JP2023032184A - tracking device - Google Patents

Abstract

To obtain a tracking device for selecting a positional relation having consistency as arrangement of a true target while reducing the amount of computation in a situation in which a plurality of targets exist.SOLUTION: A tracking device includes: a target candidate position calculation part 10 for measuring a plurality of targets with a plurality of sensors and calculating positions of a plurality of target candidates including a true target and ghosts; a true target information prediction processing part 201 for calculating future position information about the true target as prediction position information; a ghost information prediction processing part 301 for calculating future position information about the ghosts as prediction position information; and a discrimination part 40 for generating a hypothesis about positions of target candidates as a combination of positions of the target candidates, and discriminating whether the target candidates are the true target or the ghost on the basis of a residual between the positions of the target candidates and the prediction position information of the true target and a residual between the positions of the target candidate and the prediction position information of the ghosts by using the hypothesis.SELECTED DRAWING: Figure 1

Description

The present disclosure relates to a tracking device that measures a plurality of tracking targets with a plurality of sensors and calculates positions and the like.

Conventionally, there is a method of measuring a tracked target with a plurality of sensors and obtaining the position, speed, etc. of the tracked target by triangulation. When this method is used, a false image, a so-called ghost, is generated as a calculation result of the position of the tracking target.

Patent Literature 1 discloses a tracking device that removes ghosts. In a situation involving two sensors and two targets, each of sensors 101, 102 measures a tracked target, and position calculator 103 calculates the positions of multiple target candidates by triangulation. The true target information prediction processing unit 201 calculates future position information regarding the true target as predicted position information based on the position information regarding the target candidate and the motion model of the true target. The ghost information prediction processing unit 301 calculates future position information about the ghost as predicted position information based on the position information about the target candidate and the motion model of the ghost. The ghost determination processing unit 404 determines that each target candidate is a true target based on the residual between the position of the target candidate and the predicted position information of the true target and the residual between the position of the target candidate and the predicted position information of the ghost. or a ghost.

Japanese Patent No. 6303254

However, in the method described in Patent Document 1, the situation in which it is possible to determine whether a target candidate is a true target or a ghost is limited to a case where the number of targets is two. On the other hand, assuming that N is an integer of 3 or more and that there are N targets with respect to two sensors, N ² pieces of position information are generated as target candidates, and among them, true target candidates are generated. There are C (N ² , N) combinations of N pieces of position information that are simply enumerated, and the combinations include inconsistent positional relationships, that is, inconsistent positions of sensors and targets. be Therefore, as the number of targets increases, the amount of calculation involved in determining whether they are true targets or ghosts becomes enormous, and as a result of the determination, the arrangement of true targets is inconsistent. There was a problem that there is a possibility to select .

The present disclosure has been made in view of the above, and provides a tracking device that selects a consistent positional relationship as a true target arrangement while reducing the amount of calculation in a situation where there are multiple targets. With the goal.

In order to solve the above-described problems and achieve the object, the tracking device of the present disclosure calculates the positions of multiple target candidates including true targets and ghosts by measuring multiple targets with multiple sensors. and a true target information prediction processing unit that calculates future position information about the true target as predicted position information based on the position information about the true target and the motion model of the true target, a ghost information prediction processing unit that calculates future position information about the ghost as predicted position information based on future position information about the true target, a motion model of the true target, and a motion model of the ghost; For a plurality of target candidate positions including ghosts, hypotheses are generated for target candidate positions as a combination of target candidate positions, and the hypotheses are used to obtain target candidate positions and true target predicted position information. and a discrimination unit that discriminates whether each target candidate is a true target or a ghost based on the residual of and the residual between the position of the target candidate and the predicted position information of the ghost. .

The tracking device of the present disclosure has the effect of selecting a consistent positional relationship as the placement of the true target while reducing the amount of calculation in a situation where there are multiple targets.

1 is a block diagram showing a configuration example of a tracking device according to an embodiment; FIG. Flowchart showing ghost determination processing in the tracking device according to the embodiment FIG. 5 is a diagram for explaining how target candidates are handled when ghost determination processing is performed by the tracking device according to the embodiment; FIG. 4 is a diagram showing an example of a configuration of a processing circuit provided in a tracking device according to an embodiment when the processing circuit is realized by a processor and a memory; The figure which shows an example of a structure of the processing circuit in the case of comprising the processing circuit with which the tracking apparatus which concerns on embodiment is provided by hardware for exclusive use.

A tracking device according to an embodiment of the present disclosure will be described in detail below with reference to the drawings. Hereinafter, for the sake of convenience, the so-called tracked target will be referred to as a "target", the actual tracked target will be referred to as a "true target", and false images, ghosts, etc. will be referred to as "ghosts". Further, a target including a true target and a ghost obtained when the target is measured by the first sensor 101 and the second sensor 102 to be described later and the position of the target is calculated is referred to as a "target candidate". Moreover, when the first sensor 101 and the second sensor 102 are not distinguished, they may simply be referred to as "sensors." Also, in relation to electronic filing, "^" attached above alphabetic characters is written as "(hat)" after the alphabetic characters. Also, hereinafter, the number of the target direction of arrival obtained by the first sensor 101 is represented by i, the number of the target direction of arrival by the second sensor 102 is represented by j, and the number of the target candidate position by each combination is represented as [i,j].

Embodiment.
[Device configuration]
FIG. 1 is a block diagram showing a configuration example of a tracking device 1 according to this embodiment. The tracking device 1 in FIG. 1 includes a candidate target position calculator 10 , a true target position information processor 20 , a ghost position information processor 30 , a determination unit 40 , and a display device 50 . A target candidate position calculation unit 10 calculates and outputs the position of each target candidate from the target measurement result. The true target position information processing unit 20 predicts and calculates future position information of the true target using the position information of the true target and the motion model of the true target, and outputs it as predicted position information of the true target. do. The ghost position information processing section 30 predicts and calculates the future position information of the ghost using the position information of the ghost and the motion model of the ghost, and outputs it as predicted position information of the ghost. The determination unit 40 determines whether each target candidate is a true target or a ghost based on the position information of each target candidate, the predicted position information of the true target, and the predicted position information of the ghost. The display device 50 displays position information of the true target.

[Outline of processing]
FIG. 2 is a flowchart showing ghost determination processing in the tracking device 1 according to the present embodiment. When the tracking process is started, the tracking device 1 measures the arrival direction of the target (step S101) and calculates the position of the target candidate (step S102). Next, the tracking device 1 generates hypotheses about a combination of positions to be evaluated as target candidates, that is, hypotheses of target position candidates, from the positions of the target candidates calculated in step S102 (step S103). In parallel, the tracking device 1 predicts and calculates future position information, that is, predicted position information of the true target from the current position information of the true target (step S201), and calculates the current position of the ghost. Future position information, ie, predicted position information of the ghost, is predicted and calculated from the information (step S301). After that, the tracking device 1 calculates the residual between the position of the target candidate and the predicted position information of the true target (step S202), and calculates the residual between the position of the target candidate and the predicted position information of the ghost (step S202). S302).

The tracking device 1 compares these residuals calculated in steps S202 and S302 (step S401), and determines whether each target candidate is a true target or a ghost (step S501). The tracking device 1 performs smoothing of the position information, that is, smoothing processing of the true target information using the target candidate determined as the true target as the current true target (step S601). Target position information is displayed (step S701). The tracking device 1 also smoothes the position information of the target candidate determined to be a ghost as the current ghost, that is, smoothes the ghost information (step S602). If the tracking device 1 continues the tracking process (step S801: Yes), the process returns to steps S101, S201, and S301. If the tracking device 1 does not continue the tracking process (step S801: No), the process of the flowchart shown in FIG. 2 ends.

The flow of the ghost determination processing of the tracking device 1 according to the present embodiment is as described above, and the detailed operation of each component of the tracking device 1 shown in FIG. 1 will be described below. Also, the steps corresponding to the processing in FIG. 2 will be described together.

[Position calculation of target candidate]
The candidate target position calculator 10 includes a first sensor 101 , a second sensor 102 , and a position calculator 103 . The first sensor 101 and the second sensor 102 receive signals by a reception system (not shown) including an antenna, and calculate the direction of arrival of the target at time _tk (step S101). The first sensor 101 and the second sensor 102 are, for example, passive sensors such as two-dimensional angle sensors. Position calculation unit 103 calculates target candidates using triangulation based on the positions of first sensor 101 and second sensor 102 and the direction of arrival of the target calculated by first sensor 101 and second sensor 102 . For each, the three-dimensional position z _[i,j],k (i=1,...,N j=1,...,N) at time _tk is calculated (step S103). Here, it is assumed that N, which is the number of targets, is an integer of 3 or more as multiple targets. The three-dimensional position z _[i,j],k of the target candidate at this time t _k is output from the position calculation unit 103 to the determination unit 40 . In this embodiment, the tracking device 1 includes two sensors, but the number of sensors is not limited to two, and may be three or more. In this way, the candidate target position calculation unit 10 calculates the positions of a plurality of target candidates including a true target and a ghost by measuring a plurality of targets with a plurality of sensors.

The determination unit 40 includes a true target information residual calculation unit 401 , a ghost information residual calculation unit 402 , a residual comparison processing unit 403 , a ghost determination processing unit 404 , and a target candidate combination hypothesis generation unit 405 . Of these, the target candidate combination hypothesis generation unit 405 receives the three-dimensional position z _[i,j],k of the target candidate at time t _k from the position calculation unit 103, and generates a consistent combination of target candidate positions. That is, hypotheses are generated for combinations of consistent positional relationships. Detailed operations of the true target information residual calculator 401, the ghost information residual calculator 402, the residual comparison processor 403, and the ghost determination processor 404 will be described later.

[Generation of Hypotheses for Combination of Target Candidate Positions]
The target candidate combination hypothesis generating unit 405 generates hypotheses about combinations of target candidate positions based on the following concept.

(405-0) Assume that the first sensor 101 and the second sensor 102 obtain N arrival direction straight lines indicating the arrival directions of N targets, respectively.
(405-1) Focusing on two of the N observed targets, the true target pair is formed by four direction-of-arrival lines C(N, 2)×C(N, 2) Each conceivable quadrangle has a diagonal relationship.
(405-2) In the quadrangle formed by 405-1, a pair located on a diagonal line different from the true target pair becomes a ghost pair.
(405-3) Assuming that the number of true targets on one line of arrival direction is one, since the true target and the ghost are located on the same line of arrival direction, the position of the ghost is depends on the true target position.
(405-4) Since the points of intersection with the straight line of the direction of arrival of the second sensor 102 existing on the straight line of the direction of arrival of the first sensor 101 are in a dependent relationship, any one of the points of intersection is the true If they are candidates for targets, the rest are candidates for ghosts.
(405-5) From 405-1, 405-2, and 405-3, the second, third, and subsequent true target candidates are They are in a diagonal relationship on the quadrangle to be formed.
(405-6) There are P(N, N-1) combinations of intersection points of arrival direction straight lines as true target candidates that satisfy the relationships from 405-1 to 405-5. For example, when the number of targets is 3, there are 9 quadrangles formed by straight lines of arrival directions, and there are 6 combinations of intersection points of straight lines of arrival directions as true target candidates.
In this way, the target candidate combination hypothesis generation unit 405 generates a combination of target candidate positions that are consistent, that is, consistent, with respect to a plurality of target candidate positions including a true target and a ghost. generate plausible hypotheses.

[Calculation of predicted value of true target]
The true target position information processing section 20 includes a true target information prediction processing section 201 and a true target information smoothing processing section 202 . Of these, the true target information smoothing processing unit 202 treats the target candidate determined as the true target by the determination unit 40 as the current true target, and smoothes the position information such as the position and speed of the true target. Along with this (step S601), the positional information of the true target is output to the display device 50. FIG. Detailed operations of the true target information smoothing processing unit 202 will be described later.

True target information prediction processing section 201 predicts future position information of the true target at time t k based on the current position information of the true target at time t _{k −1} , that is, predicts the predicted position information of the true target. Predict and calculate (step S201). Specifically, the true target information prediction processing unit 201 converts the current position information of the true target held by the true target information smoothing processing unit 202 into the position of the true target at time t _k−1 and the velocity smoothed value x( hat)[i,j], k−1|k ⁻ ₁ (i=1, . . . , N, j=1, _{. −1|k−1} , the current position and velocity smoothed values x(hat) _{[i,j], k−1|k−1} of this true target, and the motion model assuming the motion of the true target Based on this, as the future position information of the true target, the true target position at time t _k , the velocity prediction value x (hat) _{[i, j], k|k-1} , and the prediction error covariance matrix P ^{( TGT)} _{[i, j], k|k−1} are calculated and output to the determination unit 40 .

In the true target information prediction processing unit 201, the motion model of the true target is defined by Equation (1), and the predicted position information about the true target, that is, position, velocity predicted value x (hat) _{[i, j], k |k-1} , prediction error covariance matrix P ^(TGT) _{[i,j], and k|k-1} are calculated by equation (2).

In equations (1) and (2), F(·) is a function assuming true target motion, and w _k is a 6-dimensional drive noise vector for position and velocity. For example, if the true motion of the target is assumed to be uniform linear motion, equations (1) and (2) are transformed into equations (3) and (4), respectively.

In equations (3) and (4), I _{3 × 3} is a 3-row × 3-column identity matrix, O _{3 × 3} is a 3-row × 3-column zero matrix, Q _{[i,j], k-1} is the covariance matrix of the drive noise vector.

Note that the true motion defined by the function F(·) is not limited to uniform linear motion. In the above description, a single motion model based on one type of function F(·) has been described as an example, but multiple motion models based on a plurality of functions F(·) may be used. That is, the true target motion model may be a multiple motion model combining a plurality of functions F(·). The use of the multiple motion model can flexibly deal with unknown motions of the target. In this way, the true target information prediction processing unit 201 calculates future position information about the true target as predicted position information based on the position information about the true target and the motion model of the true target. The true target information prediction processing unit 201 calculates predicted position information of the true target by regarding the target candidate determined as the true target by the determination unit 40 as the true target.

[Calculation of ghost prediction value]
The ghost position information processing section 30 includes a ghost information prediction processing section 301 and a ghost information smoothing processing section 302 . Of these, the ghost information smoothing processing unit 302 treats the target candidate determined as a ghost by the determination unit 40 as the current ghost, and smoothes the position information such as the position and speed of the ghost (step S602). The detailed operation of the ghost information smoothing processor 302 will be described later.

The ghost information prediction processing unit 301 predicts the future position information of the ghost at time t k based on the current position information of the ghost at time t _{k −1} , that is, predicts and calculates the predicted position information of the ghost (step S301). Specifically, the ghost information prediction processing unit 301 obtains the current position information of the ghost held by the ghost information smoothing processing unit 302, the position of the ghost at time tk ₋₁ , the speed smoothed value y (hat) _{[i, j [} ^_ _{_ _ i, j], k-1|k-1} , and a motion model assuming the motion of the ghost, the position of the ghost at time t _k and the predicted velocity value y (hat) _{[ i, j], k|k−1} and the prediction error covariance matrix P ^(GST) _{[i, j], k|k−1} are calculated and output to the determination unit 40 .

A specific example of the ghost motion model used by the ghost information prediction processing unit 301 will be described with reference to FIG. FIG. 3 is a diagram for explaining how target candidates are handled when ghost determination processing is performed by the tracking device 1 according to the present embodiment. FIG. 3 focuses on a combination of two of the N arrival direction straight lines obtained by the first sensor 101 and the second sensor 102, respectively.

As shown in FIG. 3, assume the respective three-dimensional positions of the first sensor 101 and the second sensor 102 are the origins (0, 0, 0) and (D _x , D _y , D _z ). Three-dimensional positions of target candidates [i, j], [i, j'], [i', j] and [i', j'] observed by the first sensor 101 and the second sensor 102 respectively (x _[i,j] , y _[i,j] , z _[i,j] ), (x _[i,j'] , y _[i,j'] , z _[i,j'] ) , (x _[i',j] , y _[i',j] , z _[i',j] ) and (x _[i',j'] , y _[i',j'] , z _{[i' , j′]} ).

Now suppose target candidates [i,j] and [i',j'] are true targets and target candidates [i,j'] and [i',j] are ghosts. In this case, the three-dimensional position (x _[i,j '], y _{[i,j ']} , z _[i,j'] ) are calculated by equation (5) as intersections of straight lines connecting each target and each sensor in space.

Also, the three-dimensional position (x _[i',j] , y _[i',j] , z _[i',j] ) of the target candidate [i',j] is calculated in the same way as in Equation (5). Calculated by

A motion model of the ghost is defined by Equation (6) using the relationship of Equation (5). Thus, the motion model of the ghost utilizes the relationship between the position of the true target and the position of the ghost in the combination indicated by the hypotheses generated by the target candidate combination hypothesis generation unit 405 .

Then, the ghost information prediction processing unit 301 obtains the position, the predicted velocity value y(hat) _{[i,j'], k|k-1} , and the prediction error covariance as the predicted position information for the ghost from equation (7). Matrix P ^(GST) _{[i, j′] and k|k−1} are calculated and output to the discriminating section 40 .

Note that the ghost information prediction processing unit 301 may use future position information about the true target, a motion model of the true target, and the like. When multiple targets are tracked by multiple sensors, the ghost's motion model is defined using the fact that the ghost's position and velocity can be determined from the sensor's position and the true target's position and velocity. Since the motion model of ghosts does not make individual assumptions about sensors, targets, etc., ghosts can be eliminated in generalized situations. That is, the ghost information prediction processing unit 301 calculates the future position information of the ghost as predicted position information based on the future position information of the true target, the motion model of the true target, and the motion model of the ghost. good too. The ghost information prediction processing unit 301 calculates the predicted position information of the ghost, regarding the target candidate determined as the ghost by the determination unit 40 as the ghost.

[Residual calculation for true target]
In the determination unit 40, the true target information residual calculation unit 401 receives the three-dimensional position z _[i,j],k of the target candidate at time _tk from the target candidate combination hypothesis generation unit 405, and performs true target information prediction. Future position information of the true target at time t _k from the processing unit 201, that is, the predicted position, the predicted velocity value x(hat) _{[i,j], k|k−1} , and the prediction error covariance matrix P ^(TGT) _{[i, j], k|k-1} are input. Then, the true target information residual calculation unit 401 that has acquired these pieces _of information calculates the _residual r ^{(TGT )} Calculate _{[i, j] and k} by equation (8). Further, true target information residual calculation section 401 calculates residual covariance matrix S ^(TGT) _{[i, j], k} and residual quadratic form q ^(TGT) _{[i, j], k} at time t _k as It is calculated by the formula (9) (step S202). After that, the true target information residual calculation unit 401 calculates the calculated residual r ^(TGT) _{[i, j], k} , the residual covariance matrix S ^(TGT) _{[i, j], k} , and the residual quadratic The form q ^(TGT) _[i,j],k is output to residual comparison processing section 403 .

Note that in equation (9), R _{[i,j], k-1} is the covariance matrix of the observed noise vector. In this way, the true target information residual calculation unit 401 calculates the residual r ^{(TGT) [i, j], k of the true target from the residual r (TGT)} _{[i, j]} between the position of the target candidate and the predicted position information of the true target. Compute the quadratic form q ^(TGT) _[i,j],k .

[Residual calculation for ghosts]
The ghost information residual calculation unit 402 receives the three-dimensional position z _[i,j],k of the target candidate at time _tk from the target candidate combination hypothesis generation unit 405, and the ghost information prediction processing unit 301 outputs time _tk future position information of the ghost at , i.e. predicted position, velocity prediction y(hat) _[i,j],k|k-1 , and prediction error covariance matrix P ^(GST) _{[i,j],k|k -1} is entered. After acquiring these pieces of information, the ghost information residual calculator 402 calculates the three-dimensional position z _[i,j] of the target candidate at time _tk , the residual r ^(GST) _{[i , j] and k} are calculated by equation (10). Further, ghost information residual calculation section 402 calculates the residual covariance matrix S ^(GST) _[i,j],k and the residual quadratic form q ^(GST) _[i,j],k at time t _k by the formula (11) is calculated (step S302). After that, the ghost information residual calculation unit 402 calculates the residual r ^(GST) _{[i, j], k} , the residual covariance matrix S ^(GST) _{[i, j], k} , and the residual quadratic form q ^(GST) _{[i, j], k} is output to residual comparison processing section 403 .

In this way, _the ghost information residual calculator 402 calculates ^the residual quadratic form q ^{( GST)} Calculate _{[i,j], k} .

[Ghost Judgment]
The residual comparison processing unit 403 receives the residual r ^(TGT) _{[i, j], k} between the target candidate and the predicted position information of the true target from the true target information residual calculation unit 401, and the residual quadratic form q ^(TGT) _{[i, j], k} are input, and the residual r ^(GST) _{[i, j], k} between the target candidate and the predicted position information of the ghost from the ghost information residual calculator 402 and the residual A quadratic form q ^(GST) _[i,j],k is input. Next, the residual comparison processing unit 403 uses the residual quadratic form q ^(TGT) _{[i, j], k} calculated from the predicted position information of the target candidate and the true target to generate the target candidate combination hypothesis For all combinations of intersections of direction-of-arrival straight lines based on all hypotheses generated by generation unit 405, residual quadratic form q ^(TGT) _{[i, j] based on the motion model of the true target, sum of k} calculate. In addition, residual comparison processing section 403 uses residual quadratic form q ^(GST) _{[i, j], k} calculated from the target candidate and the predicted position information of the ghost to generate target candidate combination hypothesis generation section 405. The residual quadratic form q ^(GST) _{[i, j] based on the motion model of the ghost and the sum of k} are calculated (step S401 ). That is, the residual comparison processing unit 403 generates a residual quadratic form q ^(TGT) _{[i, j], k} , and the sum of the residual quadratic forms q ^(GST) _{[i, j], k} for ghosts corresponding to the combinations indicated by the hypotheses generated by the target candidate combination hypothesis generation unit 405 . Then, the residual comparison processing unit 403 calculates the residual quadratic form q ^(TGT) _{[i,j] based on the calculated true target motion model, the sum of k,} and the residual quadratic formula based on the ghost motion model. The sum of q ^(GST) _{[i, j] and k} is output to ghost determination processing section 404 .

Ghost judgment processing unit 404 stores residual quadratic form q ^(TGT) _{[i, j] based on the true target motion model calculated by residual comparison processing unit 403, the sum of k,} and the ghost motion model. The residual quadratic form q ^(GST) _{[i,j] based on k} is input. Next, the ghost determination processing unit 404 generates a residual quadratic form q ^(TGT) _{[i,j] based on the true target motion model, a hypothesis that minimizes the sum of k, and} residuals based on the ghost motion model. A hypothesis that maximizes the sum of differential quadratic form q ^(GST) _{[i, j] and k} is selected (step S501).

In this way, the ghost determination processing unit 404 determines the residual r ^(TGT) _{[i, j], k} between the position information of the target candidate and the predicted position information of the true target, and the position information of the target candidate and the ghost prediction From the residual r ^(GST) _[i,j],k from the position information, it is possible to determine whether each target candidate is a true target or a ghost. That is, the ghost determination processing unit 404 performs a residual quadratic form q ^(TGT) _{[i, j] based on the true target motion model calculated by the residual comparison processing unit 403, the sum of k,} and the ghost motion model It is possible to determine whether each target candidate is a true target or a ghost by selecting a combination based on the sum of the residual quadratic forms q ^(GST) _{[i,j], k} based on can. Note that the determination of whether a target candidate is a true target or a ghost is not limited to comparison using the sum of quadratic residuals.

The ghost determination processing unit 404 determines a combination of intersections of straight lines of arrival directions based on the selected hypothesis as a true target, and outputs the residual associated with the hypothesis to the true target information smoothing processing unit 202 . Further, the ghost judgment processing unit 404 judges, of the intersections of the arrival direction straight lines, the intersections other than the intersections based on the selected hypothesis to be ghosts, and outputs the residual associated with these intersections to the ghost information smoothing processing unit 302. . Specifically, the ghost determination processing unit 404 uses information about the residual between the position information of the target candidate and the predicted position information of the true target corresponding to the number of the target candidate determined to be the true target. A certain residual r ^(TGT) _{[i, j], k} and a residual covariance matrix S ^(TGT) _{[i, j], k} are output to true target information smoothing section 202 . In addition, ghost determination processing section 404 performs residual r ^(GST) , which is information about the residual between the position information of the target candidate and the predicted position information of the ghost, corresponding to the number of the target candidate determined to be a ghost. _{[i, j], k} and residual covariance matrix S ^(GST) _{[i, j], k} are output to ghost information smoothing processing section 302 .

[True target smoothing]
True target information smoothing processing section 202 applies residual r ^(TGT) _{[i, j], k} and residual covariance matrix S ^(TGT) _{[i, j], k} input from ghost judgment processing section 404 to , according to the Kalman filter algorithm, the current position information of the true target by equation (12), i.e. position, velocity smoothed value x(hat) _{[i,j], k|k} and smoothed error covariance matrix P ^(TGT) _{[ i, j] and k|k} are calculated (step S601). Furthermore, the true target information smoothing unit 202 processes the true target position, velocity smoothed value x (hat) _{[i, j], k|k} , smoothed error covariance matrix P ^(TGT) _{[i, j], k} Position information such as _|k is output to the true target information prediction processing unit 201 and the display device 50 .

[Ghost smoothing]
Based on the residual r ^(GST) _{[i, j], k} and the residual covariance matrix S ^(GST) _{[i, j], k} input from the ghost determination processing unit 404, the ghost information smoothing unit 302 performs , according to the Kalman filter algorithm, the ghost's current position information, i.e. position, velocity smoothed value y(hat) _{[i,j], k|k} and smoothed error covariance matrix P ^(GST) _{[i,j ] and k|k} are calculated (step S602). Furthermore, the ghost information smoothing processing unit 302 processes the position of the ghost, the velocity smoothed value y (hat) _{[i, j], k|k} , the smoothed error covariance matrix P ^(GST) _{[i, j], k|k,} etc. position information is output to the ghost information prediction processing unit 301 .

The display device 50 displays the true target position, velocity smoothed value x(hat) _{[i,j], k|k,} etc. input from the true target information smoothing unit 202 (step S701). In this embodiment, the display device 50 is configured to display the true target position, velocity smoothed value x (hat) _{[i, j], k|k,} etc. input from the true target information smoothing unit 202. However, it is also possible to input from the ghost information smoothing processing unit 302 to the display device 50 and display the position where the display device 50 is determined to be a ghost with an alarm to the effect that it is a ghost. good.

As described above, according to the present embodiment, the tracking device 1 detects a true target or a ghost in a general situation where there are a plurality of targets for a plurality of sensors, for example, three or more targets. It is possible to select a consistent positional relationship as the true target arrangement while reducing the amount of calculation involved in the determination.

Next, the hardware configuration of the tracking device 1 according to this embodiment will be described. In the tracking device 1, the first sensor 101 and the second sensor 102 are passive sensors such as two-dimensional angle sensors, as described above. The display device 50 is a display such as an LCD (Liquid Crystal Display). Other configurations of the tracking device 1 are implemented by processing circuitry. The processing circuit may be a memory that stores a program and a processor that executes the program stored in the memory, or may be dedicated hardware. Processing circuitry is also called control circuitry.

FIG. 4 is a diagram showing an example of the configuration of the processing circuit 90 when the processing circuit included in the tracking device 1 according to this embodiment is realized by the processor 91 and the memory 92. As shown in FIG. A processing circuit 90 shown in FIG. 4 is a control circuit and includes a processor 91 and a memory 92 . When the processing circuit 90 is composed of the processor 91 and the memory 92, each function of the processing circuit 90 is implemented by software, firmware, or a combination of software and firmware. Software or firmware is written as a program and stored in memory 92 . In the processing circuit 90, each function is realized by the processor 91 reading and executing the program stored in the memory 92. FIG. That is, the processing circuitry 90 includes a memory 92 for storing programs that result in the processing of the tracking device 1 being executed. This program can also be said to be a program for causing the tracking device 1 to execute each function realized by the processing circuit 90 . This program may be provided by a storage medium storing the program, or may be provided by other means such as a communication medium.

Here, the processor 91 is, for example, a CPU (Central Processing Unit), a processing device, an arithmetic device, a microprocessor, a microcomputer, or a DSP (Digital Signal Processor). The memory 92 is a non-volatile or volatile memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable ROM), EEPROM (registered trademark) (Electrically EPROM). A semiconductor memory, a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD (Digital Versatile Disc) is applicable.

FIG. 5 is a diagram showing an example of the configuration of the processing circuit 93 when the processing circuit included in the tracking device 1 according to the present embodiment is composed of dedicated hardware. The processing circuit 93 shown in FIG. 5 is, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof. thing applies. The processing circuit 93 may be partially realized by dedicated hardware and partially realized by software or firmware. Thus, the processing circuitry 93 can implement each of the functions described above by dedicated hardware, software, firmware, or a combination thereof.

The configurations shown in the above embodiments are only examples, and can be combined with other known techniques, or can be combined with other embodiments, without departing from the scope of the invention. It is also possible to omit or change part of the configuration.

1 tracking device 10 target candidate position calculator 101 first sensor 102 second sensor 103 position calculator 20 true target position information processor 201 true target information prediction processor 202 true target information smoothing process Section 30 Ghost position information processing section 301 Ghost information prediction processing section 302 Ghost information smoothing processing section 40 Discrimination section 401 True target information residual calculation section 402 Ghost information residual calculation section 403 Residual comparison processing section , 404 ghost determination processing unit, 405 target candidate combination hypothesis generation unit, 50 display device.

Claims (6)

a target candidate position calculation unit that calculates the positions of a plurality of target candidates including a true target and a ghost by measuring a plurality of targets with a plurality of sensors;
a true target information prediction processing unit that calculates future position information of the true target as predicted position information based on the position information of the true target and the motion model of the true target;
a ghost information prediction processing unit that calculates future position information about the ghost as predicted position information based on future position information about the true target, a motion model of the true target, and a motion model of the ghost;
generating hypotheses about the positions of the target candidates as a combination of the positions of the target candidates for the positions of the plurality of target candidates including the true target and the ghost; and the predicted position information of the true target and the residual between the position of the target candidate and the predicted position information of the ghost, each said target candidate is a true target or a ghost a determination unit that determines whether
A tracking device comprising: The determination unit generates the hypothesis consistent with the positions of the target candidates as a combination of the positions of the target candidates with respect to the positions of the plurality of target candidates including the true target and the ghost. ,
The tracking device according to claim 1, characterized by: The true target information prediction processing unit calculates predicted position information of the true target using the target candidate determined as the true target by the determination unit as a true target,
The ghost information prediction processing unit calculates predicted position information of the ghost, with the target candidate determined as a ghost by the determination unit as a ghost.
3. The tracking device according to claim 1 or 2, characterized in that: The determination unit is
calculating a residual quadratic form for the true target from the residual between the position of the target candidate and the predicted position information of the true target;
calculating a residual quadratic form for the ghost from the residual between the position of the target candidate and the predicted position information of the ghost;
calculating the sum of the residual quadratic forms for the true target corresponding to the hypothesized combination and the sum of the residual quadratic forms for the ghost corresponding to the hypothesized combination;
determining whether each target candidate is a true target or a ghost by selecting the combination based on the calculated sum;
The tracking device according to any one of claims 1 to 3, characterized in that: The ghost motion model utilizes the relationship between the true target position and the ghost position in the combination shown in the hypothesis.
The tracking device according to any one of claims 1 to 4, characterized in that: The true target motion model is a multiple motion model combining multiple functions,
The tracking device according to any one of claims 1 to 5, characterized in that:

Images