JP6641534B2 - Target tracking device - Google Patents

Description

  The present invention relates to a target tracking device that tracks a plurality of targets.

  In the angle measurement of a plurality of targets by the high-resolution angle measurement, there is a problem that when the arriving waves arriving from the targets have a high correlation and the respective targets are close to each other, the plurality of targets are represented as one observation value. . This is hereinafter referred to as degeneration of the observed value. When the tracking processing is performed using the degenerated observation values, an error in the target position estimated value and the speed estimated value increases, and a false navigation track is generated. Furthermore, this causes a problem that the number of wakes, that is, the target number is incorrectly estimated. Therefore, the solution of the degeneration of the observation value in the target tracking device is an important issue.

  Conventionally, as a technique for addressing such a problem, for example, there is a technique disclosed in Patent Document 1. In the technique described in Patent Document 1, angle measurement is performed by the MUSIC method. Since a target number is required to perform angle measurement by the MUSIC method, the target number is generally estimated using eigenvalues of a correlation matrix. In the angle measurement by the MUSIC method, the observation value may degenerate if the estimation of the target number is erroneous. Therefore, the technique described in Patent Document 1 provides the MUSIC method with the target number estimated using the wake. To prevent degeneration.

JP 2004-309209 A

  However, the conventional technique described in Patent Document 1 is based on the assumption that the correlation between incoming waves is low, and it has been difficult to avoid degeneration of observed values when the correlation between incoming waves is high.

  The present invention has been made in order to solve such a problem, and an object of the present invention is to provide a target tracking device that can avoid degeneration of an observed value even when the correlation between incoming waves is high.

A target tracking device according to the present invention includes a sensor processing unit that acquires observation signals of a plurality of targets, an observation value extraction unit that generates an observation value from the observation signals acquired by the sensor processing unit, and an observation value extraction unit that generates an observation value. Using the obtained observation values, a correlation hypothesis including the wake, the observation value, and the evaluation value of the correlation hypothesis of a plurality of targets is obtained by using the set position estimation method , and the presence or absence of degeneration of the observation value is determined from the correlation hypothesis. When the multi-target tracking processing unit to be determined and the multi-target tracking processing unit determine that there is degeneration of the observation value , the multi-target tracking processing unit A decompression elimination unit that obtains observation values without degeneration of multiple targets using a position estimation method that performs target estimation different from the used position estimation method, and a multi-target tracking processing unit, if there is no degeneration of observation values If it is determined, the degeneration of the observation The track of the correlation hypothesis determines the Most updates with observations included in the correlation hypothesis, if it is determined that there is degeneracy observations, the track of the correlation hypothesis is determined that there is degeneracy of the observations The update is performed using the observation values without degeneration that are included in the correlation hypothesis and obtained by the degeneration elimination unit.

The target tracking device according to the present invention is used by the multi-target tracking processing unit based on the predicted value of the multi-target for one observation value when the multi-target tracking processing unit determines that the observed value is degenerated. The multi-target tracking processing unit has determined that there is no degeneration of the observed value by using a degeneration eliminating unit that obtains observation values without degeneration of multiple targets using a position estimation method that performs target estimation different from the obtained position estimation method . In this case, the wake of the correlation hypothesis determined to have no degeneration of the observation value is updated using the observation value included in the correlation hypothesis, and the wake determined to have degeneration of the observation value is included in the correlation hypothesis. The update is performed using the observed value without degeneration obtained by the degeneration elimination unit. As a result, even when the correlation between the incoming waves is high and the observed value degenerates, the influence can be avoided.

FIG. 1 is a configuration diagram illustrating a target tracking device according to a first embodiment of the present invention. FIG. 2 is a hardware configuration diagram of the target tracking device according to the first embodiment of the present invention. 5 is a flowchart illustrating an operation of a tracking processing unit in the target tracking device according to the first embodiment of the present invention. FIG. 3 is an explanatory diagram illustrating a maximum likelihood method of the target tracking device according to the first embodiment of the present invention. FIG. 9 is a configuration diagram illustrating a target tracking device according to a second embodiment of the present invention. FIG. 9 is a configuration diagram illustrating a target tracking device according to a third embodiment of the present invention. FIG. 13 is a configuration diagram illustrating a target tracking device according to a fourth embodiment of the present invention. 14 is a flowchart illustrating an operation of a tracking processing unit in the target tracking device according to the fourth embodiment of the present invention.

Hereinafter, in order to explain this invention in greater detail, the preferred embodiments of the present invention will be described with reference to the accompanying drawings.
Embodiment 1 FIG.
FIG. 1 is a configuration diagram showing a target tracking device according to the present embodiment.
The target tracking device according to the present embodiment includes a sensor processing unit 10, an observation value extraction unit 20, and a tracking processing unit 30. The sensor processing unit 10 is a processing unit that acquires a target observation signal, and includes a reception antenna unit 11 and a reception circuit 12. The receiving antenna unit 11 is an antenna for receiving a radio wave from a target, and includes a plurality of antennas for performing super-resolution angle measurement. The receiving circuit 12 is a circuit that converts an analog signal from the receiving antenna unit 11 into a digital signal and outputs the digital signal. The observation value extraction unit 20 is a processing unit that generates an observation value from the observation signal output from the reception circuit 12. The tracking processing unit 30 includes a multi-target tracking processing unit 40 and a degeneration cancellation unit 50. The multi-target tracking processing unit 40 is a processing unit that manages the position specification estimates of a plurality of targets using the observation values obtained by the observation value extraction unit 20, and includes a correlation hypothesis generation unit 41 and an observation value degeneration determination unit 42. , A correlation hypothesis evaluation unit 43, a track update unit 44, a correlation hypothesis storage unit 45, and a track determination unit 46.

  The correlation hypothesis generation unit 41 is a processing unit that generates a new correlation hypothesis based on the observation value from the observation value extraction unit 20 and the correlation hypothesis from the correlation hypothesis storage unit 45. The observation value degeneracy determination unit 42 determines whether or not the observation value is degenerated in the correlation hypothesis generated by the correlation hypothesis generation unit 41, and associates the determined correlation hypothesis with the determination result of the correlation hypothesis. This is a processing unit that outputs. The correlation hypothesis evaluation unit 43 acquires the correlation hypothesis from the observation value degeneracy determination unit 42 or the correlation hypothesis update unit 52 of the degeneration cancellation unit 50, evaluates the likelihood of the correlation hypothesis, and updates the evaluation value included in the correlation hypothesis. The processing unit outputs the correlation hypothesis with the updated evaluation value. The track updating unit 44 is a processing unit that acquires a correlation hypothesis from the correlation hypothesis evaluation unit 43 and updates each track included in the correlation hypothesis using a correlated observation value. The correlation hypothesis storage unit 45 is a storage unit for storing all the correlation hypotheses output from the track update unit 44. The track determination unit 46 is a processing unit that outputs a track included in the correlation hypothesis having the highest evaluation value among all correlation hypotheses output from the track update unit 44 as a target track.

  The degeneration elimination unit 50 is a processing unit that obtains an observation value without degeneration as a position specification estimated value by using a position estimation method of performing target estimation based on multi-target prediction values for one observation value, A super-resolution angle measurement unit 51 and a correlation hypothesis updating unit 52 are provided. When the correlation hypothesis determined by the observation value degeneracy determination unit 42 and the determination result of the correlation hypothesis are output, the super-resolution angle measurement unit 51 obtains a digital signal from the reception circuit 12 upon receiving the correlation hypothesis and the digital hypothesis. Using the signal and the track included in the correlation hypothesis, super-resolution angle measurement is performed, which is a position estimation method that estimates the target based on the predicted values of multiple targets for one observation value. This is a processing unit that outputs the correlation hypothesis from the value degeneracy determination unit 42. The correlation hypothesis updating unit 52 generates a correlation hypothesis in which the observation value included in the correlation hypothesis acquired from the super-resolution angle measurement unit 51 is replaced with an observation value calculated from the angle measurement result acquired from the super-resolution angle measurement unit 51. , And outputs the generated correlation hypothesis to the multi-target tracking processing unit 40.

FIG. 2 is a diagram illustrating a hardware configuration for realizing the observation value extraction unit 20 and the tracking processing unit 30 in the target tracking device according to the present embodiment.
The observation value extraction unit 20 and the tracking processing unit 30 in the target tracking device include a processor 101, a memory 102, an input / output I / F (input / output interface) 103, a storage 104, and a bus 105. The processor 101 is a processor for implementing the observation value extraction unit 20 and the tracking processing unit 30 by executing a program corresponding to the functions of the observation value extraction unit 20 and the tracking processing unit 30. The memory 102 is a storage unit such as a program memory for storing various programs and a ROM and a RAM used as a work memory used when the processor 101 performs data processing. The input / output I / F 103 is a communication interface for inputting / outputting an observation signal from the sensor processing unit 10 and outputting a processing result of the multi-target tracking processing unit 40 to / from the outside. The storage 104 is a storage device for storing various data and programs corresponding to the functions of the observation value extraction unit 20 and the tracking processing unit 30. The bus 105 is a communication path for interconnecting the processors 101 to the storage 104.

  The processing performed by the observation value extraction unit 20 and the tracking processing unit 30 in FIG. 1 is performed by the processor 101 expanding a program stored in the storage 104 on the memory 102 and executing the program. The digital signal output from the receiving circuit 12 of the sensor processing unit 10 is stored in the memory 102 or the storage 104 via the input / output I / F 103. The processing result of the tracking processing unit 30 is stored in the storage 104 as needed, and is output to the outside via the input / output I / F 103.

Next, the operation of the target tracking device according to the first embodiment will be described.
The receiving antenna unit 11 receives a radio signal output from a target for communication or the like, and outputs an analog signal corresponding to the received radio signal. The receiving circuit 12 converts an analog signal acquired from the receiving antenna unit 11 into a digital signal, and outputs the digital signal to the observation value extracting unit 20 and the super-resolution angle measuring unit 51 of the degeneration eliminating unit 50.
The observation value extraction unit 20 calculates an angle measurement value using the digital signal input from the reception circuit 12, and outputs an estimated position of the target on the coordinates. This estimated position is hereinafter referred to as an observation value. Observations are represented by vectors.

Next, the processing of the tracking processing unit 30 will be described. FIG. 3 is a flowchart showing the processing of the tracking processing unit 30 according to the first embodiment of the present invention. FIG. 3 shows a processing flow at time k.
The correlation hypothesis generator 41 acquires all correlation hypotheses at time k−1 from the correlation hypothesis storage 45. Further, the observation value at time k is obtained from the observation value extraction unit 20. The correlation hypothesis includes a track, an observation value, and an evaluation value of the correlation hypothesis. The wake is an estimated value of the position and speed of the target at each time, and is represented using a vector. In the correlation hypothesis generation unit 41, each input correlation hypothesis is called a parent correlation hypothesis. The correlation hypothesis generation unit 41 performs the following processing for each parent correlation hypothesis. That is, a wake is extracted from a certain parent correlation hypothesis, and the observation value input from the observation value extraction unit 20 is assigned to each wake. At this time, there may be a track to which no observation value is assigned, or a plurality of tracks to which the same observation value is assigned.

相関仮説生成部４１から出力された相関仮説毎に次のステップＳＴ２〜ステップＳＴ６の処理を行う。There are several ways to assign the wakes and observations. When the number of allocation methods is N, N new correlation hypotheses are generated using each allocation method and the evaluation value of the parent correlation hypothesis. This correlation hypothesis is called the child correlation hypothesis.

The following steps ST2 to ST6 are performed for each correlation hypothesis output from the correlation hypothesis generation unit 41.

The observation value degeneracy determination unit 42 determines whether or not the observed value is degenerated in the correlation hypothesis acquired from the correlation hypothesis generation unit 41, and outputs the determined correlation hypothesis and the determination result of the correlation hypothesis in association with each other. (Step ST2). For example, when the same observation value is assigned to a plurality of tracks, a determination result indicating that the observation value has degenerated is output (step ST2-with degeneration). Otherwise, a determination result indicating that the observed value has not degenerated is output (step ST2-no degeneration).
The super-resolution angle measurement unit 51 in the degeneration cancellation unit 50 acquires the correlation hypothesis from the observation value degeneration determination unit 42 when the determination result output from the observation value degeneration determination unit 42 indicates the degeneration of the observed value, A digital signal is obtained from the receiving circuit 12. The super-resolution angle measurement unit 51 performs super-resolution angle measurement using the acquired digital signal and the track included in the acquired correlation hypothesis, and outputs the angle measurement result and the correlation hypothesis input from the observation value degeneration determination unit 42. (Step ST3).

方位角の尤度関数は例えば次のように表現される。

最尤法による測角は、尤度関数を最大とするような方位角ベクトルθ_ｋの探索によって行われる。探索方法としては、例えばＥＭ（Ｅｘｐｅｃｔａｔｉｏｎ Ｍａｘｉｍｉｚａｔｉｏｎ）アルゴリズムがある。

The azimuth likelihood function is expressed, for example, as follows.

Angle measurement by the maximum likelihood method is performed by searching for an azimuth vector θ _k that maximizes the likelihood function. As a search method, for example, there is an EM (Expectation Maximization) algorithm.

  FIG. 4 shows an image of the maximum likelihood method using the azimuth angle of the expected value of the predicted position of the wake. FIG. 4 shows a case where the number of tracks included in the correlation hypothesis is 2 (tracks 401 and 402). In this figure, for the sake of explanation, it is assumed that a target exists in two dimensions of x and y. I have. The axis orthogonal to this plane is an axis representing the likelihood of the azimuth. Here, as an example, the likelihood function of the maximum likelihood method is represented by a solid line 403, and the likelihood function of the FB-SSP MUSIC method is represented by a dotted line 404. ing. In each angle measurement method, a peak portion of the likelihood function is a candidate for an angle measurement value. The likelihood function of the FB-SSP MUSIC method in FIG. 4 reduces the correlation between the signals of each target by the spatial averaging method when the signals output from each target have high correlation and are close to each other. An example is shown where it is not possible. From FIG. 4, when the angle measurement is performed by the FB-SSP MUSIC method, since the shape of the likelihood function is unimodal, the obtained angle measurement value is one (the mark 405 in FIG. 4), and the observed value is degenerated. You can see that In addition, when the search is started near the azimuth which becomes the local solution by the maximum likelihood method, the maximum likelihood method outputs the local solution (mark 406 in FIG. 4) as the angle measurement value. Here, when the predicted positions of the wakes 407 and 408 are sufficiently close to the global solution (marks 409 and 410 in FIG. 4), the azimuths of the predicted positions of the wakes are set to the initial values 411 and 412 and the maximum value of the likelihood function is searched. (Arrows 413 and 414), the angle measurement values 409 and 410 corresponding to the global solution of the likelihood function of the maximum likelihood method are obtained, and it is considered that the observed value does not degenerate. In the drawing, circles 415 and 416 indicate the convergence values of the search.

  In the above, the method of using the azimuth of the expected value of the predicted position of the track as the initial value of the maximum likelihood method has been described. May be used, and the value having the highest likelihood among the angle measurement values may be used as the angle measurement result of the super-resolution angle measurement unit 51.

  The correlation hypothesis updating unit 52 replaces the observation value included in the correlation hypothesis input from the super-resolution angle measurement unit 51 with the observation value calculated from the angle measurement result input from the super-resolution angle measurement unit 51. The generated correlation hypothesis is output (step ST4). The method of correlating the observed value with the track is determined, for example, by storing the track used as an initial value for each angle measurement value obtained by the super-resolution angle measuring unit 51 to determine the observation value to be assigned to each track. There is a way to do it.

  The correlation hypothesis evaluation unit 43 acquires the correlation hypothesis from the observation value degeneracy determination unit 42 or the correlation hypothesis update unit 52, evaluates the likelihood of the correlation hypothesis, updates the evaluation value included in the correlation hypothesis, and updates the evaluation value. The updated correlation hypothesis is output (step ST5). Evaluation of the correlation hypothesis is described, for example, in the literature: BT Vo and BN Vo, “Labeled Random Finite Set and Multi-Object Conjugate Priors”, IEEE Trans. On Signal Processing, vol. 61, no. 13, pp. 3460-3475, It can be performed in the same way as the multi-target tracking algorithm in 2013.

The track updating unit 44 acquires the correlation hypothesis from the correlation hypothesis evaluation unit 43, and updates each track included in the correlation hypothesis using the correlated observation value (step ST6). For example, the wake can be updated using a Kalman filter as in the following equation (3).

The above steps ST2 to ST6 are performed for each correlation hypothesis.
The correlation hypothesis storage unit 45 performs a process of receiving all correlation hypotheses output from the wake update unit 44 and outputting them to a memory or the like (step ST7).
The track determination unit 46 receives all correlation hypotheses output from the track update unit 44, and outputs a track included in the correlation hypothesis having the highest evaluation value as a target track at time k (step). ST8). The output track is output, for example, to a display (not shown) as a display track.

  As described above, the target tracking apparatus according to Embodiment 1 performs super-resolution angle measurement in the hypothesis that the observation value is degenerated, and updates the correlation hypothesis. Since it is possible to update the track using unobserved observation values, errors in estimating the target position and speed can be reduced. As a result, the occurrence of a wrong track can be prevented, so that the estimation error of the target number can be reduced.

  As described above, according to the target tracking device of the first embodiment, the sensor processing unit that acquires observation signals of a plurality of targets, and the observation value extraction unit that generates observation values from the observation signals acquired by the sensor processing unit , Using the observation values generated by the observation value extraction unit, determine the position specification estimates of multiple targets using the set position estimation method, and determine whether the observed values have degenerated from the position specification estimates When the multi-target tracking processing unit determines that there is degeneration of the observed value, a position estimation method that performs target estimation based on the predicted value of the multi-target for one observation value. A multivariate tracking processing unit, when it is determined that there is degeneration of the observed value, the multi-target tracking processing unit uses a correlation hypothesis that determines that there is degeneration of the observed value. The wake is updated with the position specification estimated value Since the way, in the case where the correlation of the incoming waves is higher, it is possible to track updates using the observed values does not degenerate, hence, it is possible to reduce the estimation error of the target position and velocity. As a result, the occurrence of a wrong track can be prevented, and the estimation error of the target number can be reduced.

  Further, according to the target tracking device of the first embodiment, the multi-target tracking processing unit determines a correlation hypothesis that a plurality of wakes correlates with a single observation value for the observation values and the wakes input from the observation value extraction unit. The generation and degeneration elimination unit updates the observation value using the wake included in the correlation hypothesis generated by the multi-target tracking processing unit, so even if the correlation between the incoming waves is high, the observation value degeneration is performed. Can be avoided.

  According to the target tracking device of the first embodiment, the degeneration elimination unit updates the observation value using the expected value of the wake prediction input from the multi-target tracking processing unit. Even if the correlation is high, it is possible to avoid degeneration of the observed value.

  Further, according to the target tracking device of the first embodiment, the degeneration cancellation unit samples a plurality of times from the distribution of predicted wakes input from the multi-target tracking processing unit, performs an update process of an observation value for each sample, Since the likely observation value is output as the updated observation value in the update processing result, degeneration of the observation value can be avoided even when the correlation between the incoming waves is high.

  Further, according to the target tracking apparatus of the first embodiment, since the position estimation method for estimating a target based on the prediction values of multiple targets for one observation value is a super-resolution angle measurement, the distance between arriving waves Even when the correlation is high, degeneration of the observed value can be avoided.

  Further, according to the target tracking device of the first embodiment, since the super-resolution angle measurement is the angle measurement by the maximum likelihood method, it is possible to avoid degeneration of the observed value even when the correlation between the incoming waves is high.

Embodiment 2 FIG.
The first embodiment is an example in which a signal output from the target itself is received and the target is tracked. However, the sensor processing unit has a transmission unit, irradiates the target with a signal, and outputs the target from the signal reflected by the target. May be performed, and this example will be described below as a second embodiment.
FIG. 5 is a configuration diagram of the target tracking device according to the second embodiment. The sensor processing unit 10a according to the second embodiment includes a transmission unit 14 and a reception unit 13. The transmission section 14 includes a transmission antenna section 15 and a transmission circuit 16. The transmission antenna unit 15 is an antenna that converts an analog signal generated by the transmission circuit 16 into a radio wave and radiates the radio wave into space. The transmission circuit 16 is a circuit that generates an analog signal of a radio wave radiated from the transmission antenna unit 15 into space. The receiving unit 13 includes a receiving antenna unit 11 and a receiving circuit 12, and the receiving antenna unit 11 and the receiving circuit 12 have the same configuration as in the first embodiment. Since the configurations of the observation value extraction unit 20 and the tracking processing unit 30 are the same as those in the first embodiment, the corresponding parts are denoted by the same reference numerals and description thereof will be omitted. Further, regarding the hardware configuration of the target tracking device according to the second embodiment, a program corresponding to the function of the transmission circuit 16 is stored in the storage 104 in the hardware configuration of FIG. The second embodiment is the same as the first embodiment except that the function of the transmission circuit 16 is realized by executing the above.

Next, the operation of the target tracking device according to the second embodiment will be described.
The transmitting circuit 16 generates an analog signal for irradiating the target, and the transmitting antenna unit 15 converts the analog signal into a radio wave and radiates it to space. In the second embodiment, the reception circuit 12 of the reception unit 13 detects a portion corresponding to the radio wave reflected at the target from the digital signal generated by the reception circuit 12 by referring to the signal generated by the transmission circuit 16. , And outputs only the detected signal. Note that, similarly to the receiving circuit 12 of the first embodiment, in the second embodiment, a radio signal output from a target for purposes such as communication may also be output.
The observation value extraction unit 20 calculates the distance measurement value and the angle measurement value using the digital signal input from the receiving circuit 12, and outputs the target estimated position on the coordinates. The operation of the tracking processing unit 30 is the same as that of the first embodiment, and a description thereof will be omitted.

  As described above, according to the target tracking device of the second embodiment, the sensor processing unit includes the transmission unit that outputs the signal for irradiating the target, and monitors the signal output from the transmission unit and reflected from the target. Since the signal is transmitted from the target tracking device, the signal is reflected by the target and the target is tracked, so that the target is tracked even when the target itself does not output a signal. It can be performed.

Embodiment 3 FIG.
In the first embodiment and the second embodiment, all the correlation hypotheses are stored in the correlation hypothesis storage unit 45. However, only a part of the correlation hypotheses may be stored. Embodiment 3 will be described below.

  FIG. 6 is a configuration diagram of the target tracking device according to the third embodiment. The multi-target tracking processing unit 40a of the tracking processing unit 30a according to the third embodiment includes a correlation hypothesis selecting unit 47. The correlation hypothesis selector 47 is configured to select a correlation hypothesis to be stored in the correlation hypothesis storage 45 based on the evaluation value of the correlation hypothesis. Other configurations in the multi-target tracking processing unit 40a and the configurations of the sensor processing unit 10, the observation value extraction unit 20, and the degeneration elimination unit 50 are the same as those of the first embodiment shown in FIG. Corresponding parts have the same reference characters allotted, and description thereof will not be repeated.

  In the hardware configuration of the target tracking device according to the third embodiment, a program corresponding to the function of the correlation hypothesis selecting unit 47 is stored in the storage 104 in the hardware configuration in FIG. 2, and the processor 101 stores the program in the memory 102. The configuration is the same as that of the first embodiment except that the function of the correlation hypothesis selecting unit 47 is realized by reading and executing.

  The correlation hypothesis selecting unit 47 of the third embodiment uses the evaluation value for each correlation hypothesis output from the wake updating unit 44 to select a correlation hypothesis to be stored in the correlation hypothesis storage unit 45. For example, an operation of selecting only a correlation hypothesis having an evaluation value exceeding a threshold value determined in advance based on a simulation or the like is performed. As a result, only the correlation hypothesis selected by the correlation hypothesis selection unit 47 is stored in the correlation hypothesis storage unit 45, and the correlation hypothesis generation unit 41 operates based on the correlation hypothesis stored in the correlation hypothesis storage unit 45. . Other operations are the same as those in the first embodiment, and a description thereof will not be repeated.

  In the above example, the example in which the correlation hypothesis selecting unit 47 is applied to the configuration of the first embodiment has been described. However, the example may be applied to the configuration of the second embodiment.

  As described above, according to the target tracking device of the third embodiment, the multi-target tracking processing unit includes a correlation hypothesis selecting unit that selects a correlation hypothesis to be stored based on the evaluation value of the correlation hypothesis, and a selected correlation hypothesis. A correlation hypothesis generation unit that generates a new correlation hypothesis using the hypothesis and the observed value is provided, and the presence or absence of degeneration of the observed value is determined based on the correlation hypothesis generated by the correlation hypothesis generation unit. The capacity of the memory for storing the hypothesis can be reduced. Further, the number of correlation hypotheses to be processed by the target tracking device at each time can be reduced, so that the calculation load can be reduced.

Embodiment 4 FIG.
In the first to third embodiments, a wake is updated using a Kalman filter on the assumption that a one-to-one correlation between an observed value and a wake is used. However, a particle filter is used as a wake update method that achieves one-to-many correlation. An example is shown below as Embodiment 4.
FIG. 7 is a configuration diagram of a target tracking device according to the fourth embodiment.
FIG. 7 is a diagram in which the super-resolution angle measurement unit 51 and the correlation hypothesis updating unit 52 of the degeneration eliminating unit 50 in FIG. The degenerate hypothesis track updating unit 53 updates the trajectory of the degenerate hypothesis using the particle filter, which is a nonlinear filter, when the observation value degeneration determining unit 42 of the multi-target tracking processing unit 40b determines that the observed value has degenerated. It is a processing unit. That is, in the fourth embodiment, a particle filter is used as a position estimation method for estimating a target based on a multi-target prediction value for one observation value, and the degeneracy eliminating unit 50a updates the position specification estimation value. Outputs the track that was performed. Accordingly, the correlation hypothesis evaluation unit 43a is configured to evaluate the correlation hypothesis using the likelihood calculation result of the degenerate hypothesis track update unit 53. The other configurations of the sensor processing unit 10 and the observation value extraction unit 20 are the same as those of the first embodiment, and a description thereof will not be repeated.

  Further, regarding the hardware configuration of the target tracking device of the fourth embodiment, in the hardware configuration of FIG. 2, a program corresponding to the function of the correlation hypothesis evaluating unit 43a of the degeneration eliminating unit 50a and the multi-target tracking processing unit 40b is stored. The configuration is the same as that of the first embodiment except that the configuration is such that the functions of the degeneration eliminating unit 50a and the correlation hypothesis evaluating unit 43a are realized by the processor 101 reading the program into the memory 102 and executing it.

  Next, the operation of the target tracking device according to the fourth embodiment will be described. The operations of the sensor processing unit 10 and the observation value extraction unit 20 are the same as those in the first embodiment, and thus description thereof will be omitted, and the operation of the tracking processing unit 30b will be described. FIG. 8 is a flowchart illustrating the operation of the tracking processing unit 30b according to the fourth embodiment. First, the operations of step ST1 to step ST2 (no degeneration) to step ST5 to step ST6 to step ST7 to step ST8 are the same as those in the first embodiment. That is, when the degeneration of the observed value does not occur, the operation is the same as the operation of the tracking processing unit 30 of the first embodiment.

As described above, in order to cope with the non-linear degeneration observation process in the fourth embodiment, the wake of the degeneracy hypothesis is updated using the particle filter. Can be estimated. In addition, the wake update of the hypothesis that the observation value does not degenerate has the same configuration as in the first embodiment, so that the calculation time can be reduced.
In the above example, the example in which the degeneration eliminating unit 50a is applied to the configuration of the first embodiment is described. However, the example may be applied to the configuration of the second or third embodiment.

  As described above, according to the target tracking apparatus of the fourth embodiment, the position estimation method for performing target estimation based on the prediction values of multiple targets with respect to one observation value is a particle filter. Even when the correlation is high, it is possible to update the track using a nonlinear filter corresponding to the degeneration, and therefore, it is possible to reduce the estimation error of the target position and speed. As a result, the occurrence of a wrong track can be prevented, and the estimation error of the target number can be reduced.

  In the present invention, any combination of the embodiments, a modification of an arbitrary component of each embodiment, or an omission of an arbitrary component in each embodiment is possible within the scope of the invention. .

  As described above, the target tracking device according to the present invention relates to a configuration for switching the position estimation method of a target according to the presence or absence of degeneration of an observed value, and is suitable for tracking a plurality of targets.

  10, 10a sensor processing section, 11 receiving antenna section, 12 receiving circuit, 13 receiving section, 14 transmitting section, 15 transmitting antenna section, 16 transmitting circuit, 20 observation value extracting section, 30, 30a, 30b tracking processing section, 40, 40a, 40b Multi-target tracking processing unit, 41 correlation hypothesis generation unit, 42 observation value degeneracy determination unit, 43, 43a correlation hypothesis evaluation unit, 44 wake update unit, 45 correlation hypothesis storage unit, 46 wake determination unit, 47 correlation hypothesis selection , 50, 50a Degeneration cancellation unit, 51 Super-resolution angle measurement unit, 52 Correlation hypothesis update unit, 53 Degeneration hypothesis track update unit.

Claims (9)

A sensor processing unit for acquiring observation signals of a plurality of targets,
An observation value extraction unit that generates an observation value from the observation signal acquired by the sensor processing unit,
Using the observation value generated by the observed value extractor, multiple targets using the set position estimation method, the correlating hypothesis comprising a track and the observed value and the evaluation value of the correlation hypothesis, the correlation hypothesis A multi-target tracking processing unit for determining whether or not the observed value has degenerated from
If it is determined that there is degeneration of the observation value in the multi-target tracking processing unit, based on the prediction value of the multi-target for one observation value, the position estimation method used by the multi-target tracking processing unit Comprises a degeneracy elimination unit that obtains an observation value without degeneration of a plurality of targets using a position estimation method that performs different target estimations,
The multi-target tracking processing unit, when it is determined that there is no degeneration of the observation value, updates the wake of the correlation hypothesis determined that there is no degeneration of the observation value using the observation value included in the correlation hypothesis. the case where it is determined that there is a degeneration of the observations, the track of the correlation hypothesis is determined that there is degeneracy of the observations, are included in the correlation hypothesis, the no degeneracy obtained in a degraded resolution block observations A target tracking device characterized by updating using a target. The multi-target tracking processing unit, for the observation values and wakes input from the observation value extraction unit, the correlation hypothesis that allocates a plurality of wakes per observation value to the observation value input from the observation value extraction unit Generate
The no observed value degeneracy obtained by degeneration canceller, the multi-target tracking process target tracking apparatus according to claim 1, wherein the determined using the track contained in the generated correlation hypothesis unit. The no observed value degeneracy obtained by degeneration canceller, the target tracking apparatus according to claim 1, characterized in that it is determined using the expected value of the prediction of the trajectory input from the multi-target tracking process section. 2. The observation value without degeneration determined by the degeneration cancellation unit is sampled a plurality of times from the distribution of predicted wakes input from the multi-target tracking processing unit, and is obtained for each sample. Target tracking device. 2. The target tracking device according to claim 1, wherein the position estimation method used by the degeneration cancellation unit is performed using super-resolution angle measurement.   The target tracking device according to claim 5, wherein the super-resolution angle measurement is an angle measurement by a maximum likelihood method.   The said sensor processing part is provided with the transmission part which outputs the signal irradiated to a target, The signal output from the said transmission part and reflected from the target is made into the said observation signal, The said observation signal. The target tracking device according to claim 1. The multi-target tracking processing unit includes a correlation hypothesis selection unit that selects a correlation hypothesis to be stored based on the evaluation value of the correlation hypothesis ,
The target according to any one of claims 1 to 6 , wherein the track obtained by the multi-target tracking processing unit is obtained using a correlation hypothesis selected by the correlation hypothesis selecting unit. Tracking device. 2. The target tracking apparatus according to claim 1, wherein the position estimating method used by the degeneration cancellation unit is performed using a particle filter.

Images