the reliability of the hypothesis from the means 131 to 132, the observed value vector from the observing means, and the transition probability Pab between the motion models, and first calculating the goodness of fit βk, a of each motion model. .

Next, further details of each of the above components will be described. The hypothesis setting means 12 uses a plurality of observation value vectors Ck,
Set a hypothesis about which of i (i = 1, 2, ..., Mk) is the observation value vector from the tracking target.
For example, a plurality of observation value vectors Ck, i (i = 1, 2,
..., mk) Hypothesis 0; all unnecessary signals. Hypothesis 1; Ck, 1 is the signal from the tracking target. Hypothesis 2; Ck, 2 is the signal from the tracking target. ... Hypothesis N; Ck and mk are signals from the tracking target. Establish a hypothesis such as.

The correlating means 131 to 132 are the predicting means 21.
Predicted value vector B _{k for} each motion model, which is the output of
a (−) and the prediction error covariance matrix P _k, a (−) are input, and the reliability βk, i, a of the hypothesis established by the hypothesis setting means 12 is output. The smoothing means 141-142 are
A plurality of observation signals from the observation means 1 and the correlation means 131-
The reliability of the hypothesis βk, i, a from 132 and the prediction means 2
The predicted state for each motion model from 1 to 22 is input, and the smoothed value vector Bk, a (+) and the smoothed error covariance matrix Pk, a are set as the target smoothed state based on each motion model.
Outputs (+).

The smooth value integrating means 15 is a smooth state based on the motion model from the smoothing means, a predictive state based on the motion model from the predicting means 21 to 22, and correlating means 131 to 132.
By inputting the reliability of the hypothesis, the observation value vector from the observation means, and the transition probability Pab between the motion models, first, the fitness βk, a of each motion model is calculated. Next, the goodness of fit β
The smoothed state for each motion model is integrated into the smoothed value vector Bk (+) and the smoothed error covariance matrix Pk (+) by k and a.

Next, referring to FIGS. 13 and 14, and particularly according to the flowchart of FIG. 14, the sixth embodiment of the present invention will be described.
The operation procedure of the tracking device in FIG. The description of the operation of the tracking device in each step from step ST1 to step ST4, which is similar to that in the first embodiment, will not be repeated.

The operation is started, steps ST1 to ST4 are executed, and the process proceeds from step ST4 to step ST12. In step ST12, the hypothesis setting means 12 determines which of the plurality of signals in the gate is the tracking target. A hypothesis is established as to whether or not the signal is. In step ST13, the correlation means 131 to 132 calculate the reliability βk, i, a of the hypothesis for each motion model.

Next, in step ST14, the smoothing means 141-142 input the observation value vectors Ck, i, weight the observation value vectors with the calculated reliability βk, i, a, and calculate the smooth value vector Bk. , A (+) and the smoothing error covariance matrix Pk, a (+) are output.

Finally, in step ST15, the smoothing value integrating means 15 integrates the smoothing states based on the respective motion models according to the fitness of the motion models, and the smoothing value vector Bk (+) and the smoothing error covariance matrix Pk ( +) Is obtained.

Hereinafter, the operation of the tracking device to be executed in each of the steps ST1 to ST15 will be described in more detail based on the following equations (65) to (79). The target motion model, the observation model, and the prediction process are the same as those in the first embodiment.
Since it is the same as that in, the description is omitted.

[0105]

[Equation 10]

[Equation 11]

The prediction state obtained in step ST2 is input to the gate calculating means 4, the prediction observation value vector Ck (-) is calculated using the equation (66), and the prediction observation value error is calculated using the equation (67). The covariance matrix Sk (-) is calculated. With these values, a region in which the next observation value vector is likely to be obtained is set according to the equation (65). Here, d is determined in advance from the χ2 distribution table with three degrees of freedom. The observed value vector Ck, i satisfying the conditional expression (65) is the target of the correlation processing.

In the correlating means 131 to 132, the reliability βk, i, a of the correlation hypothesis between the observed value vector obtained by the hypothesis setting means 12 and the tracking target is calculated by the equations (68) to (70).
Calculate according to. Here, PD, PG, and PE are parameters obtained from the radar.

In the smoothing means 141-142, the prediction value vector Bk, a (-) for each motion model, which is the output of the prediction means 21-22, and the prediction error covariance matrix Pk, a.
(-), The observed value vector Ck, i, and the hypothesis reliability β
k, i, a are input, the gain matrix Kk, a is calculated according to the equation (71), and the target smoothed value vector Bk, a (+) based on the motion model is calculated according to the equations (72), (74), and (75). ,
According to the equations (73) to (76), the smooth error covariance matrix Pk,
Calculate a (+).

The smoothing value integrating means 15 includes smoothing means 141-141.
Smooth state B for each motion model obtained in 142
k, a (+), Pk, a (+) and correlation means 131 to 1
The hypothesis reliability βk, i, a obtained in 32 is input, and the fitness βk, a of each motion model is calculated according to the equation (77). From the goodness of fit βk, a of the motion model thus obtained, the smooth state of each motion model is calculated by the equation (78).
And (79) to obtain the smoothed value vector Bk
(+) And the smooth error covariance matrix Pk (+) are calculated.

According to the tracking apparatus according to the sixth embodiment of the present invention described above, even when there are a plurality of observation signals due to the remaining undesired signals, the correlation processing of which signal is the target signal is performed. By providing the hypothesis setting means and the correlating means for performing , in addition to the effect of the first embodiment, the target is correctly tracked even in an environment where undesired signal remains unerased.
can do.

Seventh Embodiment Next, a tracking device according to the seventh embodiment of the present invention will be described with reference to FIGS. 15 and 16. First, the configuration of the tracking device in the present embodiment will be described with reference to FIG. Note that, in FIG. 15, the same components as those in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted.

In FIG. 15, 16 is second hypothesis setting means for setting a hypothesis about which observation signal from a plurality of targets is a signal from a tracked target, and 181 to 182 are prediction value vectors for each motion model and Second correlator 171 to 172, which inputs the prediction error covariance matrix and outputs the reliability βk, i, a of the hypothesis, is a plurality of observation signals from the observer 1, and second correlator 181 to 181. The reliability βk, i, a of the hypothesis from 182 and the prediction state for each motion model from the prediction means 21 to 22 are input, and the smoothed value vector Bk, a is set as the smooth state of the target t based on each motion model.
^The smoothing means for outputting ^t (+) and the smoothing error covariance matrix Pk, a ^t (+), 19 is a smooth state based on the motion model from the second smoothing means 171 to 172, and based on the motion model from the predicting means. Predicted state, second correlator 181-18
It is a smoothed value integrating means for inputting the reliability of the hypothesis from 2, the observed value vector from the observing means, and the transition probability Pab between the motion models, and calculating the goodness of fit βk, a of each motion model.

Next, further details of each of the above-mentioned components will be described. The second hypothesis setting means 16 determines which observation signal is the tracked target t (t = 1, 1) in an environment where a plurality of targets exist.
2, ..., Tk), a hypothesis is set. Further, the second correlating means 181 to 182 input the prediction value vector and the prediction error covariance matrix for each motion model, which are the outputs of the predicting means 21 to 22, and output the reliability βk, i, a of the hypothesis. To do.

The second hypothesis setting means 16 sets a hypothesis about which of the plurality of observation value vectors Ck, j (j = 1, 2, ..., Mk) is the observation value vector from the tracking target. Set. At this time, the entire hypothesis is matrix W ^k = [w ^k
jt]. j is an observation value vector (j = 1, 2, ...
, Mk), and t is a tracking target (t = 0, 1, ...
., Tk). Tk represents the total target number.

Further, the second hypothesis setting means 16 uses the entire hypothesis Wk to the partial hypotheses W ^{k, i} = [w ^{k , i} jt] (j = 1,
2, ..., Mk, t = 0, 1, ..., Tk) are output. The constraint conditions at this time are: At most one observation value vector Ck, j for each tracking target
Corresponds to 2. Only one target corresponds to each observation vector, which results in Ik hypotheses about correlations.

Further, the second correlating means 181 to 182 are
The prediction value vector and prediction error covariance matrix for each motion model, which are the outputs of the prediction means 21 to 22, are input, and further, the correlation hypothesis established by the second hypothesis setting means 16 is input, and the hypothesis regarding motion and correlation The reliability βk, i, a of is output.

The second smoothing means 171 to 172 have a plurality of observation signals from the observing means 1 and the second correlating means 181 to 181.
182 of the hypothesis reliability βk, i, a and the prediction means 2
The target state (t = 1, 2, ..., T) based on each motion model is input by inputting the prediction state for each motion model from 1 to 22.
k) as the smooth state of the smooth value vector Bk, a ^t (+)
And a smooth error covariance matrix Pk, a ^t (+) is output.

The smoothing value integrating means 19 is the second smoothing means 1
Smooth states based on the motion model from 71 to 172, prediction states based on the motion model from the predicting means 21 to 22, reliability of the hypothesis from the second correlating means 181 to 182, an observation value vector from the observing means 1, and By inputting the transition probability Pab between the motion models, first, the goodness of fit βk of each motion model,
Calculate a.

Next, the smoothness for each motion model is integrated into the smoothed value vector Bk ^t (+) and the smoothed error covariance matrix Pk ^t (+) according to the goodness of fit βk, a.

Next, referring to FIGS. 15 and 16, and in particular according to the flowchart of FIG. 16, the seventh embodiment of the present invention will be described.
The operation procedure of the tracking device in FIG. The description of the operation of the tracking device in each step from step ST1 to step ST4, which is similar to that in the first embodiment, will not be repeated.

The operation is started, steps ST1 to ST4 are executed, and the process proceeds from step ST4 to step ST16. In step ST16, the second hypothesis setting means 16 determines from the result of the gate calculating means 4 a plurality of observation signals. Set a hypothesis about which of these is the signal from the tracked target. As a result, the matrix W ^{k of the} entire hypothesis
And Ik hypothesis matrices W ^{k, i} are obtained.

In step ST17, the second correlating means 181 to 182 calculate the reliability βk, i, a of the hypothesis. Next, in step ST18, the second smoothing means 171 to 172 input the observation value vectors Ck, j, weight each observation value vector with the calculated reliability βk, i, a, and calculate the smoothing value vector. Bk, a ^t (+) and smoothing error covariance matrix Pk, a ^t (+) are output.

Finally, from step ST18 to step S
Proceeding to T19, the smooth states based on each motion model are integrated according to the fitness of the motion model, and Bk ^t (+) and Pk
^{Get t} (+).

Hereinafter, the operation of the tracking device to be executed in each of the steps ST1 to ST19 will be described in more detail based on the following equations (80) to (121).

[0125]

[Equation 12]

[Equation 13]

[Equation 14]

[Equation 15]

[Equation 16]

[Equation 17]

[Equation 18] First, the motion model and the observation model of the tracking target t (t = 1, 2, ..., Tk) are obtained from Expressions (80) to (84). In step ST1, the smoothed state obtained in step ST19 is input to the prediction means 21 to 22, and the predicted value vector Bk based on each motion model of the unit time ahead of the target t is calculated.
a ^t (−) is output using equation (85). Further, the prediction error covariance matrix Pk, a ^t (-) and outputs using Equation (86).

In step ST2, step ST1
The prediction state obtained in step S21 is input to the prediction value integrating means 3 to integrate the prediction states, and the target prediction value vector Bk ^t (−) is calculated using the equation (87). The prediction error covariance matrix Pk ^t (−) ) Is calculated according to equation (88) and output.

[0127] Enter the predicted state obtained in step ST2 to gate calculation means 4, the predicted observation value vector Ck ^t using Equation (90) (-) is calculated, the predicted observation value error covariance by the formula (91) The matrix Sk ^t (−) is calculated. From these values, a region in which the next observation value vector is likely to be obtained is set according to the equation (89). Where d is 3 degrees of freedom
It is determined in advance from the χ2 distribution table of.

The second hypothesis setting means 16 calculates the matrix of the entire hypothesis from the results of the gate calculation means 4 using the equations (92) to (95).
And the hypothesis matrix is output using equations (96) to (105).

The second correlating means 181 to 182 have the reliability of the hypothesis βk, i, a and the tracking target t (t = 1,2 ,.
..., reliability βk hypothesis for Tk), i, a a ^t is calculated according to the equation (106) - (110).

In step ST18, the second smoothing means 171 to 172 have the observation value vectors Ck, i (i = 1, 1).
2, mk), and the reliability βk, i, a at which each observation value vector is calculated using equations (111) to (118)
The, βk, i, a ^t the smoothed value vector Bk of the tracking target t based on the motion model performs weighting, a
^{The t} (+) and the smoothed error covariance matrix Pk, a ^t (+) are output.

Finally, in step ST19, the smooth states based on the respective motion models are integrated according to the fitness of the motion models, and Bk ^t (+) and Pk ^t (+) are calculated by the equation (119).
~ (121) to calculate.

According to the tracking apparatus according to the seventh embodiment of the present invention described above, in the multi-target environment in which there are a plurality of targets, the correlation processing of which observation signal is the signal from the already-tracked target is performed. By providing the hypothesis setting means and the second correlation means of, in addition to the effect of the first embodiment ,
It is possible to track the correct target even in a multi-target environment.
It

[0133]

The tracking device according to the first aspect of the present invention is configured as described above, and in particular, the prediction value for weighting and integrating the prediction states calculated for each motion model according to the fitness of the motion model. Since the output of the unifying means can be given to the gate calculating means and an appropriate gate size can be obtained according to the movement of the tracking target, a tracking device that simultaneously enhances the tracking performance with respect to the turning target and the tracking accuracy with respect to the straight ahead target by a synergistic effect is obtained. be able to.

The tracking device according to the second aspect of the present invention is configured as described above, and in particular , the motion model control means (motion model number switching means) for switching the number of motion models to one or more in accordance with the sampling interval. that) with a
Therefore, in addition to the effect of the invention according to claim 1, it is possible to obtain a tracking device capable of providing a plurality of motion models only when the sampling interval is wide.

The tracking device according to the third aspect of the present invention is configured as described above, and particularly the second motion model control for switching the number of motion models to one or a plurality (N) in accordance with the tracking time. comprising means (number of motion model switching means)
Therefore, in addition to the effect of the invention according to claim 1, there is provided a tracking device capable of making the motion model only a constant-velocity rectilinear motion at the start of tracking, and having a plurality of motion models when the tracking time is a predetermined time or more. Obtainable.

The tracking device according to a fourth aspect of the present invention is configured as described above, and in particular has constant input setting means for defining a target motion model.
In addition to the effect of the invention described above, it is possible to obtain a tracking device capable of defining a plurality of motion models with the same state variable vector and the same state transition matrix.

The tracking device according to a fifth aspect of the present invention is configured as described above, and is provided with an observation matrix calculation means for calculating an observation matrix in consideration of the arrangement position of the radar.
As a result, in addition to the effect of the invention according to claim 1, it is possible to obtain a tracking device that can accurately use the observation information of a plurality of radars having different arrangement positions.

The tracking device according to the sixth aspect of the present invention is configured as described above, and which signal is the target signal even when there are a plurality of observation signals due to the remaining undesired signals. The provision of the hypothesis setting means and the correlation means for performing the correlation processing of 1.
In addition to the above , it is possible to obtain a tracking device that can correctly track a target even in an environment where undesired signal remains unerased.

The tracking device according to the seventh aspect of the present invention is configured as described above, and in a multi-target environment in which there are a plurality of targets, the correlation processing of which observation signal is the signal from the already-tracked target is performed. than that with the second hypothesis setting means and a second correlation means for performing, the invention according to claim 1
In addition to the effect, it is possible to obtain a tracking device that can track a correct target even in a multi-target environment.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a tracking device according to a first embodiment of the present invention,

2 is a flow chart showing the operation of the tracking device shown in FIG.

FIG. 3 is a block diagram showing a configuration of a tracking device according to a second embodiment of the present invention,

4 is a flow chart showing the operation of the tracking device shown in FIG.

FIG. 5 is a block diagram showing the configuration of a tracking device according to a third embodiment of the present invention,

FIG. 6 is a flow chart showing the operation of the tracking device shown in FIG.

FIG. 7 is a block diagram showing a configuration of a tracking device according to a fourth embodiment of the present invention,

8 is a flow chart showing the operation of the tracking device shown in FIG. 7,

FIG. 9 is a diagram showing a motion coordinate system used in the fourth embodiment;

FIG. 10 is a diagram showing a target motion direction used in Embodiment 4;

FIG. 11 is a block diagram showing a configuration of a tracking device according to a fifth embodiment of the present invention,

FIG. 12 is a flowchart showing the operation of the tracking device shown in FIG.

FIG. 13 is a block diagram showing a configuration of a tracking device according to a sixth embodiment of the present invention,

FIG. 14 is a flowchart showing the operation of the tracking device shown in FIG.

FIG. 15 is a block diagram showing a configuration of a tracking device according to a seventh embodiment of the present invention,

16 is a flow chart showing the operation of the tracking device shown in FIG.

FIG. 17 is a block diagram showing a configuration of a conventional tracking device,

18 is a flowchart showing the operation of the conventional tracking device shown in FIG.

[Explanation of symbols]

1 Observation Means, 21-22 Prediction Means, 3 Prediction Value Integration Means 4, Gate Calculation Means, 51-52 Smoothing Means, 6
Smooth value integrating means, 71 to 72 delay element, 8 motion model controlling means, 12 hypothesis setting means, 9 second motion model controlling means, 16 second hypothesis setting means, 2 observation matrix calculating means, 10 constant input setting means , 41
Gate selection means, 131-132 Correlation means, 141
-142 smoothing means, 15 smoothing value integrating means, 19 smoothing value integrating means, 31-32 gate calculating means, 18
1-182 Second Correlation Means, 171-272 Second
Smoothing means.

Continuation of front page (56) Reference JP-A-4-113293 (JP, A) JP-A-4-198886 (JP, A) JP-A-6-43241 (JP, A) JP-A-6-94823 (JP , A) JP-A-9-297176 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G01S ⁷ /00-7/42 G01S 13/00-13/95

Claims (7)

(57) [Claims] 1. A tracking device for turning and straight-ahead targets.
In the above , by inputting the prediction means for outputting the respective prediction states based on the plurality of motion models , the prediction value integrating means for integrating and outputting the plurality of prediction states, and the integrated prediction states. Gate calculation means for outputting the area where the next observation value vector is obtained, observation means for observing the tracking target in the area and outputting the observation value vector, and the observation value vector from the observation means and the prediction means comprising a smoothing means for outputting a plurality of smoothing the target state based enter a plurality of predicted state to each motion model, a smoothing value integrating means for outputting integrated plurality of smooth state, a plurality of The predictor vector and predictor
A tracking device , characterized in that the measurement error covariance matrices are weighted and integrated according to the fitness of each motion model. 2. The tracking device according to claim 1, further comprising a motion model control means for switching the number of the motion models to one or a plurality according to a sampling interval. 3. The tracking device according to claim 1, further comprising a second motion model control means for switching the number of the motion models to one or a plurality according to the tracking time. 4. The tracking device according to claim 1, further comprising constant input setting means for defining a plurality of desired motion models and setting N kinds of independent same-dimensional constant input vectors. 5. The tracking device according to claim 1, further comprising an observation matrix calculation means for processing observation information in consideration of arrangement positions of a plurality of radars having different arrangement positions. 6. A hypothesis setting means and a correlation means for performing correlation processing of which signal is a target signal when a plurality of observation signals are generated due to the remaining undesired signals. Item 1. The tracking device according to item 1. 7. In a multi-target environment in which multiple targets exist, a second hypothesis setting means and a second correlation means are provided for performing correlation processing of which observed signal is a signal from a tracked target. The tracking device according to claim 1, which is characterized in that