JP4702010B2 - Tracking processing system - Google Patents

Description

  The present invention relates to an apparatus for performing observation using information obtained by a sensor for observing a wide area in a sensor system that tracks a target locus using a sensor mounted on a moving body.

  Sensor data integration means that integrates observation data from multiple sensors with sensor type and time and incorporates identifiers, and observation vectors from existing tracks included in a hypothesis table in a cluster that is a collection of target prediction existence ranges Gate calculation means to obtain possible physical area, and in-gate decision matrix to obtain all the possibility that the observation vector is an error signal, existing track, and new target for each new and existing cluster with respect to the observation vector extracted above An apparatus including a calculating means, a wake correlation matrix calculating means, a hypothesis updating means, and a wake determining device is disclosed (for example, see Patent Document 1).

Japanese Patent Laid-Open No. 9-318741 (FIG. 1)

  The conventional tracking processing system is configured as described above, and is based on the premise that only observation data observed by a plurality of sensors is used. Existence could not be calculated, and there was a problem that the possibility that two targets could be identified became low until the time when two observation values were continuously input.

  The present invention has been made to solve the above-described problems. By using observation information on the number of tracks of a wide area sensor, wide-area information with low accuracy (cluster information corresponding to observation information of the wide-area sensor is used. The output of the wide area information correlating means to which is added) is corrected for the reliability of the tracking process, and the object is to improve the quality of the track output by the tracking process using the observation value of the mounted sensor with high accuracy.

The tracking processing system according to the present invention includes a wide area information obtaining unit that obtains target observation information from a wide area sensor, a wake-related data storage unit that stores wake-related data in a cluster unit that is a set of hypotheses, The cluster update means for creating and updating a new cluster from the target information from the on-board sensor and the existing cluster information stored in the track related data storage means, and the track stored in the track related data storage means With reference to related data, a wide area information correlation unit that selects a target cluster having a relative distance between the center of gravity position of the cluster and the target observation position obtained by the wide area information acquisition unit, and the wide area information correlation Trust that improves the reliability of hypotheses that are observed in the number of wakes greater than the number of observations in the global information among the hypotheses included in the target cluster selected by means And correction means, based on the reliability was improved by the reliability correcting means is for and a correlation-integration means for determining the reliability of the hypothesis.

According to the present invention, by using the observation information of the number of tracks of the wide area sensor, the reliability of the tracking process is corrected for the information of the wide area sensor with low accuracy, and the tracking value using the observation value of the high accuracy mounted sensor is used. There is an effect that the quality of the wake output by the processing can be improved.
In addition, since the observation information of the wide area sensor is not added to the sequence of the tracking processing main body, tracking processing such as MHT (Multiple Hypothesis Tracking) can be realized without increasing the number of observation points to be processed, and the processing load is also reduced. .

There is a system in which a sensor system is mounted on a moving body such as an aircraft or a ship, and measures such as observation are performed by approaching a target. In these mobile objects, it is conceivable that a communication device is mounted and a system for tracking a target trajectory is constructed by using information obtained by a sensor for observing a wide area in addition to information on the mounted sensor.
At this time, the method of tracking the target using observation information obtained from a plurality of sensors greatly affects the performance of the entire system.

In general, sensors mounted on a moving object are usually mounted on a moving object because of the short distance to the observation target and the concentration of sensor resources on the target or area of interest. Compared with information from sensors that observe a wide area, the target observation accuracy is high.
In order to track the target, it is necessary to continuously point the sensor in the target direction. However, in the sensor mounted on the moving body, the distance to the observation target is short. The change becomes larger and the update rate of observation is faster than the information of sensors that observe a wide area.

On the other hand, there are restrictions on weight, volume, power, etc. in sensors mounted on a moving body due to the influence on the movement performance of the moving body, etc., which often imposes restrictions on downsizing and mounting position. .
For this reason, in general, a sensor mounted on a moving body has a narrow observation coverage and a short detection distance compared to a sensor that observes a wide area.
In addition, sensors for observing a wide area have a larger storage capacity because the restrictions on the mounting conditions are more gradual than those of moving objects, so that not only the target position and speed at the present time but also the target moves. It is possible to grasp and record the route.

As described above, a sensor mounted on a moving body and a sensor that can observe a wide area have different advantages.
For this reason, there is a need for a sensor system mounted on a moving body that uses observation information from a wide-area sensor that can make use of each feature.
In a sensor system for a moving body and a sensor system for observing a wide area, a configuration in which information of a plurality of sensors is combined in each system is considered.

In this configuration, in the combination of a plurality of sensors, not only the same type of sensor but also a configuration in which different types of sensors (for example, radar and optical sensors as radio wave sensors) are combined is considered.
There is one for constructing a system for tracking a target locus using observation information from a plurality of sensors.
There are those that use observation information of sensors arranged in a wide area, and those that combine radar and optical sensors.

It collects the detection data observed by a plurality of sensors and performs the tracking process (generates the wake by the tracking filter). The correlation / integration process is performed by the detection data integration type to track the target locus.
Further, when creating a wake from detection data, algorithms such as JPDA (Joint Probabilistic Data Association) and wake-type MHT (Multiple Hypothesis Tracking) are used.
This method is characterized in that it can use all detection data observed by the sensor and has high tracking ability.

In the tracking method, all detection data observed by the sensor can be used.
However, when a plurality of sensors are used for observation, the observation accuracy, update cycle, and observable items may differ depending on the sensor.
In order to deal with such a case, for example, a filter for a radio wave sensor that can observe the target azimuth, distance, and speed and an optical sensor that observes only the target azimuth are provided, depending on the sensor used in the observation. The processing method is used.
In the tracking method described above, it is possible to use detection data with inferior observation accuracy or information with a slow update cycle (information input with a delay from the tracking processing execution time), but generally contributes to the improvement of tracking capability. I know it ’s small.

For this reason, for example, when data at the same time is input from a plurality of sensors, only detection data of a sensor with a high priority is selected, or a hypothesis is deleted by giving priority to information from a highly reliable sensor ( The wake by the sensor with low reliability is deleted), and means for generating the wake with priority given to the information from the sensor with high observation accuracy is provided.

Also, there is provided means for selecting detection data and hypothesis information to be used in wake generation and management, such as deleting those with a long hypothesis update cycle.
On the other hand, there is a difference in observation accuracy between a sensor mounted on a mobile object, which is a sensor targeted in this patent, and a sensor that observes a wide area (in general, a sensor mounted on a mobile object has high accuracy). .

In addition, since it takes time to collect observation information and transmit it to the moving body, information on sensors that observe a wide area is delayed compared to sensors mounted on the moving body.
When the target is capable of high-speed motion with high speed like an aircraft, it is necessary to use information with high accuracy and little delay time.

For this reason, when trying to integrate the information of the sensor mounted on the mobile body and the sensor that observes a wide area, the observation information of the wide area sensor with low accuracy and large delay is rarely used.
A sensor system for observing a wide area can have a large storage capacity and can grasp not only the current target position and speed but also the route along which the target has moved and can be recorded.

In addition, management of information combined with pre-obtained foresight information and the like is also more suitable than the installed system. In the correlation / integration method based on the conventional detection data, the observation information of the wide area sensor is not used due to the accuracy and delay time of the sensor, and the advantages of the wide area sensor system cannot be fully utilized.
For this reason, even if correlation and integration is performed by sensor data fusion in combination with observation information of an on-board sensor, it is difficult to improve the tracking ability.

Here, the problem assumed by the present invention will be described with reference to FIGS.
4 is a schematic diagram of a scenario in which a problem occurs, FIG. 5 is a graph showing the timing at which information is input from each sensor, and FIG. 6 is a diagram showing information (observed values) input from each sensor in time series. It is.
FIG. 7 is a schematic diagram around the time T4 when the problem occurs.

FIG. 4 shows a scenario in which two targets to be observed take a course toward the moving body.
Here, it is assumed that two different sensors are mounted on the moving body, and that information from the wide area sensor can be obtained through communication means.
The observation range (observable distance) of the sensor is that the wide area sensor is longer than the mobile sensor, and the resolution and observation accuracy are superior to the wide area sensor.

The target path is far away (a range that can be observed by a wide-area sensor) so far as it can be confirmed with two targets by a wide-area sensor. The scenario is such that the sensor is approaching to the extent that separation is difficult.
In FIG. 4, a graph showing the timing at which information is input from each sensor for the portion indicated as times T0 to T9 is highlighted in FIG. 5, and the information (observed values) input from each sensor is shown in FIG. FIG. 6 shows the series.

FIG. 5 shows the time at which the target is observed by the sensor and the time at which the observed target information is input to the correlation / integration process with respect to the observation information from the sensors A, B, and wide area sensors mounted on the moving body.
Here, due to the tendency of general mounted sensors and wide area sensors, the mounted sensors input target information to the correlation / integration process at the time the target was observed, and the wide area sensor is delayed from the observation time due to the transmission time by communication, etc. Assume that the target information is input to the correlation / integration process.
In addition, it is assumed that the on-board sensor has a shorter information update cycle than the wide-area sensor, and the on-board sensor has the same update cycle.

In FIG. 5, the information of the mounted sensor A is input at time T0, T2, T4, T6, T8 and 2 unit time periods (once every 2 unit times). The observation time at the sensor and the time input to the correlation / integration processing (information input time) are the same time with no delay.
In addition, the information of the mounted sensor B is also input at times T1, T3, T5, T7, and T9 with a two-unit time period, and the observation time and the information input time are the same.
For the wide area sensor, information is input at times T4 and T8 and 4 unit time periods (once every 4 unit times).
The delay of the information input time is 4 unit times (the observation result at time T0 is input to the correlation integration process at time T4).

FIG. 6 is a diagram showing information (observed values) input from each sensor in time series for each time. Each observation value indicates the position where the target is observed.
Here, the observation value at the target time and the past observation value are presented (at each time, the observation value newly added at the target time is presented in addition to the past observation value).
For example, at time T0, only the observation information p0 of the mounted sensor A is displayed. At the next time T1, the observation information p1 of the mounted sensor B increases, such as p2 of the mounted sensor A at time T2, and p4 of the mounted sensor B at time T3.

At time T4, in addition to p4 of the mounted sensor A, w1-1 and w1-2 of the wide area sensor are input.
Here, w1-1 and w1-2 overlap because it is impossible to separate by sensor observation, but it is known from the past observation results that there are two targets. It represents that.

The operation of the apparatus will be described with reference to FIGS.
From time T0 to T3, p0, p1, p2, and p3 and target information one by one are input from the mounted sensor.
At this time, in the wake type MHT algorithm used in the apparatus, the following hypothesis and tentative wake are generated.

In the wake type MHT, there are a confirmed wake (hereinafter referred to as a main wake) to be output to the outside and a temporary wake (hereinafter referred to as a tentative wake) managed inside the algorithm.
The temporary wake is a candidate for the main wake, and a temporary wake (a hypothesis to which the tentative wake belongs) calculated by the wake type MHT algorithm is selected and becomes the main wake.

At time T0, a temporary track “tr0-1” composed of one observation result “p0” is generated, a hypothesis “H0-1” composed of only tr0-1 and a hypothesis “H0-2” composed of an empty set. Is generated.
In the following, for the sake of simplicity, tr0-1 = {p0} represents the temporary track and component observation results, H0-1 = [tr0-1] represents the hypothesis and component track, and Φ represents the empty set. .
At time T0, tr0-1 = {p0}, H0-1 = [tr0-1], and H0-2 = [Φ] are generated.
In the wake type MHT algorithm, the reliability of the hypothesis is calculated from the observation accuracy of the sensor, the detection probability, and the like.

At time T0, the reliability calculated for H0-1 and H0-2 is set.
Next, the observation result “p1” is input at time T1.
Here, it is assumed that p1 is in the gate of tr0-1.
In the tracking process, wakes are generated by connecting related observation values.

At this time, in order to determine the related observation value, the target exists from the existing wake with position and velocity information, and the target motion is assumed by considering the observation error range of the sensor, etc. Set the range that can be used as a “gate”.
Observation values in the gate are candidates for observation values constituting the wake (a process of generating a gate and determining whether the observation value is inside or outside the gate is called a gate determination process).
The wake type MHT is characterized in that the correspondence between the observation value determined to be in the gate and the existing wake is postponed.

When p1 is in the gate of tr0-1, at time T1, provisional wakes tr1-1 = {p0, p1} and tr1-2 = {p1}, hypothesis H1-1 = [tr1-1], hypothesis H1- 2 = [tr0-1], hypothesis H1-3 = [tr0-1, tr1-2], hypothesis H1-4 = [tr1-2], hypothesis H1-5 = [Φ] are generated.
Hypothesis H1-1 is the hypothesis that observation results p0 and p1 are the same target observation value (provisional track tr1-1), hypothesis H1-2 is hypothesis H0-1 is correct, and p2 is clutter (only temporary track tr0-1 exists) ), And hypothesis H1-3 is a hypothesis of a target track having different p0 and p1 (presence tracks tr0-1 and tr1-2 exist) (hypothesis H1-1 to H1-3 are hypotheses H0-1). Is the correct assumption hypothesis).

Hypothesis H1-4 is a hypothesis that hypothesis H0-1 is correct and only p2 observes the target (only temporary track tr0-1 exists), hypothesis H1-5 is hypothesis H0-1 is correct, and p2 is clutter (target It is a hypothesis that there is no observation result, that is, there is no wake.
In the wake type MHT, when an observation value is input as described above, the process proceeds while generating a hypothesis.

In the wake type MHT, the number of hypotheses increases exponentially unless a constraint is provided, so that a highly reliable hypothesis is left and other hypotheses are rejected in order to improve processing time efficiency.
Here, as a result of proceeding with the processing at times T2 and T3, tr3-1 = {p0, p1, p2, p3}, temporary track tr3-2 = {p0, p1, p3} (observed values at time T2) Suppose that the tentative track tr3-3 = {p2} is in a state where H3-1 = [tr3-1] and H3-2 = [tr3-2, tr3-3] exist as hypotheses. .

At this time, the reliability of the hypothesis including the temporary track for which the observation value (detection data) is continuously obtained increases, the reliability of H3-1 is 90%, the reliability of H3-2 is 10%, The subsequent operation will be described using an example in which the process at time T4 is started.
At time T4, observation values p4-1 and p4-2 are input from the onboard sensor, and observation values w4-1 and w4-2 are input as wide area information.

From FIG. 6, p4-1 and p4-2 are the results of observation at time T4, and w4-1 and w4-2 are the results of observation at time T0.
At this point, the observation values of the on-board sensors at times T0 to T3 have already been processed, and the observation accuracy is better for p0 to p3, p4-1 and p4-2 than for w4-1 and w4-2.
In the tracking process, a wake is generated.
The main purposes of generating wakes in mobile systems are: 1. knowing the current position and speed of the target in order to carry out the action addressing the target; It is to predict the target position (and speed) at the next time in order to point the mounted sensor.

Further, in order to feed back the tracking processing result to the sensor control, it is required to realize the processing in real time.
In general, if the accuracy of the sensors to be observed is the same, the position / velocity prediction at the next time is more accurate when the tracking process is performed with the observation result at the latest time.
In the situation where there is information (p4-1 and p4-2) with high accuracy and new observation time, information on sensors with low observation accuracy, such as w4-1 and w4-2 at time T4, is old. , Even if it is input as an observation value for tracking processing such as MHT, it is unlikely to contribute to improving the accuracy of the track (in addition to p4-1 and p4-2, w4 Even if -1 and w4-2 are additionally input, it is unlikely that the predicted position / speed at time T4 will improve).

Further, when the number of input observation values increases, there is a disadvantage that the processing time for the tracking process increases.
For this reason, functions such as w4-1 and w4-2 at time T4 that restrict the input of sensor information with old observation times and low observation accuracy, and the function that restricts the generation of hypotheses containing these observations Was set up.
The operation when the observation information of w4-1 and w4-2 at time T4 is not used is shown.
Here, in order to simplify the description, an example in which only representative provisional tracks and hypotheses are generated will be described.
At time T4 of the conventional system, the temporary track and hypothesis are updated using only the observation values p4-1 and p4-2.

  Here, on the basis of hypothesis H3-1, hypothesis H4-1 = [tr4 that generates tr4-1 = {p0, p1, p2, p3, p4-1} by adding observed value p4-1 to tr3-1 −1], tr3-1 plus the observed value p4-2 to generate tr4-2 = {p0, p1, p2, p3, p4-2} Hypothesis H4-2 = [tr4-2], tr4-1 And the hypothesis H4-3 = [tr4-1, tr4-3], which generates the temporary track tr4-3 = {p4-2} consisting only of the observed value p4-2 and the observed value p4-1 The hypothesis H4-4 = [tr4-2, tr4-4] for generating the tentative track tr4-4 = {p4-1} is generated.

  Based on Hypothesis H3-2, tr4-5 = {p0, p1 p3, p4-1} obtained by adding the observed value p4-1 to the temporary track tr3-2 and the observed value p4-2 on the temporary track tr3-2. Tr4-6 = {p2, p4-2} with hypothesis H4-5 = [tr4-5, tr4-6], tr4-7 = Hypothesis H4-6 = [tr4-7, tr4 to generate tr4-8 = {p2, p4-1} by adding the observed value p4-1 to the temporary track tr3-2 and {p0, p1 p3, p4-2} -8] is generated.

As the reliability within each hypothesis, hypothesis H4-1: hypothesis H4-2: hypothesis H4-3: hypothesis H4-4 is 3: 3: 2: 2, hypothesis H4-5: hypothesis H4-6 is 7: 3 Suppose that
In this case, since the reliability of the reference H3-1 was 90% and the reliability of H3-2 was 10%, the hypothesis H4-1 was 27%, the hypothesis H4-2 was 27%, and the hypothesis H4-3 was 18%, hypothesis H4-4 is 18%, hypothesis H4-5 is 7%, and hypothesis H4-5 is 3%.
If the top two hypotheses are selected in this state, only hypothesis H4-1 and hypothesis H4-2 are selected, and the hypothesis with two tentative tracks close to the true value as candidates is rejected.
Due to this influence, there is a problem that the possibility that the two targets can be identified becomes low until the time T8 and T9 at which two observation values are continuously input have passed.

Since the tracking processing system calculates a wake based on observation values, the presence of two targets cannot be calculated unless two observation values from the sensor are input.
Further, in the process of generating a wake using an observation value as an input, it is advantageous to limit the input of information of a sensor having an old observation time and low observation accuracy.
Here, at time T4, the input of observation values from the wide area sensor is not limited, and information that the two targets obtained by the wide area sensor exist cannot be used.
The present invention has been made on the basis of the above situation and problems, and the reliability of the hypothesis is determined in the hypothesis generation / quasi-optimization phase based on the reliability correction result of the reliability correction means. The correlation / integration means is provided. Hereinafter, more detailed examples will be described.

Embodiment 1 FIG.
FIG. 1 is a block diagram of a first embodiment of the present invention. 1 is a tracking processing system, 2 is observation data selection means, 3 is cluster update means, 4 is wake-related data storage means, 5 is correlation / integration means, 6 Is a cluster management means, 7 is a prediction means, 8 is a track determination / display means, 9 is a sensor control means, 10 is a wide area information correlation means, 11 is a reliability correction means, 12-1 is an on-board sensor 1, and 12-2 is The mounted sensors 2 and 12-N are mounted sensors N, 13 is a display device, and 14 is a wide area information obtaining unit.

  In the first embodiment, the tracking processing system 1, the mounted sensor 1 (12-1), the mounted sensor 2 (12-2),..., The mounted sensor N (12-N), the display device 13, and the wide area information obtaining means 14 are: A device mounted on a moving body such as an aircraft.

In order to simplify the description, an example in which the moving body is an aircraft will be described below.
The tracking processing system 1 is a system that performs tracking processing calculation.
Mounted sensor 1 (12-1), mounted sensor 2 (12-2), ..., mounted sensor N (12-N) are sensors mounted on an aircraft.
In this patent, the number of mounted sensors can be set to an arbitrary number of one or more, but an example of two mounted sensors, which is the minimum number when a plurality of sensors are mounted, will be described.
These mounted sensors are referred to as mounted sensor 1 (12-1) and mounted sensor 2 (12-2).
The sensor performance (observation accuracy, type of observable information, sensor update period, delay time from observation to information output, etc.) of the mounted sensor 1 (12-1) and the mounted sensor 2 (12-2) are equivalent. This will be explained with an example.

The types of information that can be observed will be described using examples of target position and speed.
The wide area information obtaining means 14 is an apparatus that communicates with the outside of the aircraft using means such as radio waves, obtains observation information of the wide area sensor, and transmits the contents to the tracking processing system 1.
The display device 13 is a device for presenting the processing result of the tracking processing system 1 to an operating person.

  2 is a flowchart of the tracking processing system 1, FIG. 3 is a flowchart of the wide area information correlation unit 10 and the reliability correction unit 11, and FIGS. 4 to 7 are diagrams for explaining an outline of a scenario in which a problem occurs in the system. .

The operation of the tracking processing system 1 will be described with reference to FIG.
In the tracking processing system 1, when target information from an on-board sensor is input, the target input phase of STEP1-1 is first entered.
Here, the observation data selection means 2 in the tracking processing system 1 accepts observation information from the on-board sensor.

This observation data selection means 2 uses the tracking algorithm such as MHT for the input information from the performance (observation accuracy, etc.) of the sensor that inputs information and the observation time (difference between current time and observation time). Thus, it is means for determining whether or not to use in the processing of STEP1-2.
In the first embodiment, in order to simplify the description, all information from the on-board sensor will be described as an example used in the processing of STEP1-2.

In STEP 1-1, when target information is input from the mounted sensor 1 (12-1) and the mounted sensor 2 (12-2), the observation data selection unit 2 sends all the input information to the cluster update unit 3.
At this time, the tracking processing system 1 makes a transition from the target input phase of STEP 1-1 to the cluster update phase of STEP 1-2.

In the cluster update phase of STEP 1-2, the existing cluster information is compared with the input target information from the sensor, and update processing such as new cluster creation / integration is performed.
In the wake type MHT algorithm, hypotheses, provisional wakes, and observation results (detection results of onboard sensors) are managed in logical units called clusters. The wake-type MHT algorithm manages past observation information in units of clusters in order to perform postponement determination, and records this information in the wake-related data storage means 4.

For this reason, the cluster update means 3 updates the new creation / integration of clusters from the target information from the sensor input from the observation data selection means 2 and the existing cluster information stored in the wake-related data storage means 4. Process.
After the update is completed, the processing of STEP 1-3 to STEP 1-6 is performed for each cluster.

After the cluster update in STEP 1-3, the processing of STEP 1-3 to STEP 1-7 is performed for each cluster.
STEP 1-3 to STEP 1-5 are executed by the correlation / integration means 5, STEP 1-6 is executed by the cluster management means 6, and STEP 1-7 is executed by the prediction means 7.
Of these, STEP1-3, STEP1-4, STEP1-6, and STEP1-7 are processed, and the track-related data storage means 4 is referred to / updated.

In this apparatus, in the hypothesis generation / quasi-optimization phase of STEP 1-5, the correlation / integration unit 5 determines the reliability of the hypothesis in cooperation with the reliability correction unit 11.
In the tracking processing system 1, when the target information from the mounted sensor is input, the above STEP1-1 to STEP1-7 are performed to update the cluster information (target track information).
On the other hand, information for presenting the processing result of the tracking processing system 1 to an operating person is generated.
Since it is not always necessary to synchronize with the information input from the sensor, an example in which output information is generated at a constant period independently of the sensor will be described here.

In the tracking processing system 1, the processes of STEP 1-8 and STEP 1-9 are called at a constant cycle.
At this time, the track determination / display unit 8 selects the track to be displayed by referring to the track related data storage unit 4 in STEP 1-8 (in the track type MHT, a plurality of track candidates called temporary tracks are held. In step 1-9, the track to be displayed (main track) is sent to the display device 13. The display device 13 displays the sent information on a display or the like, and presents it to the operating person.

The operations of the wide area information correlation unit 10 and the reliability correction unit 11 will be described with reference to FIG.
When the target information from the wide area sensor is input, first, the phase of the wide area target input of STEP2-1 is entered.
Here, the wide area information acquisition means 14 receives the information, sends the information to the wide area information correlation means 10, and transits to STEP2-2.

In STEP 2-2, the wide area information correlation unit 10 refers to the wake related data storage unit 4 and selects the cluster closest to the observation information from the wide area sensor as the “target cluster”.
Here, in each cluster, the current position of the temporary track in the cluster is calculated based on the observation time of the observation information from the wide area sensor, and the center of gravity position of the cluster is calculated based on the result of weighting with the reliability of the hypothesis. The relative distance between the center of gravity position and the observation position of the wide area sensor is calculated, and the closest one is selected. The observation information from the temporary track and the wide area sensor includes information on the target position, speed, and observation time, and the observation position at the specified time can be calculated by external insertion using the speed.

Here, if the relative distance to the nearest cluster is equal to or greater than a preset threshold, it is assumed that there is no cluster corresponding to observation information from the wide area sensor.
In STEP 2-3, the reliability correction means 11 includes hypotheses included in the target cluster designated by the wide area information correlation means 10 and those whose number of tracks coincides with the observation value of the wide area information and hypotheses that are observed further. Improve reliability.
Based on the correction result of the confidence in the phase of hypothesis generation and quasi optimization Step 1-5, the correlation and integration means 5, to determine the reliability of the hypothesis.

The operation when the present apparatus described above is applied to the problem scenario described with reference to FIGS.
In this apparatus, the operations of STEP2-1 to STEP2-3 and STEP1-5 are different from those of the conventional apparatus.
For this reason, the observation information of w4-1 and w4-2 at time T4 in FIGS. 3 to 7 is updated by the wide area information correlation unit 10 and the reliability correction unit 11 in the update of the hypothesis of STEP 1-5, where the number of tracks is 2 or more. Correct the hypothesis confidence.

Here, when hypotheses with 2 or more wakes and hypotheses with less than 2 are mixed, the reliability (total) of hypotheses with 2 or more wakes is the reliability (sum of) of hypotheses with 2 or less. An example of performing correction to 6 to 4 will be described.
At time T4 in FIGS. 3 to 7, first, the reliability of hypothesis H3-1 and hypothesis H3-2 that are the basis for hypothesis generation is corrected.
Here, since the number of wakes of hypothesis H3-1 is 1 and the number of wakes of hypothesis H3-2 is 2, the reliability of hypothesis H3-2, which is 2 or more, is improved, hypothesis H3-1 is 40%, hypothesis H3-2 is set to 60%.

Next, of the hypotheses generated at time T4, hypothesis H4-1 and hypothesis H4-2 have a wake number of 1, hypothesis H4-3 and hypothesis H4-4 have a wake number of 2, so the reliability ratio is 2. : 2: 3: 3 is corrected.
Since hypothesis H4-5 and hypothesis H4-6 both have two wakes, the reliability ratio remains 7: 3.
As a result, the reliability after the processing at time T4 is 8% for hypothesis H4-1, 8% for hypothesis H4-2, 12% for hypothesis H4-3, 12% for hypothesis H4-4, and hypothesis H4-5. Is 42% and Hypothesis H4-5 is 18%.
By using the information of the wide area sensor, the reliability of the hypothesis holding two tracks close to the true number of tracks is improved.

As a result, even when a higher reliability hypothesis is selected, a hypothesis that retains two wakes that was rejected in the conventional method is selected (because the reliability is relatively high).
The wake based on the selected hypothesis (the main wake) is sent to the display device 13 by the wake determination / display means 8 with reference to the wake related data storage means 4, and the wake is displayed on the display by the display device 13. Presented to the person operating the state of the target.

In this way, this device uses the observation information of the number of tracks of the wide area sensor, corrects the reliability of the tracking process for the information of the wide area sensor with low accuracy, and uses the observation value of the mounted sensor with high accuracy. There is an advantage that the quality of the track output by the tracking process can be improved.
Further, since the observation information of the wide area sensor is not added to the sequence of the tracking processing main body, the tracking processing such as MHT can be realized without increasing the number of observation points to be processed, so that the processing load can be reduced.

  In the first embodiment, the observation data selection means 2 has been described as an example in which the observation values of all the mounted sensors are used. However, for the information of the mounted sensors, the input of the sensor information having a low observation accuracy and the observation time is old. It is also possible to provide a function for limiting and a function for limiting the generation of hypotheses including these observation values, and the same effects as in the first embodiment can be produced.

In the first embodiment, the configuration of the two mounted sensors, the mounted sensor 1 (12-1) and the mounted sensor 2 (12-2), has been described. However, the number of mounted sensors is realized by an arbitrary number of one or more. It is possible and can produce the same effect as the embodiment.
In addition, although the mounted sensor 1 (12-1) and the mounted sensor 2 (12-2) are sensors having the same performance capable of observing the position / velocity, even in a combination of sensors having different performances such as different observation accuracy. The effect similar to Example 1 can be emitted.
In addition, the same effect as that of the first embodiment can be achieved also in a combination of different types of sensors such as an infrared sensor capable of observing only an angle and a sensor that receives and observes radio waves emitted by the other party.

  In the first embodiment, the method of calculating the center of gravity position of the cluster based on the provisional track in the cluster and the reliability of the hypothesis has been described as the selection criterion for the corresponding cluster. Any correspondence selection method can be used as long as it can be determined on a one-to-one basis, and the same effect as in the first embodiment can be produced.

  In the first embodiment, an explanation is given of an example in which the reliability is corrected to 6 to 4 with a hypothesis that the number of wakes is greater than or equal to the number of wakes observed by the wide area sensor and a hypothesis that is less than the number of wakes observed by the wide area sensor. As a correction method, an arbitrary correction method in which a hypothesis that the number of wakes is more than the number of wakes observed by the wide area sensor can be selected, and the same effect as in the first embodiment can be produced.

Embodiment 2. FIG.
FIG. 8 is a block diagram of Embodiment 2 of the present invention, 15 is a reliability distribution means, 1 to 14 are the same as those in FIG. 1, and FIG. 9 is a cluster selection in the reliability distribution means 15. It is the figure which showed the outline | summary.
Here, in STEP 2-2, the wide area information correlation unit 10 sends the result to the reliability distribution unit 15 at the stage of obtaining the relative distance between the observation values of each cluster and the wide area sensor. Further, the reliability distribution unit 15 obtains cluster information such as the number of tracks included in the cluster from the track related data storage unit 4.

The reliability distribution means 15 selects a plurality of clusters corresponding to the wide area information, and distributes the number of tracks to the selected plurality of clusters with weighting. The number of tracks for each cluster whose distribution is determined by the reliability distribution unit 15 is sent to the reliability correction unit 11.
The reliability correction unit 11 uses the number of wakes received from the reliability distribution unit 15 in the same manner as the number of wakes obtained from the wide area information correlation unit 10 in the first embodiment.
Distribute weighted number of tracks.
Here, as in the first embodiment, a method based on the relative distance between the center of gravity of the cluster and the observation value of the wide area sensor will be described.

The reliability distribution means 15 distributes the clusters within a certain distance range by weighting the number of tracks according to the relative distance.
In the example of FIG. 9, there are clusters 1 to 3 as distribution targets, and all exist at the relative distance of the target to distribute the number of tracks.
The ratio of the relative distances is 1: 1: 2 for cluster 1: cluster 2: cluster 3.
In the wide area sensor, five tracks are densely present.
In this case, two are assigned to cluster 1, two to cluster 2, and one to cluster 3 according to the relative distance.

As described above, this apparatus has an effect that the observation information of the number of tracks of the wide area sensor can be distributed not only to one specific cluster but also to a plurality of surrounding clusters.
The observation value of the wide area sensor often has a large possible range because the update rate is often slow. For this reason, if only one specific cluster is selected, there is a risk that weighting information is concentrated on the wrong cluster.
In this apparatus, the risk when this incorrect correspondence is selected can be reduced.

In the second embodiment, an example in which the number of tracks is distributed in units of one or more is shown. However, in addition to being able to distribute the number of tracks after the decimal point, for example, the total number is set so that two tracks are allocated to the cluster 3 as well. Then, it is possible to assign more than the number of observations of the wide area sensor, and the same effect as in the second embodiment can be produced.
In the second embodiment, the number of wakes of the wide area sensor is collectively distributed. However, it is possible to distribute the weights by determining and weighting each wake, and the same effect as in the second embodiment can be obtained. it can.

In addition, in the case where the individual wakes and adjacent wake groups are mixed for the observation result of the wide area sensor, by distributing each wake group and the wake group that summarizes multiple wakes, It is possible to implement a mix of both.
Also in this case, the same effect as in the second embodiment can be produced.
In the second embodiment, the method of calculating the center of gravity position of the cluster based on the temporary track in the cluster and the reliability of the hypothesis has been described as the selection criterion for the corresponding cluster. Any method for selecting a correspondence relationship can be used as long as the weighting ratio set for each correspondence relationship can be determined, and the same effect as in the second embodiment can be produced.
In addition, in consideration of the number of wakes included in the current cluster, it is possible to implement a case where a large number of wakes of observation results of wide area sensors are assigned to a cluster including many wakes, and the same effect as in the second embodiment is produced. Can do.

Embodiment 3 FIG.
10 is a block diagram of Embodiment 3 of the present invention, 16 is a track determination support means, and 1 to 13 are the same as those in FIG.
Here, in STEP 2-3, the reliability correction unit 11 obtains information on the number of tracks included in the hypothesis and the number of tracks observed by the wide area sensor (the number of observations by the wide area sensor) for each cluster, and the track determination support means. 16 to send. The wake determination support means 16 improves the reliability of a new wake or a temporary wake with a small number of observation points when the number of wakes included in the hypothesis is smaller than the number of observations of the wide area sensor.

In general, sensor signals include false signals (clutter). For this reason, in the correlation / integration means 5, only one observation value does not give a high reliability as a wake, and the wake is composed of a plurality of observation values (the observation time interval may be considered). Improve reliability at the time and select it as a wake.
On the other hand, in this apparatus, information that there is a high possibility that the target is in the vicinity where the on-board sensor has obtained the observation value is obtained from the wide area sensor.
The wake determination support means 16 corrects the reliability calculation in the correlation / integration means 5 based on the observation result of the wide area sensor.
For this reason, the reliability is improved faster than when only the observation value of the on-board sensor is used, and the track can be selected in a short time.

In this way, this device can improve the reliability of new tracks and temporary tracks with a small number of observation points when the number of tracks included in the hypothesis is less than the observation information of the number of tracks of wide area sensors. The reliability can be improved.
This makes it possible to establish a wake with a smaller number of detections. By reducing the number of detections until the wake determination, there is an advantage that the wake can be established at a long distance.
In the third embodiment, the example in which the wake determination unit 16 is added to the first embodiment has been described. However, even in the configuration in which the wake determination unit 16 is added to the apparatus of the second embodiment, the same effect as in the third embodiment can be produced. .

Embodiment 4 FIG.
FIG. 11 is a block diagram of Embodiment 4 of the present invention, 17 is a cluster wake determination support means, and 1 to 14 are the same as those in FIG. The reliability correction unit 11 sends the number of tracks observed by the wide area sensor to the cluster track determination support unit 17 for each cluster.
When the target flies in the form of a dense corps, etc., it becomes difficult to separate individual targets using sensors. In this case, the observation value of the sensor is observed as a large target in a lump or the number of targets fluctuates, resulting in an unstable observation state.

In order to deal with such a situation, the correlation / integration unit 5 may select a countermeasure for tracking a target as a group of clusters (cluster) instead of individual targets.
The cluster wake determination support means 17 holds information on the number of wakes observed by the wide area sensor.
The correlation / integration unit 5 makes an inquiry to the cluster track determination support unit 17 when it is necessary to determine whether or not to deal with the cluster target.

At this time, the cluster wake determination support means 17 refers to the information on the number of wakes observed by the wide area sensor for the target cluster, and the target number captured in the cluster, which is the observation result of the mounted sensor, is observed by the wide area sensor. When the number is smaller than the target number, a determination result that the cluster target countermeasure is advantageous is returned.
Conversely, if the number of targets captured in the cluster, which is the observation result of the on-board sensor, is the same as or less than the number of targets observed by the wide area sensor, a determination result that the cluster target handling is disadvantageous is returned.
The correlation / integration unit 5 refers to the determination result of the cluster wake determination support unit 17 to determine whether or not cluster handling is possible.

  In this way, this device uses information on the number of wakes observed by a wide-area sensor as individual wakes separated from a specific temporary wake group or as a cluster target with multiple wakes (targets) overlapped. It can support the judgment. The track management in cluster units is a method used as an application of the track type MHT, and has an advantage in dealing with a dense target. At this time, there are cases in which it is difficult to determine whether or not the target is a dense target only by the information of the mounted sensor, and there is an advantage that this determination can be easily performed by giving a target number by a wide area sensor.

In the fourth embodiment, the example in which only the advantages / disadvantages of cluster target management are returned as the determination result of the cluster wake determination support means 17 has been described. Yes, the same effect as in the fourth embodiment can be produced.
It is also possible to set a predicted value of the number of tracks included as a cluster target from the cluster track determination support means 17, and in this case, in tracking of the cluster target of the correlation / integration means 5, the target included in the cluster There is an advantage that the number can be estimated more accurately.

  In the fourth embodiment, the example in which the cluster wake determination support unit 17 is added to the first embodiment has been described. However, the configuration in which the cluster wake determination support unit 17 is added to the apparatus of the second or third embodiment may Similar effects can be produced.

Embodiment 5 FIG.
FIG. 12 is a block diagram of Embodiment 5 of the present invention, 18 is a cluster deletion support means, and 1 to 14 are the same as those in FIG.
Here, in STEP 2-3, the wide area information correlation unit 10 sends the number of wakes observed by the wide area sensor to the cluster wake determination support unit 8 for each cluster. The cluster deletion support means 18 informs the cluster management means 6 of whether or not there is an observation target by a wide area sensor around a certain cluster (whether the number of tracks is 0).
The cluster management means 6 deletes a cluster whose information is not updated for a certain time.
When there is no observation target by a wide area sensor, the time until this deletion is shortened.

Thus, in this apparatus, it is possible to delete a cluster using information observed by a wide area sensor.
This has the advantage that the cluster can be deleted faster than when only the information of the on-board sensor is used.
If an unnecessary cluster can be deleted immediately, there is an advantage that the calculation load of the tracking processing system can be reduced.
In the fifth embodiment, the example in which the cluster deletion support means 18 is added to the first embodiment has been described. However, the same effect as that of the fifth embodiment can be obtained by adding the cluster deletion support means 18 to the apparatuses of the second to fourth embodiments. Can be issued.

Embodiment 6 FIG.
FIG. 13 is a block diagram of Embodiment 6 of the present invention, 19 is identification support means, and 1 to 14 are the same as those in FIG.
In this example, the on-board sensor 3 (12-3) has a sensor that receives the other party's radiated radio wave and outputs the target direction and identification result from the radio wave specifications. The result of the fusion with the sensor is recorded in the wake related data storage means 4.

The identification support means 19 refers to the wide area sensor information and the cluster correlation information from the wide area information correlation means 10 and the wake related data storage means 4, and associates the track information of the mounted sensor with the observation information of the wide area sensor.
By associating the radio parameters observed by the onboard sensor with the past route information of the target target, for example, by distinguishing the aircraft that flew from the airport in area A and the aircraft that flew from the airport in area B, the target (aircraft) Improve identification accuracy such as affiliation.

As described above, this apparatus has an advantage that identification of the observation target can be improved by associating the information observed by the wide area sensor with the observation information of the on-board sensor.
Observation of radio wave specifications etc. with an onboard sensor is advantageous, and observation of past target route information etc. with a wide area sensor is advantageous.
This advantageous effect is obtained by combining the advantageous information.
In the sixth embodiment, the example in which the identification support means 19 is added to the first embodiment has been described. However, even in the configuration in which the identification support means 19 is added to the apparatuses in the second to fifth embodiments, the same effect as in the sixth embodiment is produced. Can do.

Embodiment 7 FIG.
FIG. 14 is a block diagram of Embodiment 7 of the present invention, 20 is a foreseeing information utilization means, 1 to 14 are the same as FIG. 1, and 19 is the same as FIG.
The foresight information utilization means 20 holds information obtained in advance separately from the sensor observation.
With reference to this information, the wide- area sensor information from the wide- area information correlation means 10 and the correlation information of the cluster, information related to identification is sent to the identification support means 19 and information related to the number of tracks is sent to the reliability correction means 11.

For example, if there is model information (e.g., it can be limited to several types) for a target flying from a certain point (orientation), it is sent to the identification support means 19 in association with observation information from a wide area sensor.
In addition, if there is information that the probability that a target flying from a certain point (azimuth) will fly in units of four aircraft is 80% or more, the wide-area sensor is reliable for up to four targets even if two targets are observed. An instruction to improve the degree is sent to the reliability correction means 11.

  As described above, the present apparatus has an advantage that the observation target target identification and the reliability of the wake can be corrected by using the foresight information to associate the information observed by the wide area sensor with the observation information of the on-board sensor. .

It is a block diagram of Example 1 of this invention. It is a flowchart figure of the tracking processing system. It is a flowchart figure of a wide area information correlation means and a reliability correction | amendment means. It is a figure of the outline | summary of the scenario which a subject generate | occur | produces. It is a graph showing the timing when information is input from each sensor. It is the figure which showed the information (observation value) input from each sensor in time series. It is a figure of the outline | summary near time T4 when a subject generate | occur | produces. It is a block diagram of Example 2 of this invention. It is the figure which showed the outline | summary of the cluster selection of a reliability distribution means. It is a block diagram of Example 3 of this invention. It is a block diagram of Example 4 of this invention. It is a block diagram of Example 5 of this invention. It is a block diagram of Example 6 of this invention. It is a block diagram of Example 7 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Tracking processing system, 2 Observation data selection means, 3 Cluster update means, 4 Track related data storage means, 5 Correlation / integration means, 6 Cluster management means, 7 Prediction means, 8 Track determination / display means, 9 Sensor control means, DESCRIPTION OF SYMBOLS 10 Wide area information correlation means, 11 Reliability correction means, 12-1 Mounted sensor 1, 12-2 Mounted sensor 2, 12-N Mounted sensor N, 13 Display apparatus, 14 Wide area information acquisition means, 15 Reliability distribution means, 16 Wake determination support means, 17 cluster wake determination support means, 18 cluster deletion support means, 19 identification support means, 20 foresight information use means, 21 sensor control support means.

Claims (7)

Wide area information acquisition means to obtain target observation information from wide area sensors;
Wake-related data storage means for storing wake-related data in cluster units that are aggregates of hypotheses;
Cluster update means for newly creating and updating a cluster from target information from a plurality of mounted sensors and existing cluster information stored in the track-related data storage means,
Referring to the wake-related data stored in the wake-related data storage means, the cluster having the closest relative distance between the center of gravity position of the cluster and the target observation position obtained by the wide area information obtaining means is selected as the target cluster. Wide area information correlation means,
Among the hypotheses included in the target cluster selected by the wide area information correlation means, a reliability correction means for improving the reliability of the hypothesis observing the number of tracks more than the number of observations of the wide area information,
Correlation / integration means for determining the reliability of the hypothesis based on the reliability improved by the reliability correction means,
A tracking processing system comprising: Receives the relative distance between each cluster obtained by the above-mentioned wide area information correlation means and the observation value of the wide area sensor, selects a plurality of clusters corresponding to the wide area information, and distributes the number of tracks to the selected multiple clusters by weighting A hypothesis confidence distribution means to
2. The tracking processing system according to claim 1, wherein the number of tracks for each cluster whose distribution is determined by the reliability distribution means is sent to the reliability correction means. 2. The tracking processing system according to claim 1, further comprising wake determination support means for correcting the reliability calculation based on the reliability correction result of the reliability correction means. Cluster wake determination support means for determining whether or not the cluster target can be dealt with by comparing the target number captured in the cluster and the target number observed by the wide area sensor based on the reliability correction result of the reliability correction means. The tracking processing system according to any one of claims 1 to 3, further comprising: 2. A cluster deletion support unit for transmitting information on whether or not there is an observation target by a wide area sensor around a predetermined cluster obtained by the wide area information correlation unit to a cluster management unit. 5. The tracking processing system according to any one of items 4. An identification support means for associating the track information of the on-board sensor and the observation information of the wide area sensor with reference to the correlation information of the wide area sensor information and the cluster from the wide area information correlation means and the track related data storage means The tracking processing system according to any one of claims 1 to 5. Means for referring to the wide area sensor information from the wide area information correlating means and the correlation information of the cluster, and sending information relating to identification to the identification supporting means and information relating to the number of tracks to the reliability correcting means. The tracking processing system according to any one of claims 1 to 6.

Images