JP5701152B2 - TRACKING DEVICE, COMPUTER PROGRAM, AND TRACKING METHOD - Google Patents

Description

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

A tracking device that estimates the target track and eliminates the erroneous track due to the unnecessary signal by determining which of the detection data obtained from the sensor is the target and which is the unnecessary signal, and tracks the target. is there. The tracking device determines whether the detection data is “currently tracking target”, “newly detected target”, or “unnecessary signal”. This is called “correlation determination”. The tracking device estimates the track of the target based on the detection data determined to be the same target by the correlation determination, and tracks the target. In the correlation determination, for example, a correlation likelihood of each interpretation is calculated, and an interpretation having a large correlation likelihood is employed.
In addition, there is a tracking device that does not determine correlation immediately after obtaining detection data, but holds a plurality of hypotheses and determines which hypothesis is adopted after the detection data is accumulated.

JP 2002-168949 A

D. B. Reid, “An algorithm for tracking multiple targets”, IEEE Transactions on Automatic Control, vol. 24, no. 6, pp. 843-854, 1979.

When the reliability of the estimated track is low, such as when the target has just started tracking, the correct correlation may not be determined. In order to avoid this, when holding a plurality of hypotheses, the processing load increases as the number of hypotheses increases.
The present invention has been made, for example, in order to solve the above-described problems, and an object thereof is to carry out correct correlation determination.

The tracking device according to the present invention is
A detection data acquisition unit, a wake storage unit, a prediction unit, a likelihood calculation unit, a correlation unit, an update unit, and a reliability determination unit,
The detection data acquisition unit acquires detection data representing an observed value of the state quantity of the target and an observation time when the sensor observed the target for the target estimated to be observed by the sensor,
The track storage unit stores track data representing an estimated value of the state quantity of the target for the target estimated to be observed by the sensor,
The prediction unit determines whether the detection data is based on an estimated value of a target state quantity represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. Calculate a predicted value of the state quantity of the target represented by the track data at the observation time of the target represented,
The likelihood calculation unit includes a predicted value of the target state quantity calculated by the prediction unit and the detection data for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. A likelihood that the target represented by the track data and the target represented by the detection data are the same based on the observed state value of the target represented by
The correlation unit is a target represented by the track data based on the likelihood calculated by the likelihood calculation unit for a set of the track data stored by the track storage unit and the detection data acquired by the detection data acquisition unit. And whether or not the target represented by the detection data is the same,
The update unit is configured such that for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit, a target represented by the track data and a target represented by the detection data are the same. When the correlator determines, the wake data is updated based on the observed value of the target state quantity represented by the detection data,
The reliability determination unit determines whether the reliability of the wake data is high or low for the wake data stored in the wake storage unit,
The likelihood calculation unit predicts a target state quantity calculated by the prediction unit for a set of track data determined by the reliability determination unit and detection data acquired by the detection data acquisition unit when reliability is high. The larger the difference between the value and the observed value of the target state quantity represented by the detection data, the smaller the likelihood is calculated. For a set with the acquired detection data, a constant likelihood is calculated regardless of the difference between the predicted value of the target state quantity calculated by the prediction unit and the observed value of the target state quantity represented by the detection data. It is characterized by.

  According to the tracking device according to the present invention, the likelihood calculating unit is configured to calculate the prediction unit for the set of the track data determined by the reliability determining unit and the detection data acquired by the detection data acquiring unit when the reliability is low. Regardless of the difference between the predicted value of the target state quantity calculated by and the observed value of the target state quantity represented by the detection data, a certain likelihood is calculated, so that a correct correlation can be determined.

1 is a system configuration diagram illustrating an example of an overall configuration of a tracking system 800 according to Embodiment 1. FIG. FIG. 3 is a hardware diagram illustrating an example of a hardware configuration of the tracking device 100 according to the first embodiment. FIG. 3 is a block diagram illustrating an example of a functional block configuration of the tracking device 100 according to the first embodiment. FIG. 6 is a flowchart showing an example of a flow of tracking processing S600 in the first embodiment. The figure which shows an example of the observation condition of detection data. FIG. 4 is a graph showing an example of correlation likelihood calculated by a correlation likelihood calculation unit 133 in the first embodiment. The flowchart figure which shows an example of the flow of 1st likelihood calculation process S636 in Embodiment 2. FIG. FIG. 18 is a detailed block diagram illustrating an example of a configuration of a wake data update unit according to the sixth embodiment. FIG. 20 is a flowchart showing an example of the flow of tracking processing S600 in the sixth embodiment. The figure which shows an example of the observation condition of detection data. FIG. 18 is a block diagram illustrating an example of a configuration of a tracking device 100 according to a seventh embodiment. FIG. 20 is a flowchart showing an example of the flow of tracking processing S600 in the seventh embodiment. The figure which shows an example of the observation condition of detection data.

Embodiment 1 FIG.
The first embodiment will be described with reference to FIGS.

FIG. 1 is a system configuration diagram showing an example of the overall configuration of a tracking system 800 in this embodiment.
The tracking system 800 includes, for example, a sensor 801, a tracking device 100, and a display device 802.

The sensor 801 is, for example, a radar or an infrared camera. The sensor 801 observes the target and outputs detection data. The detection data is data representing the observed value of the target state quantity observed by the sensor 801 and the observation time. The target state quantity represented by the detection data is, for example, the target position and speed. The information that can be observed differs depending on the type of sensor 801, and accordingly, the target state quantity represented by the detection data also differs. For example, in the case of a sensor 801 capable of observing not only the position of the target but also the speed in the distance direction between the sensor 801 and the target, such as Doppler radar, the detection data represents the position of the target and the speed in the distance direction. Further, in the case of a sensor 801 that can observe only the two-dimensional direction in which the target can be seen, not the three-dimensional position of the target, such as an infrared camera, the detection data represents the direction in which the target can be seen from the sensor 801. The detection data need not represent all the state quantities that can be observed by the sensor 801, but may be configured to represent only the state quantities necessary for the tracking device 100 to track the target.
The observation time indicated by the detection data is the time when the sensor 801 observes the state quantity. For example, the detection data includes data representing the elapsed time from the predetermined time in seconds. Alternatively, in the case of a sensor 801 that repeatedly performs observation at a constant cycle, such as an infrared camera, the detection data may be configured to include data representing a frame number, for example. Further, when the observation time (frame number) is expressed by outputting the detection data as a group for each frame, the detection data does not include data indicating the frame number or the like. May be.
The sensor 801 generates detection data based on the observation result obtained by observing the target.

However, because of the influence of observation noise and the like, the detection data does not necessarily represent the target observed by the sensor 801. This is called an “unnecessary signal”. For example, the sensor 801 generates detection data on the assumption that the target is observed when the observed signal is stronger than a preset signal strength threshold. However, even when the target does not actually exist, the signal observed by the sensor 801 may exceed the signal strength threshold due to the observation noise. If the signal strength threshold is set high, the generation of unnecessary signals can be suppressed, but instead, there is a high possibility that the target will be missed when the target actually exists. For this reason, it is desirable to set the signal strength threshold to a relatively low value. The generation probability of the unnecessary signal is a function of the signal strength threshold.
In addition, because of the influence of observation error, the observed value of the target state quantity represented by the detection data may be different from the true value of the target state quantity.
The sensor 801 may be configured to output observation data representing the observed raw observation result. Alternatively, the sensor 801 may be configured to output intermediate data in the middle of generating detection data from observation data.

The tracking device 100 receives detection data (or observation data) output from the sensor 801. The sensor 801 and the tracking device 100 are connected directly or via a network or the like. The tracking device 100 inputs detection data (or observation data or the like) directly from the sensor 801 or via a network or the like. The tracking device 100 tracks the target based on the input detection data. When the sensor 801 is configured to output observation data and intermediate data, the tracking device 100 generates detection data based on the input data, for example, and tracks the target based on the generated detection data. Alternatively, a device that generates detection data from observation data or intermediate data is interposed between the sensor 801 and the tracking device 100, and the tracking device 100 may be configured to input the detection data generated by the device. .
For example, the tracking device 100 determines whether or not the detection data is an unnecessary signal, and estimates a target track based on the detection data determined to be not an unnecessary signal. The tracking device 100 determines whether a plurality of detection data based on observation results observed by the sensor 801 at different observation times are observed for the same target or different targets. The tracking device 100 estimates state quantities such as the position and speed of a target that changes over time based on detection data that is determined to have observed the same target.
The tracking device 100 outputs wake data. The wake data represents an estimated value of the target state quantity calculated by the tracking device 100.

The display device 802 inputs the track data output from the tracking device 100. The display device 802 displays state quantities such as the target position and speed on the display screen based on the input track data.
The display device 802 is an example of a device that uses track data output by the tracking device 100. The track data output from the tracking device 100 can be used by various devices other than the display device 802. The tracking system 800 may be configured to include another device that inputs and uses the wake-up data output from the tracking device 100 instead of the display device 802 or in addition to the display device 802.

FIG. 2 is a hardware diagram illustrating an example of a hardware configuration of the tracking device 100 according to this embodiment.
The tracking device 100 is, for example, a computer. The tracking device 100 includes, for example, a storage device 914, a processing device 911, an input device 912, and an output device 913.
The storage device 914 stores a computer program executed by the processing device 911, data processed by the processing device 911, and the like. The storage device 914 is, for example, a volatile memory, a nonvolatile memory, a semiconductor memory, a magnetic disk device, or the like.
The processing device 911 processes the data by executing the computer program stored in the storage device 914 and controls the entire tracking device 100.
The input device 912 inputs information from outside the tracking device 100 and converts it into data that can be processed by the processing device 911. The input device 912 inputs, for example, an analog-digital conversion device that converts an analog signal into digital data, a communication device (reception device) that receives and demodulates a signal transmitted by another device such as the sensor 801, and a human operation. Operation input devices such as a keyboard and a mouse. The data converted by the input device 912 may be directly processed by the processing device 911 or may be temporarily stored by the storage device 914.
The output device 913 converts the data processed by the processing device 911 and the data stored in the storage device 914 into a format that can be output to the outside of the tracking device 100 and outputs the converted data. The output device 913 is, for example, a digital-analog conversion device that converts digital data into an analog signal, a communication device (transmission device) that transmits a signal representing data to another device such as the display device 802, a human-visible image or a human A device that converts data into sound that can be heard and outputs it.

The functional blocks of the tracking device 100 described below are realized by the processing device 911 executing a computer program stored in the storage device 914, for example.
However, this is an example, and the functional block of the tracking device 100 is realized by a dedicated integrated circuit, an electric circuit configured by digital parts / analog parts, or other configurations such as a mechanical configuration. There may be.
In addition, the tracking device 100 does not have to be a physically integrated device, and each function block is realized by a different device, and a plurality of devices operate in a coordinated manner, so that tracking is performed as a whole. A configuration that functions as the device 100 may be used.

FIG. 3 is a block diagram illustrating an example of a functional block configuration of the tracking device 100 according to this embodiment.
The tracking device 100 includes, for example, a detection data acquisition unit 110, a track data storage unit 120, a gate calculation unit 131, a gate inside / outside determination unit 132, a correlation likelihood calculation unit 133, a track data update unit 134, a track An establishment determination unit 135 and a wake output unit 190 are included.

  The detection data acquisition unit 110 acquires the detection data 511 using the input device 912. For example, the detection data acquisition unit 110 uses the input device 912 to input detection data output by the sensor 801 or the like. Alternatively, the detection data acquisition unit 110 uses the input device 912 to input observation data output from the sensor 801 and the like, and uses the processing device 911 to generate detection data from the input observation data. The detection data acquisition unit 110 outputs the acquired detection data 512 using the processing device 911.

  The wake data storage unit 120 (wake storage unit) stores the wake group 521 using the storage device 914. The wake group 521 is composed of one or more wake data. One track data represents an estimated value of the state quantity for one target tracked by the tracking device 100. The track group 521 includes the same number of track data as the number of targets tracked by the tracking device 100.

The gate calculation unit 131 (prediction unit) uses the processing device 911 to calculate a gate for each target tracked by the tracking device 100 based on the track group 521 stored in the track data storage unit 120. The gate represents a range of observation values of the state quantity of the target in which the target represented by the track data and the target represented by the detection data may be the same target. For example, the gate calculation unit 131 uses the processing device 911 to calculate a predicted value of the target state quantity at the observation time of the detection data based on the estimated value of the target state quantity represented by the track data. The gate calculation unit 131 uses the processing device 911 to calculate a gate centered on the predicted value based on the calculated predicted value of the target state quantity. The gate calculation unit 131 outputs the calculated gate 531.
There may be a case where a plurality of detection data exist at one observation time, such as when the sensor 801 repeats observation at a predetermined cycle and the observation time of detection data is represented by a frame number or the like. In that case, the gate calculation unit 131 calculates the gate at the observation time. When the observation time is different for each detection data, the gate calculation unit 131 calculates a gate at the observation time of the detection data for each detection data. The gate calculation unit 131 may be configured to regard the predetermined time as the observation time of the detection data in which the observation time falls within a predetermined period centered on that time, and calculate the gate at that time.

When the wake data is configured to include the target position and speed as state quantities, the gate calculation unit 131, for example, has a constant target speed based on an estimated value of the target position and speed represented by the wake data. The target position at the observation time is predicted under the assumption that. Actually, the target speed is not constant, and the position of the target is affected by the estimation error of the speed. Therefore, the position where the target is observed may be different from the position predicted by the gate calculation unit 131. For example, the prediction error is estimated in advance, and the gate size is set in advance based on the estimation result. The gate calculation unit 131 calculates a range of a preset size around the predicted position as a gate.
In addition, when the detection data includes a target position as a state quantity and does not include a speed as a state quantity, such as when the sensor 801 observes only the target position, for a new target that has just started tracking, I don't know the speed. Since it is not known in which direction the target travels, the gate calculation unit 131 calculates a gate centered on the position where the target was observed last time. In addition, since the target speed is unknown, the gate calculation unit 131 calculates a gate having a size that includes a position that is considered to be observed when the target travels at the assumed maximum speed. Such a gate is called an “initial gate”.

  For example, the gate calculation unit 131 calculates a gate 531 represented by the following expression.

G = {z | (z−z ^* ) ^T S ⁻¹ (z−z ^* ) ≦ d}

However, G is a set and represents the gate 531. z is a vertical vector and represents an observed value of the target state quantity. For example, when the detection data 512 uses a target three-dimensional position as a state quantity, the vertical vector z is a three-dimensional vector whose elements are target coordinates in a three-dimensional orthogonal coordinate system. z ^* is a vertical vector having the same dimension as the vertical vector z, and represents a predicted value of the target state quantity. The superscript “T” represents transposition of the matrix, and (z−z ^* ) ^T represents a horizontal vector obtained by transposing the vertical vector (z−z ^* ). S is a square matrix having the same dimension as the vertical vector z, and represents the size and shape of the gate. The value of the element of S is, for example, a preset constant. The superscript “−1” represents the inverse matrix, and S ⁻¹ represents the inverse matrix of the square matrix S. d is a positive real number and represents the size of the gate. The value of d is, for example, a preset constant.

When the gate 531 is an initial gate, the gate calculation unit 131, for example, a vertical vector z ^* whose elements are coordinates in the three-dimensional orthogonal coordinate system at the position where the target was observed last time, and a square represented by the following equation: The gate 531 is calculated using the matrix S.

^{_{S = (V max · ΔT)}} 2/9 · I + R

However, V _max is a positive real number, representing the maximum rate envisaged goals. ΔT is a positive real number and represents the elapsed time from the time when the target was observed last time to the target observation time indicated by the detection data 512. I is a unit matrix having the same dimensions as the square matrix S. R is a square matrix having the same dimension as the square matrix S, and is an observation error covariance matrix representing the observation error covariance of the sensor 801.

  The gate inside / outside determination unit 132 uses the processing device 911 to detect the detection data 512 output from the detection data acquisition unit 110 based on the gate calculated by the gate calculation unit 131 for each target tracked by the tracking device 100. Determine if you are in the gate. The gate inside / outside determination unit 132 outputs the determined gate inside / outside determination result 532 using the processing device 911.

  The correlation likelihood calculation unit 133 (likelihood calculation unit, reliability determination unit) uses the processing device 911 to determine the gate internal / external determination result 532 determined by the gate internal / external determination unit 132 for each target tracked by the tracking device 100. Based on, the correlation likelihood 533 is calculated for the detection data 512 entering the target gate 531. Correlation likelihood 533 is an index representing the likelihood that the target represented by the track data and the target represented by detection data 512 are the same. The larger the correlation likelihood 533, the higher the possibility that the target represented by the track data and the target represented by the detection data 512 are the same. Correlation likelihood calculation section 133 is a combination of each of the wake data included in wake group 521 and each of detection data 512 output from detection data acquisition section 110, and the wake at the target observation time represented by that detection data 512. The correlation likelihood 533 is calculated for the combination in which the detection data 512 enters the target gate 531 represented by the data, and the correlation likelihood is not calculated for the other combinations. Note that the correlation likelihood calculation unit 133 indicates a correlation likelihood (for a combination in which the detection data 512 does not enter the gate 531) indicating that the target represented by the wake data and the target represented by the detection data 512 are not the same. For example, “0” may be calculated.

  The correlation likelihood calculation unit 133 calculates the correlation likelihood 533 using different calculation formulas depending on whether the reliability of the track data is high or low. For example, when the track data represents a new target that has just started tracking, the correlation likelihood calculating unit 133 determines that the reliability of the track data is low, and in other cases, the reliability of the track data is low. Judge as high.

  When it is determined that the reliability of the track data is high, the correlation likelihood calculation unit 133 calculates, for example, a correlation likelihood 533 represented by the following expression.

_{g = P d / √ [det} (2πS)] · exp [-1/2 · (z-z *) T S -1 (z-z *)]

However, g is a positive real number and represents the correlation likelihood 533. P _d is a real number larger than 0 and smaller than 1, and represents a target detection probability for detecting a target. det represents a determinant. π is the circumference ratio. exp represents an exponential function with the Napier number e as the base.

The correlation likelihood 533 calculated using this equation is smaller as the difference between the vertical vector z and the vertical vector z ^* is larger. For example, when the vertical vectors z and z ^* are three-dimensional vectors whose elements are target coordinates in a three-dimensional orthogonal coordinate system, the correlation likelihood calculating unit 133 determines the position of the observed target and the predicted target. The smaller the likelihood between the positions is, the smaller the correlation likelihood 533 is calculated. The correlation likelihood 533 is a monotonically decreasing function having the distance between the predicted position of the tracking track and the detection data as a variable.

  When it is determined that the reliability of the track data is low, the correlation likelihood calculating unit 133 calculates, for example, a correlation likelihood 533 represented by the following expression.

g = max {β _FT , β _NT }

However, max represents a maximum value. β _FT is a positive real number and represents the likelihood that the detection data 512 is an unnecessary signal. The value of β _FT is, for example, a predetermined constant, and is set by calculating the number of unnecessary signals generated per unit volume based on a signal intensity threshold set in the sensor 801. β _NT is a positive real number and represents the likelihood that the detection data 512 is a newly generated target. The value of β _NT is, for example, a predetermined constant and the same value as β _FT . When the number of targets to be observed is known from the detection data 511 acquired by the detection data acquisition unit 110, the correlation likelihood calculation unit 133 tracks the number of targets to be observed and the tracking device 100. The configuration may be such that _βNT is calculated based on the number of targets.

The correlation likelihood 533 calculated using this equation is a constant value regardless of the difference between the vertical vector z and the vertical vector z ^* . For example, when the vertical vectors z and z ^* are three-dimensional vectors whose elements are target coordinates in a three-dimensional orthogonal coordinate system, the correlation likelihood calculating unit 133 determines the position of the observed target and the predicted target. Regardless of the distance to the position, a constant correlation likelihood 533 is calculated.

  The track data update unit 134 (correlation unit, update unit, new target determination unit, new track generation unit, unnecessary signal determination unit) uses the processing device 911 to update the track data stored in the track data storage unit 120. For example, in the wake data update unit 134, the target represented by the wake data stored in the wake data storage unit 120 and the target represented by the detection data 512 are the same based on the correlation likelihood 533 calculated by the correlation likelihood calculation unit 133. It is determined whether or not there is. The wake data update unit 134 updates the wake data stored in the wake data storage unit 120 based on detection data representing a target determined to be the same as the target represented by the wake data. The wake data update unit 134 outputs the updated wake group 523 including the updated wake data using the processing device 911. The track data storage unit 120 stores the updated track group 523 output from the track data update unit 134 using the storage device 914.

For example, the track data update unit 134 uses the processing device 911 to correlate a set of track data and the detection data 512 from the track data stored in the track data storage unit 120 for one detection data 512. The track data having the largest correlation likelihood 533 calculated by the likelihood calculation unit 133 is extracted. The track data update unit 134 determines that the target represented by the extracted track data and the target represented by the detection data 512 are the same. However, if the maximum correlation likelihood 533 is smaller than β _FT or β _NT , the wake data update unit 134 determines that the target represented by the wake data and the target represented by the detection data 512 are not the same. When β _FT is the maximum, the track data update unit 134 determines that the detection data 512 is an unnecessary signal. When β _NT is the maximum, the track data update unit 134 determines that the detection data 512 is a newly detected target. When it is determined that the detection data 512 is a newly detected target, the track data update unit 134 uses the processing device 911 to generate new track data based on the detection data 512. For example, the wake data update unit 134 sets the observed value of the target position represented by the detection data 512 to the estimated value of the target position, and generates wake data in which the estimated value of the target speed is set to “0”. To do.

When the maximum correlation likelihood 533 and β _FT are equal, the track data update unit 134 may indicate that the target represented by the extracted track data and the target represented by the detection data 512 are the same, and the detection data 512 is It is determined that the possibility of being an unnecessary signal is approximately the same. If the maximum correlation likelihood 533 is equal to β _NT , the wake data update unit 134 may detect that the target represented by the extracted wake data and the target represented by the detection data 512 are the same, and the detection data. It is determined that the possibility that 512 is a newly detected target is similar.
In such a case, the tracking device 100 parallels two hypotheses, the hypothesis that the target represented by the extracted track data and the target represented by the detection data 512 are the same, and the hypothesis that the two targets are not the same. To consider. In that case, the wake data storage unit 120 uses the wake data updated by the wake data update unit 134 based on the hypothesis that the two targets are the same, and the wake data update unit 134 based on the hypothesis that the two targets are not the same. Two track data are stored together with the track data before updating. The two track data represent the same target, one of which is correct and one of which is incorrect. The tracking device 100 adopts one of the two track data and rejects the other by determining the reliability of the two track data. Even if the difference in reliability between the two track data is small and it is not possible to immediately determine which one is correct, by repeating the update several times, the reliability of the correct track data becomes higher and the wrong one is incorrect. The reliability of wake data becomes low. Therefore, when the difference between the reliability levels of the two track data becomes large, it is determined which is correct, and the track data storage unit 120 deletes the track data determined to be incorrect.

There may be a plurality of detection data 512 determined to represent the same target as the target represented by the track data for one track data. When the separation of the targets is not considered, the detection data 512 representing the same target as the target represented by one track data is observed at the same time (including the case where a time within a predetermined period is regarded as the same time; the same applies hereinafter). There is only one of the detected detection data 512.
Even in such a case, the tracking device 100 examines a plurality of hypotheses in parallel. The track data storage unit 120 stores a plurality of track data in which the track data update unit 134 updates one track data using different detection data 512. Thereafter, the reliability of the plurality of track data is compared to determine correct track data, and the track data storage unit 120 deletes the track data determined to be incorrect.
At this time, the reliability of each track data corresponds to the correlation likelihood 533 in which the target represented by the track data and the target represented by the detection data 512 are the same. That is, if the correlation likelihood 533 is large, the reliability of the updated track data is high, and if the correlation likelihood 533 is small, the reliability of the updated track data is low. For this reason, the track data storage unit 120 may be configured to store only track data with high reliability from the beginning. For example, the wake data storage unit 120 stores only wake data updated by the wake data update unit 134 using the detection data 512 having the largest correlation likelihood 533. Alternatively, the track data storage unit 120 stores only the track data updated by the track data update unit 134 using the detection data 512 having a relatively large correlation likelihood 533, and the detection data 512 having a relatively small correlation likelihood 533 is stored. The track data updated by the track data update unit 134 is not stored.

Further, when there is a plurality of detection data 512 determined to represent the same target as the target represented by the track data for one track data, other than the detection data 512 representing the same target as the target represented by the track data This detection data 512 represents either the same target as the target represented by different track data, an unnecessary signal, or a newly detected target. Therefore, when there is a plurality of detection data 512 determined to represent the same target as the target represented by the track data for one track data, the track data update unit 134 correlates with the track data 533. The configuration may be such that the track likelihood data 533 calculated by the correlation likelihood calculating unit 133 is extracted with the second largest likelihood. In that case, the tracking device 100 has a hypothesis that the target represented by one detection data 512 is the same target as the target represented by the track data having the maximum correlation likelihood 533, and the track having the second largest correlation likelihood 533. Consider the hypothesis that the goal is the same as the goal represented by the data. The track data update unit 134 updates each track data using the same detection data 512. The track data storage unit 120 stores a plurality of track data updated by using one detection data 512.
In that case, at the first time point, the track data having the largest correlation likelihood 533 has higher reliability, but thereafter, the reliability of the track data having the second largest correlation likelihood 533 becomes high, and the reliability is reversed. there is a possibility. For example, the track data storage unit 120 determines that the track data having the second largest correlation likelihood 533 is correct when the reliability is reversed, and uses the detection data 512 for the track data having the largest correlation likelihood 533. Delete the updated track data. Conversely, it is found that the target represented by the track data having the maximum correlation likelihood 533 is the same as the target represented by the other detection data 512, and the track data having the maximum correlation likelihood 533 is determined as the detection data 512. When the track data updated by use is deleted, it is determined that the track data having the second largest correlation likelihood 533 is correct. Further, when it is determined that the target represented by the track data having the maximum correlation likelihood 533 is the same as the target represented by the detection data 512 due to the relationship with the other detection data 512, the correlation likelihood 533 is determined. It is determined that the track data updated using the detection data 512 for the second largest track data is not correct, and the track data storage unit 120 deletes the track data.

  It should be noted that, with respect to one track data 512, whether or not there is a plurality of detection data 512 determined to represent the same target as the target represented by the track data, the correlation likelihood for one track data 512 is determined. A configuration may be adopted in which a plurality of pieces of track data having a relatively high degree 533 are extracted, and one of the targets represented by the plurality of extracted track data is determined to be the same as the target represented by the detection data 512. In that case, the wake data update unit 134 updates each extracted wake data using one detection data 512. When duplication of targets is not considered, there is only one track data representing the same target as the target represented by one detection data 512. Therefore, the tracking device 100 examines a plurality of hypotheses in parallel. The track data storage unit 120 stores a plurality of track data updated by the track data update unit 134 using one detection data 512. Thereafter, the reliability of the track data is compared to determine correct track data, and the track data storage unit 120 deletes the track data determined to be incorrect.

  The track establishment determination unit 135 (reliability determination unit) uses the processing device 911 to correctly track the target for each track data based on the updated track group 523 output from the track data update unit 134. Is almost certain (established). The wake establishment determination unit 135 uses the processing device 911 to output the establishment wake information 534 representing the determined determination result. The wake data storage unit 120 uses the storage device 914 to store the established wake information 534 output from the wake establishment determination unit 135 together with the wake data.

  The wake output unit 190 uses the output device 913 to convert the wake group 521 stored in the wake data storage unit 120 into a format to be output to the outside of the tracking device 100, and as a wake group 522 to the display device 802 or the like. Output.

FIG. 4 is a flowchart showing an example of the flow of the tracking process S600 in this embodiment.
In the tracking process S600, the tracking device 100 tracks the target based on the detection data acquired by the detection data acquisition unit 110. The tracking process S600 includes, for example, a detection data acquisition step S610, a track selection step S631, a gate calculation step S632, a detection data selection step S633, a gate inside / outside determination step S634, a reliability determination step S635, and a first likelihood. A degree calculation step S636, a second likelihood calculation step S637, a wake data update step S660, and a wake establishment determination step S670. The tracking device 100 starts the tracking process S600 from the detection data acquisition step S610.

  In the detection data acquisition step S610, the detection data acquisition unit 110 acquires detection data using the input device 912. When there are a plurality of detection data observed at the same time, the detection data acquisition unit 110 acquires a plurality of detection data observed at the same time.

In the wake selection step S631, the gate calculation unit 131 uses the processing device 911 to select one wake data not yet selected from the wake group 521 stored in the wake data storage unit 120.
If all the track data has been selected and there is no track data to be selected, the gate calculation unit 131 advances the tracking process S600 to the track data update step S660.
If there is wake data that has not been selected, and the wake data is selected, the gate calculation unit 131 advances the tracking process S600 to the gate calculation step S632.

In the gate calculation step S632, the gate calculation unit 131 uses the processing device 911 to detect the detection data acquired by the detection data acquisition unit 110 in the detection data acquisition step S610 for the target represented by the wake data selected in the wake selection step S631. A predicted value of the target state quantity at the observation time is calculated.
The gate calculation unit 131 uses the processing device 911 to calculate a gate centered on the calculated predicted value.

In the detection data selection step S633, the gate inside / outside determination unit 132 uses the processing device 911 to search the gate calculation unit 131 in the track selection step S631 from the detection data acquired by the detection data acquisition unit 110 in the detection data acquisition step S610. For the track data selected by, detection data not yet selected is selected.
When all the detection data has been selected and there is no detection data to be selected, the gate inside / outside determination unit 132 returns the tracking process S600 to the track selection step S631, and the gate calculation unit 131 selects the next track data. .
When there is detection data that has not yet been selected, and the detection data is selected, the gate inside / outside determination unit 132 advances the tracking process S600 to the gate inside / outside determination step S634.

In the gate inside / outside determination step S634, the gate inside / outside determination unit 132 uses the processing device 911 to input the detection data selected in the detection data selection step S633 into the gate calculated by the gate calculation unit 131 in the gate calculation step S632. It is determined whether or not.
When it is determined that the detection data is in the gate, the gate inside / outside determination unit 132 advances the tracking process S600 to the reliability determination step S635.
When it is determined that the detection data is outside the gate, the gate internal / external determination unit 132 returns the tracking process S600 to the detection data selection step S633, and selects the next detection data.

In the reliability determination step S635, the correlation likelihood calculation unit 133 uses the processing device 911 to determine the reliability of the track data selected by the gate calculation unit 131 in the track selection step S631. For example, if the track data is track data for a target that has just started tracking, and there is only one detection data from which the state quantity of the target is estimated, the correlation likelihood calculating unit 133 Determines that the reliability of the track data is low. Conversely, if the track data is not track data for a target that has just started tracking, and there are two or more detection data from which the state quantity of the target is estimated, the correlation likelihood calculating unit 133 It is determined that the reliability of the track data is high.
When it is determined that the reliability of the track data is high, the correlation likelihood calculating unit 133 advances the tracking process S600 to the second likelihood calculating step S637.
When it is determined that the reliability of the track data is low, the correlation likelihood calculating unit 133 advances the tracking process S600 to the first likelihood calculating step S636.

In the first likelihood calculation step S636 (initial correlation likelihood calculation step), the correlation likelihood calculation unit 133 uses the processing device 911 and the target represented by the track data selected by the gate calculation unit 131 in the track selection step S631. In the detection data selection step S633, a correlation likelihood is calculated in which the target represented by the detection data selected by the gate inside / outside determination unit 132 is the same target. Since the reliability of the wake-up data is low, the correlation likelihood calculation unit 133 causes the gate state calculation unit 132 to calculate the predicted value of the target state quantity calculated by the gate calculation unit 131 in the gate calculation step S632 and the detection data selection step S633. Regardless of the difference from the observed value of the target state quantity represented by the selected detection data, a certain correlation likelihood is calculated. For example, the correlation likelihood calculating unit 133 sets the larger value of β _FT and β _NT as the correlation likelihood. The correlation likelihood calculation unit 133 uses the storage device 914 to set the combination of the track data selected by the gate calculation unit 131 in the track selection step S631 and the detection data selected by the gate inside / outside determination unit 132 in the detection data selection step S633. The calculated correlation likelihood is stored in association with.
The correlation likelihood calculation unit 133 returns the tracking process S600 to the detection data selection step S633, and the gate internal / external determination unit 132 selects the next detection data.

In the second likelihood calculation step S637 (updated correlation likelihood calculation step), the correlation likelihood calculation unit 133 uses the processing device 911 and the target represented by the track data selected by the gate calculation unit 131 in the track selection step S631. In the detection data selection step S633, a correlation likelihood is calculated in which the target represented by the detection data selected by the gate inside / outside determination unit 132 is the same target. Since the reliability of the wake-up data is high, the correlation likelihood calculation unit 133 causes the gate state calculation unit 132 to calculate the predicted value of the target state quantity calculated by the gate calculation unit 131 in the gate calculation step S632 and the detection data selection step S633. The smaller the difference from the observed value of the target state quantity represented by the selected detection data, the smaller the correlation likelihood is calculated. For example, the correlation likelihood calculation unit 133 includes a vertical vector z representing an observed value of the target state quantity represented by the detection data selected by the gate inside / outside determination unit 132 in the detection data selection step S633, and a gate calculation unit in the gate calculation step S632. A difference vector (z−z ^* ) from the vertical vector z ^* representing the predicted value of the target state quantity calculated by 131 is calculated. The correlation likelihood calculating unit 133 multiplies the inverse matrix S ⁻¹ of the square matrix S by the calculated difference vector (z−z ^* ) from the right, and transposes the calculated difference vector (z−z ^* ) (z− z ^* ) A real value obtained by multiplying ^T by -1/2 to a real value obtained by multiplying ^T from the left is calculated. The correlation likelihood calculating unit 133 calculates a real value obtained by raising the Napier number e to the calculated real value. Correlation likelihood calculating unit 133, a real value calculated, the product of the target detection probability P _d, divided by the square matrix S by the square root of the determinant of a square matrix 2πS that scalar multiplication by 2 times pi π A real value that is a quotient is calculated, and the calculated real value is used as a correlation likelihood. The correlation likelihood calculation unit 133 uses the storage device 914 to set the combination of the track data selected by the gate calculation unit 131 in the track selection step S631 and the detection data selected by the gate inside / outside determination unit 132 in the detection data selection step S633. The calculated correlation likelihood is stored in association with.
The correlation likelihood calculation unit 133 returns the tracking process S600 to the detection data selection step S633, and the gate internal / external determination unit 132 selects the next detection data.

  In the wake data update step S660, the wake data update unit 134 uses the processing device 911 to set the correlation likelihood calculated by the correlation likelihood calculation unit 133 in the first likelihood calculation step S636 or the second likelihood calculation step S637. Based on this, the track data stored in the track data storage unit 120 is updated. The track data storage unit 120 uses the storage device 914 to store the track data updated by the track data update unit 134.

In the wake establishment determination step S670, the wake establishment determination unit 135 uses the processing device 911 to determine whether the wake data updated by the wake data update unit 134 has been established. The track data storage unit 120 uses the storage device 914 to store the determination result determined by the track establishment determination unit 135 together with the track data updated by the track data update unit 134.
The wake establishment determination unit 135 returns the tracking process S600 to the detection data acquisition step S610, and the detection data acquisition unit 110 acquires the detection data at the next time.

FIG. 5 is a diagram illustrating an example of an observation state of detection data.
In this figure, the observed value of the target state quantity represented by the detection data is expressed two-dimensionally.

The state quantity A1 is an observation value of the target state quantity represented by the detection data at the first observation time among the detection data acquired by the detection data acquisition unit 110. The state quantities A2 and B2 are observation values of the target state quantity represented by the detection data at the second observation time that is the observation time next to the first observation time among the detection data acquired by the detection data acquisition unit 110. is there. The state quantity A3 is an observation value of the target state quantity represented by the detection data at the third observation time that is the observation time next to the second observation time among the detection data acquired by the detection data acquisition unit 110.
It is assumed that the wake data storage unit 120 stores wake data that is generated assuming that the target of the state quantity A1 is a new target that has newly started tracking. The gate G1 is an initial gate calculated by the gate calculation unit 131 based on the track data. The gate G1 is a range centered on the state quantity A1.
The state quantities A2 and B2 are both in the gate G1. Therefore, it cannot be determined from the detection data up to the second observation time whether the wake (A1, A2) is correct or the wake (A1, B2) is correct.
The state quantity A3 is close to the extension of the line connecting the state quantity A1 and the state quantity A2. Therefore, if there is detection data up to the third observation time, it can be determined that the wake (A1, A2) is correct and the wake (A1, B2) is incorrect.

FIG. 6 is a graph showing an example of the correlation likelihood calculated by the correlation likelihood calculation unit 133 in this embodiment.
The horizontal axis is a one-dimensional representation of the observed value of the target state quantity represented by the detection data. The vertical axis represents the correlation likelihood.
A range 701 indicated by a thin line represents the gate G1. A correlation likelihood 711 indicated by a solid line represents a correlation likelihood calculated by the correlation likelihood calculation unit 133. A correlation likelihood 712 indicated by a broken line represents a correlation likelihood calculated by the correlation likelihood calculation unit in the comparative example.

The correlation likelihood calculation unit 133 calculates a constant correlation likelihood 711 regardless of the state quantity difference. Therefore, the difference between the state quantity A2 and the state quantity A1 is larger than the difference between the state quantity B2 and the state quantity A1, but the correlation likelihood calculated by the correlation likelihood calculating unit 133 for the state quantity A2 is the state quantity B2. Is equal to the correlation likelihood calculated for.
Thereby, when the detection data acquisition unit 110 acquires the detection data up to the second observation time, the wake data update unit 134 indicates that the wake (A1, A2) is correct and the wake (A1, B2) is correct. It is determined that the correct possibility is substantially equal, and wake data representing the wake (A1, A2) and wake data representing the wake (A1, B2) are generated, and the wake data storage unit 120 generates the two wake data. Remember.
Thereafter, when the detection data acquisition unit 110 acquires detection data at the third observation time and determines that the wake (A1, A2) is correct and the wake (A1, B2) is incorrect, the wake data storage unit 120 The track data representing the track (A1, B2) is deleted, and only the track data representing the correct track (A1, A2) remains.

On the other hand, the correlation likelihood calculation unit in the comparative example does not distinguish between the initial gate and the other, and calculates the correlation likelihood for the initial gate using the same calculation formula as in other cases. To do. The correlation likelihood calculation unit in the comparative example calculates a smaller correlation likelihood as the state quantity difference is larger. For this reason, the correlation likelihood which the correlation likelihood calculation part in a comparative example calculates about state quantity A2 is smaller than the correlation likelihood calculated about state quantity B2. When the correlation likelihood calculated for the state quantity A2 is smaller than β _FT , the wake data update unit 134 determines that the detection data of the state quantity A2 is an unnecessary signal, and the wake data representing the wake (A1, A2). Is generated, and only wake data representing the wake (A1, B2) is generated. Therefore, even if the detection data at the third observation time is acquired, the correct track (A1, A2) cannot be tracked.

As described above, the tracking device 100 according to this embodiment is configured so that all of the track data for the new target just started to be tracked are included in the gate regardless of the difference between the predicted value of the state quantity and the observed value. A certain correlation likelihood is calculated for the detection data. This prevents detection data having a small difference in state quantity from being erroneously adopted. Further, since the calculated correlation likelihood is not smaller than the likelihood β _FT for determining that the detection data is an unnecessary signal, the correct detection data is prevented from being erroneously rejected. Further, by re-evaluating the validity of the correlation determination in the subsequent detection data, it is possible to correctly interpret the detection data from the same target and start tracking.
Thereby, a target can be tracked correctly.

The tracking device 100 described above determines which one of the detection data obtained from the sensor 801 such as a radar is a target and which is an unnecessary signal, thereby eliminating a false track due to the unnecessary signal. Estimate the track and track the target.
The tracking device 100 includes a detection data acquisition unit 110, a track data storage unit 120, a gate calculation unit 131, a gate inside / outside determination unit 132, a correlation likelihood calculation unit 133, a track data update unit 134, and a track establishment determination. Part 135.
The detection data acquisition unit 110 acquires detection data from the sensor 801 at each time, or acquires detection data information via a network.
The track data storage unit 120 stores track data.
The gate calculation unit 131 calculates a gate based on a predicted track position calculated from the track data.
The gate inside / outside determination unit 132 determines whether the detection data enters the gate.
The correlation likelihood calculation unit 133 calculates the correlation likelihood with the second detection data based on the detection data information acquired by the detection data acquisition unit 110 when the wake is composed only of the first detection data. If the track is also composed of detection data after the initial detection data, a correlation likelihood with the detection data entering the gate is calculated based on the predicted position of the track.
The track data update unit 134 is based on the correlation likelihood between the detection data and the track, the correlation likelihood that the detection data is an unnecessary signal, and the correlation likelihood that the detection data is a new target. In addition, the correlation which is a combination of the detection data and the wake is determined, and the wake data is updated.
The wake establishment determination unit 135 establishes a wake having a high possibility of accurately tracking the target from the wake data updated by the wake data update unit 134.

  When the reliability of the wake data is low, the predicted value of the target state quantity calculated by the gate calculation unit 131 is an average of the possible values, so that the closer the observed value is to the predicted value, the higher the correlation likelihood. Not that. Therefore, it is assumed that target detection data (second detection data) exists uniformly in the gate (initial gate), and a certain correlation likelihood is calculated. Thereby, correct correlation determination can be performed and target tracking can be started reliably.

  The correlation likelihood calculation unit 133 sets a larger value between the correlation likelihood that the detection data is an unnecessary signal and the correlation likelihood that the detection data is a new target as the second detection data. Is set as the correlation likelihood.

  This prevents detection data from being erroneously determined to be an unnecessary signal and rejected or determined to be a new target, performs correct correlation determination, and reliably starts target tracking. Can do.

Embodiment 2. FIG.
The second embodiment will be described with reference to FIG.
In addition, about the part which is common in Embodiment 1, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The configurations of the tracking system 800 and the tracking device 100 in this embodiment are the same as those described in the first embodiment.

The detection data acquisition unit 110 uses the input device 912 to acquire detection data representing an observation value having the target position and the approximate speed as state quantities.
The wake data storage unit 120 (wake storage unit) uses the storage device 914 to store wake data representing estimated values having the target position and speed as state quantities.
The gate calculation unit 131 (prediction unit) uses the processing device 911 and based on the estimated value of the target position and speed represented by the wake-up data, the predicted value of the target position and speed at the observation time of the detection data Is calculated. The gate calculation unit 131 uses the processing device 911 to calculate a gate centered on the predicted value of the calculated target position and speed. For example, based on the assumption that the target speed is constant, the gate calculation unit 131 uses the processing device 911 to calculate an estimated value of the target speed represented by the track data at the target data observation time. The predicted value of.
The correlation likelihood calculation unit 133 (likelihood calculation unit, reliability determination unit) uses the processing device 911 to calculate a correlation likelihood for the detection data that enters the gate calculated by the gate calculation unit 131.

The track data update unit 134 (correlation unit, update unit, new target determination unit, new track generation unit, unnecessary signal determination unit) is based on the correlation likelihood calculated by the correlation likelihood calculation unit 133 using the processing device 911. Thus, it is determined whether the detection data represents the same target as the target represented by any of the wake data, a newly detected target, or an unnecessary signal.
When it is determined that the detection data is a newly detected target, the wake data update unit 134 uses the processing device 911 to generate wake data representing the new target. For example, the wake data update unit 134 generates wake data in which the observed values of the target position and the approximate speed represented by the detection data are estimated values of the target position and speed. The track data storage unit 120 stores the track data generated by the track data update unit 134 using the storage device 914.

When the track data is low in reliability, such as when the track data represents a new target that has just started tracking, the correlation likelihood calculation unit 133 uses the processing device 911 to observe the position of the target represented by the detection data. It is determined whether or not the time change rate of the distance between the value and the predicted value of the target position calculated by the gate calculation unit 131 is equivalent to the approximate speed.
When it is determined that the time change rate of the distance is equivalent to the approximate speed, the gate calculation unit 131 uses the processing device 911 regardless of the distance between the observed target position and the predicted target position. A certain correlation likelihood is calculated.
When it is determined that the time change rate of the distance is not equivalent to the approximate speed, the gate calculation unit 131 uses the processing device 911 regardless of the distance between the observed target position and the predicted target position. The correlation likelihood is calculated using a monotonically decreasing function with the difference between the time change rate of the distance and the approximate speed as a variable.

  For example, the correlation likelihood calculating unit 133 uses the processing device 911 to calculate the vertical vector u represented by the following expression.

u = (x−x ₁ ) / ΔT− (v + v ₁ ) / 2

However, x is a vertical vector and represents the observed value of the target position represented by the detection data. x ₁ is a vertical vector having the same dimension as the vertical vector x, and represents an estimated value of the target position represented by the tracking data. If the goal is a new target that just started tracking, x ₁ represents, for example, the observed value of the position of the target was last observed. ΔT is a positive real number and represents the elapsed time from the time when the target was observed last time to the target observation time indicated by the detection data. v is a vertical vector having the same dimension as the vertical vector x, and represents an observed value of the target velocity represented by the detection data. v ₁ is a vertical vector having the same dimension as the vertical vector x, and represents an estimated value of the target speed represented by the tracking data. If the goal is a new target that just started tracking, v _1, for example, represents the observed value of the speed of the target was last observed.

The correlation likelihood calculating unit 133 calculates a real value g _u represented by the following expression based on the calculated vertical vector u.

g _u = u ^T S _u ⁻¹ u

However, S _u is a square matrix having the same dimension as the vertical vector u.

When the calculated real value g _u is smaller than a predetermined threshold, the correlation likelihood calculating unit 133 calculates a correlation likelihood represented by the following equation.

g = max {β _FT , β _NT }

When the calculated real value g _u is larger than a predetermined threshold value, the correlation likelihood calculating unit 133 calculates a correlation likelihood represented by the following expression.

g = P _d / √ [det (2πS _u )] · exp (−½ · g _u )

FIG. 7 is a flowchart showing an example of the flow of the first likelihood calculation step S636 in this embodiment.
In the first likelihood calculation step S636, the correlation likelihood calculation unit 133 calculates a correlation likelihood when the reliability of the track data is low. The first likelihood calculation step S636 includes, for example, a speed difference calculation step S681, a determination value calculation step S682, and two likelihood calculation steps S683 and S684.

In the speed difference calculation step S681, the correlation likelihood calculation unit 133 calculates the vertical vector u using the processing device 911. For example, the correlation likelihood calculating unit 133 calculates the difference between the vertical vector x representing the observed value of the target position represented by the detection data and the vertical vector x ₁ representing the estimated value of the target position represented by the tracking data for the elapsed time ΔT. A difference vector obtained by subtracting the average of the vertical vector v representing the observed value of the target velocity represented by the detection data and the vertical vector v ₁ representing the estimated value of the target velocity represented by the tracking data from the vertical vector multiplied by the scalar by the reciprocal. The calculated difference vector is defined as a vertical vector u.

In the determination value calculation step S682, the correlation likelihood calculation unit 133 uses the processing device 911 to calculate the value g _u based on the vertical vector u calculated in the speed difference calculation step S681. For example, the correlation likelihood calculation unit 133 calculates a real value obtained by multiplying the inverse matrix S _u ⁻¹ of the square matrix S _u by the vertical vector u from the right and the horizontal vector u ^T obtained by transposing the vertical vector u from the left. Then, the calculated real value is set as a value g _u .
When the calculated value g _u is smaller than a predetermined threshold value, the correlation likelihood calculation unit 133, a first likelihood computing step S636, it proceeds to the likelihood calculation step S683.
When the calculated value _{g u} is greater than a predetermined threshold value, the correlation likelihood calculation unit 133, a first likelihood computing step S636, it proceeds to the likelihood calculation step S684.

In the likelihood calculating step S683, the correlation likelihood calculating unit 133 calculates the correlation likelihood using the processing device 911. For example, the correlation likelihood calculating unit 133 sets the larger value of β _FT and β _NT as the correlation likelihood. The correlation likelihood calculation unit 133 ends the first likelihood calculation step S636.

In the likelihood calculating step S684, the correlation likelihood calculating unit 133 calculates the correlation likelihood using the processing device 911. For example, the correlation likelihood calculation unit 133 calculates the real values multiplied by 1/2 of the value _{g u} calculated by the judgment value calculation step S682. The correlation likelihood calculating unit 133 calculates a real value obtained by raising the Napier number e to the calculated real value. Correlation likelihood calculating unit 133, a real value calculated, the product of the target detection probability P _d, divided by the square matrix S by the square root of the determinant of a square matrix 2πS that scalar multiplication by 2 times pi π A real value that is a quotient is calculated, and the calculated real value is used as a correlation likelihood.

In tracking device 100 in this embodiment,
The correlation likelihood calculation unit 133 sets the correlation likelihood with the second detection data as a certain numerical value if the temporal change rate of the distance between the first detection data and the second detection data is equivalent to the target approximate speed. Otherwise, the correlation likelihood with the second detection data is calculated by a monotonically decreasing function using the difference between the time change rate of the distance and the approximate speed as a variable.

  Thereby, since the correlation likelihood with respect to 2nd detection data is adaptively controlled according to an observation condition, correct correlation determination can be implemented and target tracking can be started reliably.

Embodiment 3 FIG.
A third embodiment will be described.
Note that portions common to Embodiment 1 or Embodiment 2 are denoted by the same reference numerals and description thereof is omitted.

  The configurations of the tracking system 800 and the tracking device 100 in this embodiment are the same as those described in the first embodiment.

The detection data acquisition unit 110 uses the input device 912 to acquire detection data representing an observation value having the target position and the approximate speed as state quantities.
The wake data storage unit 120 (wake storage unit) uses the storage device 914 to store wake data representing estimated values having the target position and speed as state quantities.

The correlation likelihood calculation unit 133 (likelihood calculation unit, reliability determination unit) uses the processing device 911 to determine the reliability of the wake data. The correlation likelihood calculation unit 133 determines that the reliability of the track data is low when the track data represents a new target that has just started tracking and the estimated target speed is greater than a predetermined threshold value. To do.
The correlation likelihood calculation unit 133 calculates the correlation likelihood according to the determination result of the reliability of the wake data. The equation for calculating the correlation likelihood is the same as the equation described in the first embodiment or the second embodiment, for example.

In tracking device 100 in this embodiment,
The correlation likelihood calculating unit 133 sets the correlation likelihood with the second detection data as a certain numerical value if the target approximate speed is equal to or higher than a predetermined threshold, and otherwise, the initial detection data and The correlation likelihood with the second detection data is calculated by a monotonously decreasing function that is a distance variable with the second detection data.

  Thereby, since the correlation likelihood with respect to 2nd detection data is adaptively controlled according to an observation condition, correct correlation determination can be implemented and target tracking can be started reliably.

Embodiment 4 FIG.
A fourth embodiment will be described.
In addition, about the part which is common in Embodiment 1- Embodiment 3, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The configurations of the tracking system 800 and the tracking device 100 in this embodiment are the same as those described in the first embodiment.

The correlation likelihood calculation unit 133 (likelihood calculation unit, reliability determination unit) uses the processing device 911 to determine the reliability of the wake data. When the track data represents a new target that has just started tracking and the distance between the estimated target position and the sensor 801 is larger than a predetermined threshold, the correlation likelihood calculating unit 133 It is determined that the reliability of wake data is low.
The correlation likelihood calculation unit 133 calculates the correlation likelihood according to the determination result of the reliability of the wake data. The equation for calculating the correlation likelihood is the same as the equation described in the first embodiment or the second embodiment, for example.

In tracking device 100 in this embodiment,
The correlation likelihood calculating unit 133 sets the correlation likelihood with the second detection data as a certain numerical value if the distance from the sensor 801 of the first detection data is equal to or greater than a predetermined threshold, and otherwise. Then, the correlation likelihood with the second detection data is calculated by a monotonically decreasing function that is a distance variable between the first detection data and the second detection data.

  Thereby, since the correlation likelihood with respect to 2nd detection data is adaptively controlled according to an observation condition, correct correlation determination can be implemented and target tracking can be started reliably.

Embodiment 5 FIG.
The fifth embodiment will be described.

  In addition, about the part which is common in Embodiment 1- Embodiment 4, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The configurations of the tracking system 800 and the tracking device 100 in this embodiment are the same as those described in the first embodiment.

The correlation likelihood calculation unit 133 (likelihood calculation unit, reliability determination unit) uses the processing device 911 to determine the reliability of the wake data. The correlation likelihood calculating unit 133 represents a new target whose track data has just started tracking, and the elapsed time from the time when the target was last observed to the target observation time indicated by the detection data is a predetermined value. If it is longer than the threshold, it is determined that the reliability of the track data is low.
The correlation likelihood calculation unit 133 calculates the correlation likelihood according to the determination result of the reliability of the wake data. The equation for calculating the correlation likelihood is the same as the equation described in the first embodiment or the second embodiment, for example.

In tracking device 100 in this embodiment,
If the observation time difference between the first detection data and the second detection data is equal to or greater than a predetermined threshold, the correlation likelihood calculation unit 133 sets the correlation likelihood with the second detection data as a certain numerical value. Otherwise, the correlation likelihood with the second detection data is calculated with a monotonically decreasing function as a distance variable between the first detection data and the second detection data.

  Thereby, since the correlation likelihood with respect to 2nd detection data is adaptively controlled according to an observation condition, correct correlation determination can be implemented and target tracking can be started reliably.

  Note that the correlation likelihood calculating unit 133 may be configured to determine the reliability of the wake data based on a determination criterion that is a combination of the determination criteria described in the first to fifth embodiments.

  The correlation likelihood calculating unit 133, when the wake is composed only of first detection data, includes a target approximate speed, a sensor of the first detection data among the detection data information acquired by the detection data acquisition unit 110. The correlation likelihood with the second detection data is set or calculated based on some of the distance from the observation time and the observation time difference between the first detection data and the second detection data.

Embodiment 6 FIG.
The sixth embodiment will be described with reference to FIGS.
In addition, about the part which is common in Embodiment 1- Embodiment 5, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The configurations of the tracking system 800 and the tracking device 100 in this embodiment are the same as those described in the first embodiment.

FIG. 8 is a detailed block diagram showing an example of the configuration of the wake data updating unit 134 in this embodiment.
The track data update unit 134 includes, for example, a correlation unit 141, an update unit 142, a new target selection unit 143, and a new track generation unit 144.

The correlation unit 141 uses the processing device 911 to determine whether the target represented by the detection data and the target represented by the tracking data are the same target based on the correlation likelihood calculated by the correlation likelihood calculation unit 133. Determine.
The update unit 142 uses the processing device 911 to update the track data stored in the track data storage unit 120 based on the determination result of the correlation unit 141.
The new target selection unit 143 (new target determination unit, reliability determination unit) is acquired by the detection data acquisition unit 110 based on the correlation likelihood calculated by the correlation likelihood calculation unit 133 using the processing device 911. It is determined whether or not the detection data is a newly detected target.
The new wake generation unit 144 uses the processing device 911 to generate new wake data based on the determination result of the new target selection unit 143. The track data storage unit 120 uses the storage device 914 to store the track data generated by the new track generation unit 144.

The new target selection unit 143 uses the processing device 911 to determine the reliability (accuracy) of the wake data. Alternatively, the new target selection unit 143 may be configured to acquire and use the reliability of the wake data determined by the correlation likelihood calculation unit 133. On the contrary, the correlation likelihood calculation unit 133 may acquire and use the reliability of the wake data determined by the new target selection unit 143.
When the new target selection unit 143 uses the processing device 911 to determine that the reliability of the track data is high, the target represented by the track data and the target represented by the detection data have the same correlation likelihood. Or low. For example, the new target selection unit 143 compares the correlation likelihood calculated by the correlation likelihood calculation unit 133 with a predetermined threshold. When the correlation likelihood is larger than the threshold value, the new target selection unit 143 determines that the correlation likelihood is high. When the correlation likelihood is smaller than the threshold value, the new target selection unit 143 determines that the correlation likelihood is low.
When it is determined that the correlation likelihood is high, the new target selection unit 143 uses the processing device 911 to determine that the target represented by the detection data is not a newly detected target. When there is track data determined to have a high correlation likelihood, the new target selection unit 143 determines that the detection data does not represent a newly detected target.
The new target selection unit 143 uses the processing device 911 to determine that the detection data may represent a newly detected target when there is no wake data determined to have a high correlation likelihood. . The new target selection unit 143 further determines whether or not the detected data represents a newly detected target for the detected data determined to have a possibility of representing a newly detected target.

  A track that can be determined to have a high possibility of tracking a target is likely to be at most one of the detection data included in the gate of the track. At this time, if the detection data with a high correlation likelihood with the wake does not consider that the target is separated on the way, the detection data is a target being tracked by the wake or an unnecessary signal, Not a new goal. Therefore, by determining that some of the detected data is not the new target, the possibility of useless correlation different from the correct answer is eliminated.

FIG. 9 is a flowchart showing an example of the flow of the tracking process S600 in this embodiment.
In addition to the steps described in the first embodiment, the tracking processing S600 further includes an initialization step S620, a reliability determination step S641, a likelihood determination step S642, and a determination storage step S643.

  After the end of the detection data acquisition step S610, the detection data acquisition unit 110 advances the tracking process S600 to the initialization step S620.

In the initialization step S620, the new target selection unit 143 uses the storage device 914 to detect the target data newly detected by the detection data for each detection data acquired by the detection data acquisition unit 110 in the detection data acquisition step S610. Is stored (data indicating “possibility of a new goal”).
The new target selection unit 143 advances the tracking process S600 to the wake selection process S631.

  After the completion of the first likelihood calculation step S636 or the completion of the second likelihood calculation step S637, the correlation likelihood calculation unit 133 advances the tracking process S600 to the reliability determination step S641.

In the reliability determination step S641, the new target selection unit 143 uses the processing device 911 to determine whether the reliability of the track data selected by the gate calculation unit 131 in the track selection step S631 is high or low.
When it determines with the reliability of track data being high, the new target selection part 143 advances tracking process S600 to likelihood determination process S642.
When it is determined that the reliability of the track data is low, the new target selection unit 143 returns the tracking process S600 to the detection data selection step S633, and the gate internal / external determination unit 132 selects the next detection data.

In the likelihood determination step S642, the new target selection unit 143 uses the processing device 911 to calculate the correlation likelihood calculated by the correlation likelihood calculation unit 133 in the first likelihood calculation step S636 or the second likelihood calculation step S637. Determine whether it is large or small.
When it is determined that the correlation likelihood is large, the new target selection unit 143 advances the tracking process S600 to the determination storage process S643.
When it is determined that the correlation likelihood is small, the new target selection unit 143 returns the tracking process S600 to the detection data selection step S633, and the gate internal / external determination unit 132 selects the next detection data.

In the determination storage step S643, the new target selection unit 143 rewrites the data about the detection data selected by the gate inside / outside determination unit 132 in the detection data selection step S633 among the data stored in the initialization step S620, and detects the detection data. Stores data indicating that it does not represent a newly detected target (“no new target possibility” data).
The new target selection unit 143 returns the tracking process S600 to the detection data selection step S633, and the gate inside / outside determination unit 132 selects the next detection data.

  In the track data update step S660, the new target selection unit 143 uses the processing device 911 to store the detection data for each detection data acquired by the detection data acquisition unit 110 in the detection data acquisition step S610. It is determined whether the “possible new target” data stored in the initialization step S620 is retained or has been rewritten to the “no possible new target” data stored in the determination storing step S643. When the data has been rewritten with “no new target possibility” data, the new target selection unit 143 uses the processing device 911 to determine that the detected data is not a newly detected target. When the “possible new target” data remains, the new target selection unit 143 uses the processing device 911 to determine whether or not the detected data represents a newly detected target.

  The criterion for determining whether the reliability is high or low in the reliability determination step S641 may be the same as or different from the criterion for determining whether the reliability is high or low in the reliability determination step S635. Also good. When the reliability criterion is the same, the reliability determination step S641 may not be performed. In that case, after the end of the first likelihood calculation step S636, the correlation likelihood calculation unit 133 returns the tracking processing S600 to the detection data selection step S633. In addition, after the second likelihood calculation step S637 ends, the correlation likelihood calculation unit 133 advances the tracking process S600 to the likelihood determination step S642.

  The reliability determination criteria in the reliability determination step S641 include, for example, the following determination criteria in addition to the determination criteria in the correlation likelihood calculation unit 133 described in the first to fifth embodiments.

  When the track data storage unit 120 stores track data based on a plurality of hypotheses, the track data based on the hypothesis having the highest reliability at that time is determined to have high reliability, and other track data Is determined to have low reliability.

  When the number of detection data constituting the track data (the number of times the update unit 142 has updated the track data using the detection data) is larger than a predetermined threshold (for example, 3), it is determined that the reliability is high, and the detection is performed. When the number of data (number of updates) is smaller than a predetermined threshold, it is determined that the reliability is low.

  The wake data determined by the wake establishment determination unit 135 is determined to have high reliability, and the other wake data is determined to have low reliability.

  Note that the correlation likelihood calculating unit 133 may determine whether the reliability of the track data is high or low based on these determination criteria, and calculate the correlation likelihood based on the determined result. Good.

  In addition, as a criterion for determining whether the correlation likelihood is large or small in the likelihood determination step S642, there are the following determination criteria in addition to the determination criterion for determining whether the correlation likelihood is larger or smaller than a predetermined threshold.

  When the correlation likelihood is the maximum among the detection data for which the correlation likelihood is calculated for one track data, the target represented by the track data and the target represented by the detection data are the same target. It is determined that the correlation likelihood is high, and in other cases, it is determined that the correlation likelihood is low.

  For one track data, when the correlation likelihood is calculated and arranged in the descending order of correlation likelihood in the detection data in which the correlation likelihood is calculated, the target represented by the track data when it falls within the top Nth (for example, second) And the target represented by the detection data are determined to have a high correlation likelihood, and otherwise, the correlation likelihood is determined to be low.

  FIG. 10 is a diagram illustrating an example of an observation state of detection data.

The detection data A, B, C, and D are detection data observed at the same time.
The wake T1 has five detection data. The gate calculation unit 131 calculates the gate of the wake T1. The gate inside / outside determination unit 132 determines that three detection data A, B, and C are contained in the gate. The correlation likelihood calculation unit 133 sets the correlation likelihood between the wake T1 and the detection data A to “1.0”, sets the correlation likelihood between the wake T1 and the detection data B to “0.5”, and sets the correlation likelihood between the wake T1 and the detection data C. Is calculated as “0.2”.
The wake T2 has two detection data. The gate calculation unit 131 calculates the gate of the wake T2. The gate inside / outside determination unit 132 determines that three detection data B, C, and D are included in the gate. The correlation likelihood calculation unit 133 sets the correlation likelihood between the wake T2 and the detection data B to “0.3”, sets the correlation likelihood between the wake T2 and the detection data C to “0.1”, and sets the wake T2 and the detection data D. Is calculated as “0.5”.

In such a situation, if the reliability determination criterion of the wake data is “determined that the reliability is high when the number of detected data is 3 or more”, the new target selection unit 143 has a high reliability of the wake T1. And the reliability of the wake T2 is determined to be low.
Also, if the correlation likelihood magnitude determination criterion is “determined that the correlation likelihood is large up to the second highest rank”, the new target selection unit 143 sets the two detection data A and B for the wake T1. It is determined that the correlation likelihood is large, and it is determined that the correlation likelihood of the two detection data B and D is large for the wake T2.
In this case, the new target selection unit 143 determines that the two detection data A and B determined to have a high correlation likelihood for the track T1 determined to have high reliability are “no possibility of new target”, and The other two detection data C and D are determined as “possible new target”.

  In this way, by determining whether or not the detection data may be a new target, the processing required to determine whether or not the detection data is a new target is reduced, and the time taken for the determination Can be shortened. This also reduces the number of hypotheses, so that the processing required to determine which hypothesis is correct can be reduced and the time required for the determination can be shortened.

The tracking device 100 in this embodiment has a new target selection unit 143.
The new target selection unit 143 selects a track with high accuracy tracking the target from the tracks stored in the track data storage unit, and is a new target among the detection data that enters the gate of the selected track. Select detection data that eliminates the possibility.
The track data update unit 134 determines the correlation based on the result of the new target selection unit selecting the detection data.

  The new target selection unit 143 selects, as the highly accurate track, a track included in the hypothesis having the highest reliability among the hypotheses that have the possibility of correlation.

  The new target selection unit 143 selects a track whose number of detection data to be configured is equal to or greater than a predetermined threshold as the highly accurate track.

  The new target selection unit 143 selects a track that has been determined to be an established track by the track establishment determination unit 135 as the highly accurate track.

  The new target selection unit 143 selects the detection data having the maximum correlation likelihood as the detection data that excludes the possibility of the new target.

  The new target selection unit 143 selects, as detection data for eliminating the possibility of being the new target, detection data included up to the Nth highest correlation likelihood for a predetermined natural number N.

  As described above, by not considering some of the detection data included in the gate of the track with high accuracy as the new target, the possibility of useless correlation different from the correct answer is eliminated. Thereby, the processing load of the tracking device 100 for performing correct correlation determination can be reduced.

  In addition, when it can be determined from the detection data acquired by the detection data acquisition unit 110 whether the target has been separated in the middle, the tracking device 100 assumes that the target has been separated in the middle based on the determination result. It may be a configuration. For example, if the observation situation is that the target is separated, the detection data included in the high-accuracy wake gate is regarded as a tracked target or a new target, and not an unnecessary signal. Such a configuration may be adopted.

  The track data update unit 134 selects some of the tracks stored in the track data storage unit 120, and some of the detection data that enters the gate of the selected track may be a tracked target. The correlation is determined after eliminating either the possibility of an unnecessary signal or the possibility of a new target.

Embodiment 7 FIG.
The seventh embodiment will be described with reference to FIGS.
In addition, about the part which is common in Embodiment 1- Embodiment 6, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The configuration of tracking system 800 in this embodiment is the same as that described in the first embodiment.

FIG. 11 is a block diagram showing an example of the configuration of the tracking device 100 according to this embodiment.
The tracking device 100 further includes a track priority setting unit 136 and a gate area storage unit 137 in addition to the configuration described in the first embodiment.

  The wake priority setting unit 136 (reliability calculation unit) uses the processing device 911 and based on the wake group 521 stored in the wake data storage unit 120, the priority (reliability) of each wake data included in the wake group 521. Degree). The wake priority setting unit 136 outputs the calculated wake priority 535 using the processing device 911.

  The gate calculation unit 131 (prediction unit) uses the processing device 911, based on the track priority 535 calculated by the track priority setting unit 136, the track data included in the track group 521 stored in the track data storage unit 120. Are calculated in descending order of the track priority 535.

  The gate area storage unit 137 stores the gate calculated by the gate calculation unit 131 using the storage device 914.

  The gate inside / outside determination unit 132 uses the processing device 911 to determine whether or not the detection data 512 is in the gate based on the gate calculated by the gate calculation unit 131. However, based on the gate stored in the gate area storage unit 137, the gate inside / outside determination unit 132 is calculated by the gate calculation unit 131 for the track data whose priority calculated by the track priority setting unit 136 is higher than the track data. The detection data 512 that has entered the gate is determined not to have entered the gate. That is, the gate inside / outside determination unit 132 determines that the detection data determined to be in the gate of a certain track data is not in the track data gate having a lower priority than the track data. Note that the gate inside / outside determination unit 132 may determine that one piece of detection data is in the gate of a plurality of track data having the same priority.

  For a wake that cannot be determined to track the target, such as a wake composed only of the first detection data, the target detection data can be reliably captured in the gate and the target tracking can be started by expanding the gate. However, when the gate is enlarged, detection data of a wake that is likely to track the target is also included in the gate. The detection data included in the track of a track that is likely to track the target is likely to be the target being tracked by that track, and therefore the correlation with the track that cannot be determined to track the target. Thinking about possibilities is useless. Therefore, the possibility of useless correlation different from the correct answer is eliminated by executing the gate inside / outside determination processing in order of priority.

FIG. 12 is a flowchart showing an example of the flow of the tracking process S600 in this embodiment.
In addition to the process described in the first embodiment, the tracking process S600 is further provided with a track priority setting process S615, an initialization process S625, a priority determination process S651, an area update process S652, and an area storage process S653. And have.

  After the end of the detection data acquisition step S610, the detection data acquisition unit 110 advances the tracking process S600 to the track priority setting step S615.

  In the track priority setting step S615, the track priority setting unit 136 uses the processing device 911 to calculate the priority of the track data for each track data stored in the track data storage unit 120. The wake priority setting unit 136 stores the calculated priority using the storage device 914.

In the initialization step S625, the gate region storage unit 137 uses the processing device 911 to initialize the set region that has been set as the gate region and the exclusion region that is excluded from the gate region, and uses the storage device 914 to The fact that there are no set areas and excluded areas is stored.
The gate area storage unit 137 advances the tracking process S600 to the wake selection process S631.

In the wake selection step S631, the gate calculation unit 131 uses the processing device 911 to store the wake stored in the wake data storage unit 120 based on the priority calculated by the wake priority setting unit 136 in the wake priority setting step S615. Among the track data that have not yet been selected, the track data with the highest priority is selected.
If all the track data has been selected and there is no track data to be selected, the gate calculation unit 131 advances the tracking process S600 to the track data update step S660.
When there is track data that has not been selected yet, the gate calculation unit 131 selects one track data with the highest priority from the track data that has not been selected yet. When there are a plurality of wake data having the highest priority, the gate calculation unit 131 selects one wake data from them. The gate calculation unit 131 advances the tracking process S600 to the priority determination step S651.

In the priority determination step S651, the gate area storage unit 137 uses the processing device 911 to perform the track priority setting unit 136 in the track priority setting step S615 on the track data selected by the gate calculation unit 131 in the track selection step S631. It is determined whether or not the calculated priority is lower than the priority calculated by the track priority setting unit 136 in the track priority setting step S615 for the track data selected by the gate calculation unit 131 in the previous track selection step S631. .
When it is determined that the priority is low, the gate area storage unit 137 advances the tracking process S600 to the area update process S652.
When it is determined that the priorities do not change and are the same, the gate area storage unit 137 advances the tracking process S600 to the gate calculation process S632.

In the area update step S652, the gate area storage unit 137 uses the storage device 914 to update the excluded area. The gate area storage unit 137 stores the set area stored in the area storage step S653 described later as an excluded area.
The gate region storage unit 137 advances the tracking process S600 to the gate calculation step S632.

  After the completion of the gate calculation step S632, the gate calculation unit 131 advances the tracking process S600 to the area storage step S653.

  In the area storage step S653, the gate area storage unit 137 uses the storage device 914 to add the area included in the gate calculated by the gate calculation unit 131 in the gate calculation step S632 to the stored set area. Remember. The gate area storage unit 137 advances the tracking process S600 to the detection data selection step S633.

  In the gate inside / outside determination step S634, the gate inside / outside determination unit 132 uses the processing device 911 to input the detection data selected in the detection data selection step S633 into the gate calculated by the gate calculation unit 131 in the gate calculation step S632. It is determined whether or not. However, when the detection data is included in the exclusion region stored in the gate region storage unit 137 in the region update step S652, the gate inside / outside determination unit 132 determines that the detection data is not in the gate.

In addition, the criterion for determining whether the priority in the wake priority setting step S615 is high or low may be the same as the reliability determination criterion described in the first to sixth embodiments, for example. May be different.
For example, there are the following criteria for determining the priority in the wake priority setting step S615.

  In the case where the track data storage unit 120 stores track data based on a plurality of hypotheses, the higher the reliability of the hypothesis, the higher the priority of the track data based on the hypothesis, according to the reliability of the hypothesis at that time. The lower the reliability of the hypothesis, the lower the priority is calculated for the track data based on the hypothesis. When one track data is included in a plurality of hypotheses, for example, the priority is calculated according to the hypothesis having the highest reliability among the plurality of hypotheses.

  In accordance with the number of detection data constituting the track data (the number of times the track data update unit 134 has updated the track data using the detection data), the higher the number of detection data, the higher the priority is given to the track data. The lower the priority of the track data, the lower the number of detection data.

  A high priority is calculated for the wake data determined by the wake establishment determination unit 135.

  A low priority is calculated for track data representing a new target that has just started tracking.

  In accordance with the number of times the target is interpreted not to be detected in the wake, a high priority is calculated for wake data with a small number of times, and a low priority is calculated for wake data with a large number of times. For example, if the observation period of the sensor 801 is one period and the detection data that the observation time falls within that period does not include the detection data used by the wake data update unit 134 to update the wake data, the wake priority setting is performed. The unit 136 determines that the target has not been detected during the period. The wake priority setting unit 136 counts the number of periods in which it is determined that the target has not been detected, and sets the number of times that the target has not been detected.

Note that the correlation likelihood calculating unit 133 may determine whether the reliability of the track data is high or low based on these determination criteria, and calculate the correlation likelihood based on the determined result. Good.
Further, the new target selection unit 143 described in the sixth embodiment determines whether the reliability of the track data is high or low based on these determination criteria, and the possibility of a new target based on the determined result. It may be configured to determine whether or not there is.

FIG. 13 is a diagram illustrating an example of an observation state of detection data.
The detection data A, B, C, and D are detection data observed at the same time.
The wake T1 has five detection data. The wake T2 has two detection data.

If the priority determination criterion of the wake data is “the higher the number of detection data, the higher the priority”, the wake priority setting unit 136 determines that the wake T1 has a higher priority than the wake T2.
The gate calculation unit 131 calculates the gate of the wake T1. The gate inside / outside determination unit 132 determines that three detection data A, B, and C are contained in the gate.
The gate calculation unit 131 calculates the gate of the wake T2. Since the gate overlaps with the gate of the wake T1 having a higher priority than the wake T2, the inside / outside determination unit 132 detects the detection data in the area (the hatched area) excluding the wake T1 gate from the gate. D is determined to be in the gate, and the detection data B and C are determined not to be in the gate.

The tracking device 100 in this embodiment includes a track priority setting unit 136 and a gate area storage unit 137.
The gate area storage unit 137 stores a gate setting area.
The track priority setting unit 136 sets a priority for a track stored in the track data storage unit 120.
The gate calculation unit 131 calculates wake gates in descending order of priority within the range of the gate setting region stored in the gate region storage unit 137, and stores the gate in the gate region storage unit 137. Delete the gate area of the track with the same priority from the gate setting area.

  The track priority setting unit 136 sets the track priority in descending order of hypothesis reliability among the hypotheses that are the possibility of correlation as the priority.

  The track priority setting unit 136 sets the track priority as the priority in descending order of the number of detection data constituting the track.

  The wake priority setting unit 136 includes a wake group determined by the wake establishment determination unit 135 as an established wake in order of priority, and a wake group formed of two or more detection data that are not established wakes. The wake priority is set so that the wake group is composed only of the first detection data.

  The wake priority setting unit 136 sets the wake priority as the priority in ascending order of the number of times that the target is not detected in the wake.

  In this way, by executing the gate inside / outside determination process with priority on the highly accurate track, the possibility of correlation of the track with low accuracy is limited, and the possibility of useless correlation different from the correct answer is eliminated. Thereby, the processing load of the tracking device for performing correct correlation determination can be reduced.

  The configurations described in the above embodiments are examples, and the tracking device 100 may have a different configuration. For example, the structure which combined the structure demonstrated in different embodiment may be sufficient, and the structure which replaced the structure of the non-essential part with the other structure may be sufficient.

The tracking device (100) described above includes a detection data acquisition unit (110), a track storage unit (track data storage unit 120), a prediction unit (gate calculation unit 131), and a likelihood calculation unit (correlation likelihood calculation). 133), a correlation unit (141; wake data update unit 134), an update unit (142; wake data update unit 134), and a reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake A priority setting unit 136; a new target selection unit 143).
The detection data acquisition unit acquires detection data representing an observed value of the state quantity of the target and an observation time when the sensor observed the target for a target estimated to be observed by the sensor (801).
The track storage unit stores track data representing an estimated value of the state quantity of the target for the target estimated to be observed by the sensor.
The prediction unit determines whether the detection data is based on an estimated value of a target state quantity represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. A predicted value of the state quantity of the target represented by the track data at the observation time of the represented target is calculated.
The likelihood calculation unit includes a predicted value of the target state quantity calculated by the prediction unit and the detection data for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. The likelihood that the target represented by the track data and the target represented by the detection data are the same is calculated based on the observed state value of the target represented by.
The correlation unit is a target represented by the track data based on the likelihood calculated by the likelihood calculation unit for a set of the track data stored by the track storage unit and the detection data acquired by the detection data acquisition unit. And whether the target represented by the detection data is the same.
The update unit is configured such that for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit, a target represented by the track data and a target represented by the detection data are the same. When the correlation unit determines, the wake data is updated based on the observed value of the target state quantity represented by the detection data.
The reliability determination unit determines whether the reliability of the track data is high or low with respect to the track data stored in the track storage unit.
The likelihood calculation unit predicts a target state quantity calculated by the prediction unit for a set of track data determined by the reliability determination unit and detection data acquired by the detection data acquisition unit when reliability is high. The larger the difference between the value and the observed value of the target state quantity represented by the detection data, the smaller the likelihood is calculated, and when the reliability is low, the track data determined by the reliability determination unit and the detection data acquisition unit For a set with the acquired detection data, a constant likelihood is calculated regardless of the difference between the predicted value of the target state quantity calculated by the prediction unit and the observed value of the target state quantity represented by the detection data.

The track data stored in the track storage unit (track data storage unit 120) represents an estimated value of the state quantity including the target position and speed.
The detection data acquired by the detection data acquisition unit (110) represents the observed value of the state quantity including the target position.
The prediction unit (gate calculation unit 131) is an estimated value of a target position and speed represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. Based on the above, a predicted value of the target position represented by the track data at the target observation time represented by the detection data is calculated.
The update unit (142; track data update unit 134) includes a target represented by the track data and the detection data for the set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. When the correlation unit (141; track data update unit 134) determines that the target to be represented is the same, the estimated value of the target position and speed represented by the track data, and the target position represented by the detection data The wake data is updated based on the observed values.
The likelihood calculation unit (correlation likelihood calculation unit 133) includes track data determined by the reliability determination unit (correlation likelihood calculation unit 133) when the reliability is high, and detection data acquired by the detection data acquisition unit. For the set of the above, the larger the difference between the predicted value of the target position predicted by the prediction unit and the observed value of the target position represented by the detection data, the smaller likelihood is calculated. For the set of track data determined by the degree determination unit and the detection data acquired by the detection data acquisition unit, the predicted value of the target position calculated by the prediction unit, and the observed value of the target position represented by the detection data, Regardless of the difference, a certain likelihood is calculated.

The track data stored in the track storage unit (track data storage unit 120) represents an estimated value of the state quantity including the target position and speed.
The detection data acquired by the detection data acquisition unit (110) represents the observed value of the state quantity including the target position and speed.
The prediction unit (gate calculation unit 131) is an estimated value of a target position and speed represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. Based on the above, a predicted value of the target position represented by the track data at the target observation time represented by the detection data is calculated.
The update unit (142; track data update unit 134) includes a target represented by the track data and the detection data for the set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. When the correlation unit (141; track data update unit 134) determines that the target to be represented is the same, the estimated value of the target position and speed represented by the track data, and the target position represented by the detection data The wake data is updated based on the observed values of the speed and the speed.
The likelihood calculation unit (correlation likelihood calculation unit 133) includes track data determined by the reliability determination unit (correlation likelihood calculation unit 133) when the reliability is high, and detection data acquired by the detection data acquisition unit. For the set, the smaller likelihood is calculated as the difference between the predicted value of the target position calculated by the prediction unit and the observed value of the target position represented by the detection data is larger, and the reliability is lower when the reliability is lower. Based on the estimated value of the target position represented by the track data and the observed value of the target position represented by the detection data for the set of the track data determined by the degree determination unit and the detection data acquired by the detection data acquisition unit The target speed is calculated, and the smaller likelihood is calculated as the difference between the calculated speed and the observed value of the target speed represented by the detection data is larger.

The tracking device 100 includes a detection data acquisition unit (110), a track storage unit (track data storage unit 120), a prediction unit (gate calculation unit 131), a likelihood calculation unit (correlation likelihood calculation unit 133), Correlation unit (141; track data update unit 134), update unit (142; track data update unit 134), new target determination unit (new target selection unit 143; track data update unit 134), and new track generation unit ( 144; wake data update unit 134) and reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143).
The detection data acquisition unit acquires detection data representing an observed value of the state quantity of the target and an observation time when the sensor observed the target for a target estimated to be observed by the sensor (801).
The track storage unit stores track data representing an estimated value of the state quantity of the target for the target estimated to be observed by the sensor.
The prediction unit determines whether the detection data is based on an estimated value of a target state quantity represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. A predicted value of the state quantity of the target represented by the track data at the observation time of the represented target is calculated.
The likelihood calculation unit includes a predicted value of the target state quantity calculated by the prediction unit and the detection data for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. The likelihood that the target represented by the track data and the target represented by the detection data are the same is calculated based on the observed state value of the target represented by.
The correlation unit is a target represented by the track data based on the likelihood calculated by the likelihood calculation unit for a set of the track data stored by the track storage unit and the detection data acquired by the detection data acquisition unit. And whether the target represented by the detection data is the same.
The update unit is configured such that for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit, a target represented by the track data and a target represented by the detection data are the same. When the correlation unit determines, the wake data is updated based on the observed value of the target state quantity represented by the detection data.
The new target determination unit determines whether or not the target represented by the detection data is a new target for the detection data acquired by the detection data acquisition unit.
The new track generation unit, when the new target determination unit determines that the target represented by the detection data is a new target for the detection data acquired by the detection data acquisition unit, the target of the target represented by the detection data Wake data representing a new target is generated based on the observed value of the state quantity.
The track storage unit stores track data generated by the new track generation unit.
The reliability determination unit determines whether the reliability of the track data is high or low with respect to the track data stored in the track storage unit.
Whether the new target determination unit has a high likelihood calculated by the likelihood calculation unit for a set of the track data determined by the reliability determination unit and the detection data acquired by the detection data acquisition unit when the reliability is high. If it is determined whether or not the likelihood is high, it is determined that the target represented by the detection data is not a new target.

  The new target determination unit (new target selection unit 143) includes the track data determined by the reliability determination unit (new target selection unit 143) and the detection data acquisition unit (detection data acquisition unit 110) when the reliability is high. The likelihood calculated by the likelihood calculation unit (correlation likelihood calculation unit 133) for the set of acquired detection data is the likelihood of the set of the track data and other detection data acquired by the detection data acquisition unit. In the likelihood calculated by the degree calculation unit, the likelihood is determined to be high when entering from a higher one to a predetermined rank.

The tracking device 100 includes a detection data acquisition unit (110), a track storage unit (track data storage unit 120), a prediction unit (gate calculation unit 131), a likelihood calculation unit (correlation likelihood calculation unit 133), Correlation unit (141; wake data update unit 134), update unit (142; wake data update unit 134), reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136) A new target selection unit 143).
The detection data acquisition unit acquires detection data representing an observed value of the state quantity of the target and an observation time when the sensor observed the target for a target estimated to be observed by the sensor (801).
The track storage unit stores track data representing an estimated value of the state quantity of the target for the target estimated to be observed by the sensor.
The prediction unit determines whether the detection data is based on an estimated value of a target state quantity represented by the track data for a set of track data stored by the track storage unit and detection data acquired by the detection data acquisition unit. A predicted value of the state quantity of the target represented by the track data at the observation time of the represented target is calculated.
The likelihood calculation unit includes a predicted value of the target state quantity calculated by the prediction unit and the detection data for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. The likelihood that the target represented by the track data and the target represented by the detection data are the same is calculated based on the observed state value of the target represented by.
The correlation unit is a target represented by the track data based on the likelihood calculated by the likelihood calculation unit for a set of the track data stored by the track storage unit and the detection data acquired by the detection data acquisition unit. And whether the target represented by the detection data is the same.
The update unit is configured such that for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit, a target represented by the track data and a target represented by the detection data are the same. When the correlation unit determines, the wake data is updated based on the observed value of the target state quantity represented by the detection data.
The reliability determination unit determines whether the reliability of the track data is high or low with respect to the track data stored in the track storage unit.
The likelihood calculation unit includes a predicted value of the target state quantity calculated by the prediction unit and the detection data for a set of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit. When the difference from the observed value of the target state quantity represented by is within a predetermined range, the likelihood is calculated, and the track data determined by the reliability determination unit when the reliability is lower than the track data; The likelihood is not calculated for the pair with the detection data.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) is stored in the wake data stored in the wake storage unit (wake data storage unit 120). When the update unit (142; track data update unit 134) has updated the track data, it is determined that the reliability of the track data is high, and the update unit has updated the track data. If there is no data, it is determined that the reliability of the track data is low.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) is updated by the update unit (142; wake data update unit 134). If the speed of the target represented by the track data is fast or slow, and if it is determined that the target speed represented by the track data is fast, the reliability of the track data is low. Is determined.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) is stored in the wake data stored in the wake storage unit (wake data storage unit 120). When the update unit (142; track data update unit 134) has not updated the track data, whether the distance between the target represented by the track data and the sensor (801) is long or short. When it is determined that the distance is long, it is determined that the reliability of the wake data is low.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) is stored in the wake data stored in the wake storage unit (wake data storage unit 120). When the update unit (142; track data update unit 134) has updated the track data, the update unit detects the detection from the target observation time indicated by the detection data used for updating the track data. If the data acquisition unit (110) determines whether the elapsed time up to the target observation time represented by the detection data acquired is long or short, and determines that the elapsed time is long, the reliability of the track data is low judge.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) is stored in the wake data stored in the wake storage unit (wake data storage unit 120). When the update unit (142; track data update unit 134) determines whether the update number of the track data is large or small, and determines that the update number is large, the reliability of the track data is If it is determined that the number of updates is small, it is determined that the reliability of the track data is low.

  The reliability determination unit (correlation likelihood calculation unit 133; wake establishment determination unit 135; wake priority setting unit 136; new target selection unit 143) includes the wake storage unit (wake data storage unit 120) for each of a plurality of periods. ) Is stored using the detection data in which the target observation time falls within the period, and the update unit (142; wake data update unit 134) determines whether or not the track data is updated. When the update unit determines whether the number of periods not updated is large or small, and when the update unit determines that the number of periods not updated is small, it is determined that the reliability of the wake data is high, When it is determined that the number of periods in which the update unit has not been updated is large, it is determined that the reliability of the track data is low.

According to the tracking device 100 described above, correct correlation determination can be performed and target tracking can be reliably started by adaptively controlling the correlation likelihood for the second detection data according to the observation state.
Further, by narrowing down the possibility of correlation with a highly accurate track and limiting hypotheses, the processing load of the tracking device for performing correct correlation determination can be reduced.

  DESCRIPTION OF SYMBOLS 100 Tracking apparatus, 110 Detection data acquisition part, 120 Track data storage part, 131 Gate calculation part, 132 Gate inside / outside determination part, 133 Correlation likelihood calculation part, 134 Track data update part, 135 Track establishment determination part, 136 Track priority Setting unit, 137 Gate area storage unit, 141 correlation unit, 142 update unit, 143 new target selection unit, 144 new track generation unit, 190 track output unit, 511, 512 detection data, 521, 522 track group, 523 update track group 531 gate, 532 gate inside / outside determination result, 533 correlation likelihood, 534 established track information, 535 track priority, 536 gate setting area, 701 range, 711, 712 correlation likelihood, 800 tracking system, 801 sensor, 802 display device , 911 processing device, 912 input device, 913 output device 914 storage device.

Claims (5)

A detection data acquisition unit, a wake storage unit, a prediction unit, a likelihood calculation unit, a correlation unit, an update unit, and a reliability determination unit,
The detection data acquisition unit, state-like more than one sensor for monitoring a respective condition amount for each first observation time after the observation time of the target including a primary goal has been observed current observation time One or more detection data including the observation value representing the quantity and the current observation time when the sensor observed the state quantity ,
The track storage unit stores a track data including an estimate representing the state of the upper Symbol primary goal for each observation time of the first observation time after,
The prediction unit, for each combines only with the detection data in which the track data track storage unit is stored with the detection data acquisition unit has acquired, based on an estimate of the track data includes, time of the detection data comprises A predicted value representing the state quantity of the first target at the observation time of
When the track data stored in the track storage unit includes the estimated value at the first observation time and does not include the estimated value at the observation time after the first observation time, It is determined that the reliability of the wake data is low, and the wake data includes the estimated value at the first observation time and the estimated value at the observation time after the first observation time. Judge that reliability is high,
The likelihood calculation unit, the case that the reliability determination unit determines that reliability is high in the track data, each combining only the detection data in which the track data and the detection data acquisition unit has acquired, the prediction It represents a predicted value of parts was calculated, based on the difference between the observed value the detection data includes a plausibility observations which the detection data is included in an observation value representing a state quantity of the first target When the likelihood is calculated and the reliability determination unit determines that the reliability of the track data is low, the prediction unit calculates for each combination of the track data and the detection data acquired by the detection data acquisition unit. Regardless of the difference between the predicted value and the observed value included in the detection data, a certain likelihood is calculated,
The correlation unit compares the likelihoods each other the track storage unit has calculated the likelihood calculator per combines only with detection data track data and the detection data acquisition unit for storing acquired, the first one or more detection data that may contain observations representing a state quantity of the target to determine the constant on the basis of the comparison result,
The updating unit, for each combine only with the detection data in which the track track data and the correlation unit to the storage unit is stored is determined that there is a possibility that include observations representing a state quantity of the first target In addition , the tracking device is characterized in that, based on the observation value included in the detection data, wake data including the estimated value at the current observation time included in the detection data is generated . The wake data stored by the wake storage unit includes the estimated values of the position and speed , which are the state quantities of the first target,
The detection data detection data acquisition unit acquires includes observations of position is a state of one of the target,
The prediction unit, for each combines only with the detection data in which the track data track storage unit is stored with the detection data acquisition unit has acquired, based on an estimate of the position and velocity of the track data includes the above Calculate the predicted value of the position included in the track data at the current observation time included in the detection data,
The updating unit, for each combine only with the detection data in which the track track data and the correlation unit to the storage unit is stored is determined that there is a possibility that include observations representing a state quantity of the first target to the estimate of the position and velocity of the track data includes, based on the observed value of the position where the detection data comprises, plus an estimate of the position and velocity at the current observation time at which the detection data comprises Generate wake data,
The likelihood calculation unit, when the upper Symbol reliability determination unit determines that reliability is high in the track data, each combining only the detection data in which the track data and the detection data acquisition unit has acquired, the and the predicted value of the position prediction section is calculated, based on the difference between the observed value of the position where the detection data includes calculating the likelihood, upper Symbol reliability determination unit determines that reliability is low in the track data If the difference in per combines only with the detection data in which the track data and the detection data acquisition unit has acquired, and the predicted value of the position of the prediction unit is calculated, and the observed value of the position where the detection data comprises 2. The tracking device according to claim 1, wherein a constant likelihood is calculated regardless of the tracking device. The wake data stored by the wake storage unit includes the estimated values of the position and speed , which are the state quantities of the first target,
The detection data acquired by the detection data acquisition unit includes observation values of position and velocity , which are state quantities of any target,
The prediction unit, for each combines only with the detection data in which the track data track storage unit is stored with the detection data acquisition unit has acquired, based on an estimate of the position and velocity of the track data includes the above Calculate the predicted value representing the position of the first target at the current observation time included in the detection data,
The likelihood calculation unit, when the upper Symbol reliability determination unit determines that the low reliability of the track data, each combining only the detection data in which the track data and the detection data acquisition unit has acquired, the calculating a velocity based on the estimated value and the observed value of the position, including the detection data of the position where the track data includes, if the calculated velocity is less than the speed threshold, the predicted value of the position of the prediction unit is calculated Regardless of the difference between the observed value of the position included in the detection data, the likelihood is calculated based on the difference between the calculated speed and the observed value of the speed included in the detection data , and the calculated speed is not less than the speed threshold value. In some cases, regardless of the difference between the predicted value of the position calculated by the prediction unit and the observed value of the position included in the detection data, a certain likelihood is calculated,
The update unit, for each combination of the track data stored in the track storage unit and the detection data determined that the correlation unit may include an observation value representing the state quantity of the first target, Based on the estimated position and speed values included in the track data and the observed position and speed values included in the detection data, the estimated position and speed values at the current observation time included in the detection data are added. The tracking device according to claim 1 or 2, wherein the track data is generated . A computer program that uses a record track storage unit that stores track data including an estimated value representing a state quantity of a first target for each observation time after the first observation time,
An observation value representing the state quantity observed at the current observation time by the sensor that performs observation of each state quantity of one or more targets including the first target at each observation time after the first observation time, and the above A detection data acquisition unit that acquires one or more detection data including the current observation time at which the sensor observed the state quantity;
For each combination of track data stored in the track storage unit and detection data acquired by the detection data acquisition unit, the first time at the current observation time included in the detection data is based on an estimated value included in the track data. A predictor that calculates a predicted value representing the target state quantity of
The reliability of the wake data is low when the wake data stored in the wake storage unit includes the estimated value at the first observation time and does not include the estimated value at the observation time later than the first observation time. If the track data includes the estimated value at the first observation time and also includes the estimated value at the observation time later than the first observation time, the reliability of the track data is determined to be high. A reliability determination unit;
When the reliability determination unit determines that the reliability of the track data is high, the prediction value calculated by the prediction unit for each combination of the track data and the detection data acquired by the detection data acquisition unit, and the Based on the difference from the observation value included in the detection data, the likelihood indicating the likelihood that the observation value included in the detection data is an observation value indicating the state quantity of the first target is calculated, and the reliability determination unit If the reliability of the track data is determined to be low, for each combination of the track data and the detection data acquired by the detection data acquisition unit, the prediction value calculated by the prediction unit and the observation included in the detection data A likelihood calculating unit for calculating a certain likelihood regardless of the difference between the values,
The likelihood calculated by the likelihood calculating unit is compared for each combination of the track data stored in the track storing unit and the detection data acquired by the detection data acquiring unit, and represents the state quantity of the first target A correlator that determines one or more detection data that may contain observations based on the comparison results;
The detection data includes for each combination of the track data stored by the track storage unit and the detection data that the correlation unit has determined to possibly include the observation value representing the state quantity of the first target. An update unit that generates track data that includes the estimated value at the current observation time included in the detection data, based on the observed data
A computer program that causes a computer to function as. In a tracking method in which a tracking device having a detection data acquisition unit, a wake storage unit, a prediction unit, a likelihood calculation unit, a correlation unit, an update unit, and a reliability determination unit tracks a target,
The detection data acquisition unit, one or more respective state quantities observed first observation time after the observation time state like the sensor is observed in this observation time at which each of the target including a first target One or more detection data including the observation value representing the quantity and the current observation time when the sensor observed the state quantity ,
The track storage unit may store the track data including an estimate representing the state of the upper Symbol primary goal for each observation time of the first observation time after,
The prediction unit, for each combines only with the detection data in which the track data track storage unit is stored with the detection data acquisition unit has acquired, based on an estimate of the track data includes, time of the detection data comprises A predicted value representing the state quantity of the first target at the observation time of
When the track data stored in the track storage unit includes an estimated value at the first observation time and does not include an estimated value at an observation time later than the first observation time, the reliability determination unit It is determined that the reliability of the wake data is low, and the wake data includes the estimated value at the first observation time and the estimated value at the observation time after the first observation time. Judge that reliability is high,
The likelihood calculation unit, the case that the reliability determination unit determines that reliability is high in the track data, each combining only the detection data in which the track data and the detection data acquisition unit has acquired, the prediction It represents a predicted value of parts was calculated, based on the difference between the observed value the detection data includes a plausibility observations which the detection data is included in an observation value representing a state quantity of the first target When the likelihood is calculated and the reliability determination unit determines that the reliability of the track data is low, the prediction unit calculates for each combination of the track data and the detection data acquired by the detection data acquisition unit. Regardless of the difference between the predicted value and the observed value included in the detection data, a certain likelihood is calculated,
The correlation unit, compares the likelihood each other the track storage unit has calculated the likelihood calculator per combines only with detection data track data and the detection data acquisition unit for storing acquired, the first one or more detection data that may contain observations representing a state quantity of the target to determine the constant on the basis of the comparison result,
The updating unit, each said track track data and the correlation unit to the storage unit and stored by the combine only with detection data determined to be likely to contain observations representing a state quantity of the first target Further, the tracking method is characterized in that, based on the observation value included in the detection data, wake data including the estimated value at the current observation time included in the detection data is generated .

Images