JP6503267B2 - Group tracking device, group tracking method and program - Google Patents

Description

  The present invention relates to a group tracking device, a group tracking method, and a program, which are suitable when a plurality of moving targets form a group.

  There is a need to monitor movement targets using sensors, such as monitoring of critical facilities and areas, monitoring of disaster areas and search or rescue support.

  Patent Document 1 discloses a method of tracking a plurality of aircraft targets flying in a group (formation) from observation data by a radar and estimating a track.

  As a track estimation method for a plurality of moving targets, a filter called MHT (Multiple Hypothesis Tracking, multiple hypothesis correlation method) is widely used, and many derivation methods thereof have been proposed. MHT is a filter that estimates the tracks of individual moving targets while maintaining multiple hypotheses about the correspondence between multiple moving targets and observation data.

  When many movement targets are close to each other, observation data may be lost because the observation data can not be associated with individual movement targets. In such a case, it is difficult to accurately estimate the track of each moving target by MHT. In Patent Document 1, there is a method of detecting that a plurality of aircraft are flying in a group, approximating the motion of each moving target with the average of the entire group and the difference from the average, and using it for MHT. It is disclosed.

  Further, Non-Patent Document 1 proposes a filter called PHD (Probability Hypothesis Density), which estimates a density distribution of moving targets. In addition, Non-Patent Document 2 proposes a filter (Extended Object and Group Tracking, hereinafter referred to as EOGT) that estimates a track by regarding a grouped group as one extended movement target.

  As described above, there are a plurality of target tracking filters having different resolutions, MHT is a filter for estimating the track of each target, PHD is a filter for estimating changes in target density distribution, and EODT is a track for the group as a whole. Patent Document 2 discloses a method for accurately estimating the track of each moving target by complementarily using filters with different resolutions.

JP 2000-505201 gazette JP, 2014-169865, A

B. Vo, S. Singh, and A. Doucet: "Sequential Monte Carlo Methods for Multi-Target Filtering with Random Finite Sets", IEEE Transactions on Aerospace and Electronic Systems, Vol. 41, No. 4, 2005 J. W. Koch: "A Bayesian Approach to Extended Object Cluster Tracking Using Random Matrices", IEEE Transactions on Aerospace and Electronic Systems, Vol. 44, No. 3, 2008

  As described above, there is a method of focusing on the group and trying to approximate the behavior of each moving target (or the monitoring target) by the average of the entire group and the difference from the average. In such a method, it is assumed that the individual moving targets behave the same as the whole group. For example, assuming that a plurality of vehicles forming a group travels in parallel along a road, the behavior of the whole group and individual moving targets are greatly different at a corner, etc., and approximations and estimations focusing on the group are accurately performed. There was a problem that it could not be done.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a group tracking device, a group tracking method, and a program for estimating a track of a group which behaves differently from the overall average.

  The present invention is a group tracking device connected to a sensor that detects an object, having a processor and a memory, the sensor information input unit receiving observation data from the sensor, and detecting a moving object from the observation data A target state estimation unit for estimating the target state of the moving object, and a target information output unit for outputting the target state estimated by the target state estimation unit, the target state estimation unit including a plurality of movements A group composed of objects is detected, the inside of the group is divided into a plurality of sub-groups, the centers of these sub-groups are calculated, and the sub-groups are connected by a broken line with the center as a bending point. Assuming that the center of the first and subsequent subgroups follows the center of the previous subgroup with a time delay, the target state of the center of the subgroup is estimated.

  According to the present invention, it is possible to represent moving objects that behave differently from the overall average by means of subgroups. Moreover, it becomes possible to express the case where the behavior of the whole group and each moving target differ greatly like the corner of a road, etc. by using a polygonal line.

It is a block diagram showing functional composition of a group tracking device in a 1st example of the present invention. It is a figure which shows the example of the behavior of the vehicle group which carries out parallel driving on the road used as the observation object of group tracking in 1st Example of this invention. It is a figure which shows the example at the time of applying the modeling (ellipse model) in a prior art to the vehicle group which carries out parallel traveling on the road shown in FIG. FIG. 5 is a diagram showing an example of the case where modeling of a group (broken line model) in the first embodiment of the present invention is applied to a vehicle group traveling in parallel on the road shown in FIG. 2; It is a figure which shows the relationship of a group, a subgroup, and a broken line in the 1st Example of this invention. It is a flowchart which shows the flow of the group tracking process performed by the control part of the group tracking apparatus in 1st Example of this invention. It is a figure which shows the 1st example of the polygonal line model used as the basis of a target state estimation part performed in the control part of the 1st Example of this invention. It is a figure which shows the 2nd example of the polygonal line model which is one Example of the target state estimation part performed in the control part of the 1st Example of this invention. It is a block diagram showing functional composition of a group tracking device in a 2nd example of the present invention. FIG. 11 is a diagram showing an example of observing a target area using a plurality of different types of sensors in the second embodiment of the present invention. FIG. 8 is a diagram showing an example of observation data when a radar is used in the second embodiment of the present invention. It is a figure which shows the example of the observation data at the time of utilizing UAV in the 2nd Example of this invention. It is a block diagram which shows the modification of the 2nd Example of this invention, and shows the function structure of a group tracking apparatus.

  According to one embodiment of the present invention, a group tracking device includes a control unit, and a receiving unit connected to the control unit, and the receiving unit receives observation data from a sensor every moment The control unit receives the observation data from the receiving unit, divides a group consisting of a plurality of moving targets (moving objects) into a plurality of sub-groups, and expresses the group as a broken line with the center of the sub-group as a bending point. , Assuming that the center of the second and subsequent subgroups of the polygonal line follows the center of the previous subgroup with a temporal delay, the target state is estimated. Do.

  In the present embodiment, the trajectory of the moving target is expressed as “track”, but any target such as a vehicle, a person or a ship may be used as long as the target targeted by the present invention is a moving object. . Further, the sensors targeted by the present invention include, for example, radar, camera (fixed type for fixed point monitoring, mobile platform mounted type, portable type, etc.), acoustic sensor, vibration sensor, counter, human (visual report, social network service registration) It may be any sensor that captures the presence or movement of a target, such as

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  A group tracking apparatus according to a first embodiment of the present invention will be described with reference to FIGS.

  FIG. 1 is a block diagram showing a functional configuration of the group tracking device 100 in the first embodiment of the present invention.

  The group tracking device 100 of the first embodiment is realized by executing a program described later on a computer such as a PC or a server. Specifically, the group tracking device 100 includes a processor 10, a memory 20, a storage 30, a reception unit 120, a display unit 130, and an instruction unit 140. The control unit 110 is loaded into the memory 20 and executed by the processor 10. The group tracking device 100 further stores a parameter storage unit 151 and a target state storage unit 152 in the storage 30.

  The control unit 110 of the first embodiment includes a sensor information (observation data) input unit 111, a target state estimation unit 112, and a target state output unit 113. These are stored in the memory 20 and included in a program executed by the processor 10. That is, the control unit 110 is realized by a computer such as a PC or a server executing a program. This computer includes the processor 10, the memory 20, the storage (external storage device) 30, the network interface and the like described above, and these components are connected to one another via a bus or the like.

  The processor 10 loads data of various programs and databases stored in the storage 30 into the memory 20 and executes the program. When the processor 10 executes a program, each function of the sensor information input unit 111, the target state estimation unit 112, and the target state output unit 113 necessary for the group tracking processing of the control unit 110 can be realized. The memory 20 stores a program to be executed by the processor 10 and data required to execute the program.

  The receiving unit 120 is an interface device that is connected to the sensor 160 and takes in observation data (Z) that is time-series data from the sensor 160 into the control unit 110. The acquisition of observational data acquired from time to time by the sensor 160 may be performed step by step or may be acquired by streaming. The sensor 160 is configured by, for example, the above-described radar or the like, and the observation data (Z) is, for example, information on a position (bearing and distance).

  The display unit 130 is a display device or the like that displays the output from the control unit 110, and is used to display the result of the group tracking process calculated by the control unit 110. The instruction unit 140 is a device such as a mouse or a keyboard that performs input to the control unit 110, and is used to input an instruction by an operator who operates the group tracking device. The display unit 130 and the instruction unit 140 may be integrally configured with input / output means as a user interface. Further, the means for inputting observation data to the group tracking device 100 is not limited to the receiving unit 120, and any means capable of providing observation data to the continuous group tracking device 100 may be used.

  The parameter storage unit 151 and the target state storage unit 152 indicate files or databases stored in the storage 30 including a storage device such as an HDD connected to the control unit 110 or data temporarily stored in the memory.

  FIG. 2 is a view showing an example of the behavior of a group of vehicles (moving targets in the figure) traveling in parallel on the road 50, which is an observation target of the group tracking device 100 of the present invention. In the drawing, T1 to T6 represent vehicles, and these vehicles T1 to T6 form a group (car train), and represent a situation in which the vehicles travel in tandem along the road. In the figure, t = k−1, t = k, and t = k + 1 indicate that time elapses at regular intervals.

  At time t = k-1 in the drawing, the leading vehicle T1 and the next vehicle T2 enter a curve. At the next time t = k, the leading vehicle T1 leaves the curve, and the next vehicle T2 and the vehicle T3 pass the curve. Furthermore, at time t = k + 1, the vehicle T4 in the middle group passes the curve, and the next vehicles T5 and T6 pass the curve.

  FIG. 3 is a diagram showing an example in which group modeling (elliptic model) in group tracking in the prior art is applied to a group of vehicles (moving targets) traveling in tandem on the road 50 shown in FIG. . In the figure, G1 represents a vehicle group corresponding to the vehicles T1 to T6 in FIG. 2, and the state of the entire group is represented by the center (position) of the group, the average velocity vector of the center, and the existing range of the vehicle (ellipse). There is.

  In the straight part of the road 50, the direction of the velocity vector is almost the same between the whole group G1 and each moving target, but at the corner, the behavior of the whole group and each moving target is largely different. There is a problem that approximations focusing on groups fail.

  That is, in FIG. 3, at time t = k-1, the velocity vector of the vehicles T4 to T6 traveling on the straight part of the road 50 is the average velocity vector at the center of the group G1, as described in the conventional example. And it is possible to approximate by the difference from the mean.

  However, the velocity vector of the vehicles T1 and T2 entering the curve (or intersection etc.) of the road 50 is estimated from the average velocity vector of the center of the group G1 at time t = k-1 shown in the prior art and the difference Is difficult because the directional components of the velocity vector are largely different.

  FIG. 4A is a diagram showing an example in which modeling of a group (broken line model) in the group tracking device 100 of the present invention is applied to a vehicle group traveling in parallel on the road 50 shown in FIG. The vehicle group corresponding to the vehicles T1 to T6 in FIG. 2 and the group G1 in FIG. 3 is divided into three sub groups g1, g2 and g3 and the state of each sub group g1 to g3 is the center of the sub group It is represented by c1 to c3, average velocity vectors V1 to V3, and an existing range (ellipse).

  When the vehicle group is expressed by broken lines L1 and L2 as shown in FIG. 4A, the behavior (velocity vector) of each moving target (vehicles T1 to T6) and the behavior of subgroups g1 to g3 corresponding to the respective vehicles T1 to T6 ( It is possible to reduce the difference between the average velocity vectors V1 to V3). Since the state of the entire group can be approximated by broken lines L1 and L2 connected with the centers c1 to c3 of the subgroups as inflection points, modeling of such a group is called a broken line model in this embodiment.

  FIG. 4B is a diagram showing the relationship between the group G1 shown in FIG. 3 and the subgroups g1 to g3 shown in FIG. 4A and broken lines.

  The broken lines L1 and L2 connecting the centers c1 to c3 of the subgroups g1 to g3 become straight lines when the train is straight and when the train passes through the curve of the road 50, the broken lines L1 and L2 bend Do. In the curve of the road 50, the center c2 of the subgroup g2 between the subgroups g1 to g3 is the inflection point.

  The broken lines L1 and L2 change to straight lines or broken lines according to the deformation of the train, but are located inside the group G1 of the conventional example regardless of the state of the train. Therefore, in the first embodiment, the broken lines L1 and L2 represent the group G1 indicating the entire vehicle train. In the first embodiment, an example in which a broken line is configured by two line segments L1 and L2 is shown, but the number of line segments changes according to the number of divisions of the sub group. Also, the number of subgroups is determined by clustering as described later.

  FIG. 5 is a flowchart showing the flow of the group tracking process executed by the control unit 110 of the group tracking device 100 according to the first embodiment. The group tracking process is a process that is sequentially performed while capturing observation data from the previous step to the current step. The process of FIG. 5 is repeatedly performed from the start to the end of the group tracking device 100.

  When the process is started, observation data from the sensor 160 is first input in the sensor information input unit 111 (step 1001). The observation data is the result of extracting the position, velocity, and the like of the movement target, but includes errors, missing data, noise, and the like. Although not clearly shown in FIG. 5, the group tracking device 100 performs adjustment such as conversion of different data format (for example, coordinate system etc.) according to the type of sensor, synchronization of observation timing, etc., if necessary. , May be converted to a data format used in the subsequent processing.

  Next, in the target state estimation unit 112, data correlation is performed to associate the observation data input in step 1001 with the target state predicted in the previous loop (step 1002). The target state estimation unit 112 performs the correlation at the group level, the correlation at the sub group level, and the correlation at the individual target (vehicle) level in stages.

  The target state estimation unit 112 determines whether the group corresponding to the observation data is present in the previously predicted target state (step 1003). The target state estimation unit 112 performs a new group detection process on observation data for which there is no corresponding group (step 1004).

  The target state estimation unit 112 further divides the detected group into a plurality of subgroups, and generates a subgroup (step 1005). The group detection in step 1004 and the generation of subgroups in step 1005 are performed based on a known or known clustering method or the like. For example, the vehicle may be divided into subgroups by a predetermined number from the leading vehicle or by a predetermined ratio to the total number.

  Next, the target state estimation unit 112 updates the target state based on the observation data (step 1006). The target state (previous prediction result) corresponding to the newly detected group has the initial value as it is. The target states corresponding to the existing groups are updated based on the observation data. Further, the target state estimation unit 112 calculates the centers c1 to c3 of the subgroups g1 to g3 and also calculates the target states (the position and the average velocity vector) of the centers c1 to c3. Note that a method of calculating the center of the group G1 of the conventional example can be used for the calculation of the centers c1 to c3 of the subgroups.

  Next, the target state output unit 113 outputs the updated target state (step 1007). The output target state is accumulated in the target state storage unit 152. The target state output unit 113 may output the updated target state to the display unit 130 for visual confirmation of the user of the group tracking device 100.

  Next, the target state estimation unit 112 determines whether the termination condition is satisfied (step 1008), and when the termination condition is not satisfied, predicts the next target state (step 1009). Return to step 1001 of and repeat the above process. The end condition is, for example, a predetermined condition such as the end of the group tracking device 100 or the acceptance of the stop of the process.

  Further, the prediction of the target state (estimation calculation) is the target state (position, speed) of each vehicle T1 to T6 after the next calculation cycle or a predetermined time, and the target state of the centers c1 to c3 of the subgroups g1 to g3. The target state estimation unit 112 estimates (position, average velocity vector). In addition, what is necessary is just to apply the technique of the patent document of the said prior art example, and a nonpatent literature about estimation operation of a target state.

  In the above example, a new group is detected for observation data in which there is no group corresponding to the previous prediction result (target state). However, the present invention is not limited to this. For example, When new observation data is detected, the subgroups may be updated. The update of this subgroup can associate new observation data with the existing subgroup or increase the number of subgroups in the group G1.

  In the above example, a new group is detected for observation data in which there is no group corresponding to the previous prediction result (target state). However, the present invention is not limited to this. For example, The subgroup may be updated when the previous target state disappears from the observation data. This sub-group update can delete existing target states from the sub-groups or reduce the number of sub-groups in group G1. In addition, when the previous target state disappears from the observation data, there is also an influence by noise etc. Therefore, when it disappears from the observation data after a predetermined time or a predetermined number of loops, the existing target state is deleted. It is also good.

  As described above, the target state estimation unit 112 detects a group which is a set of moving objects (moving targets) from observation data from the sensor 160, and further, a sub group which is a set of a plurality of moving objects in the group Divide into g1 to g3. Then, the target state estimation unit 112 calculates the center which is a representative point of the subgroups g1 to g3, estimates the target state for this center, and the center of the second and subsequent subgroups from the first subgroup is one The target state is estimated and calculated on the assumption that the center of the previous subgroup is followed with a temporal delay.

  Further, the target state estimation unit 112 calculates individual target states of the individual moving objects from the observation data from the sensor 160, and estimates the track. Then, the target state estimation unit 112 associates the target states of the subgroups with the target states of the individual moving objects by data correlation. As a result, the number of vehicles and the number of vehicles for which the centers of the plurality of subgroups follow the center of the immediately preceding subgroup with a time delay, such as a train of cars traveling on a road Tracks can be tracked accurately.

  In the target state estimation unit 112, the centers of the subgroups are connected by line segments and output to the target state output unit 113, whereby the target state estimation unit 112 can be represented by a polygonal line model with the centers of the subgroups as bending points.

  FIG. 6 is a diagram showing a first example of a broken line model, which is a basis of the target state estimation unit 112, executed in the control unit 110 of the first embodiment.

Here, x _{1, k−1 to} x _{3, k + 1} represent the state of the subgroup at each time, and x _{i, k} represents the state of the subgroup g _i at time t = k. However, in the first embodiment, an example in which i = 1 to 3 is used. Further, in the first embodiment, an example is shown in which the state x _i, k of the sub group i at time t = k is represented by the coordinates (x, y) of the center c _i of the sub group g _i and the velocity vector.

In the figure, 1 indicates the state of the subgroup g _i at time t = k, and arrows 2 and 3 indicate the propagation of the state x _{i, k} to the next time t = k + 1. This polygonal line model indicates that the state of the subgroup g _{i at} a certain time affects the next time self g _i and the next subgroup g _{i + 1} .

That is, the state x1 _{, k-1} of the subgroup g1 affects the state x2 _{, k} of the subgroup g2 of the next process, and the state x2 _{, k} of the subgroup g2 of the next process is the next It affects the state x _{3, k + 1} of the processing sub-group g ₃ . In FIG. 4A, the group G1 moving on the road 50 indicates that the next subgroup g2 passes through the point where the subgroup g1 passes, and the subgroup g3 passes through the same point.

  An example of a state equation and an observation equation based on the model of FIG. 6 will be described using the following equation. The following equations (1) to (5) show the state equation based on the model of FIG.

Formula (1) represents the state Xk of the group G1 that is the target observation target. Here, the state Xk of the group G1 is represented by a vector consisting of the states x _{i, k} of all subgroups g _i . The state x _{i, k} of the subgroup g _i is represented by the coordinates of the centers c1 to c3 of the subgroup and the velocity vector. Note that T in the equation indicates a transposed matrix.

Equation (2) is a state equation representing temporal changes in the state of the system. Here, the state Xk of the group G1 of a certain step is a first term ( _Fk X _k-1 ) depending on the state Xk of the group G1 of the immediately preceding loop and a second term by random disturbance (Gv _k ) It is assumed that it consists of Note that random disturbance is a term that represents noise or an error. Further, a to d, α, and β indicate predetermined parameters and coefficients.

Equation (3) represents the first term coefficient matrix F _k . The coefficients corresponding to arrows 2 and 3 in FIG. 6 are designed to have non-zero elements.

  Equation (4) is an identity matrix that represents the second term coefficient matrix G.

Equation (5) represents the second term random disturbance v _k . Here, it was assumed that random disturbances were normally distributed.

  The following equations (6) to (9) show observation equations based on the model of FIG.

Expression (6) represents the observation Zk of the target group G1. Here, the observations of group G1 are represented by a vector consisting of the observations of all subgroups g _i . The observation of the subgroup g _i is only the coordinates of the centers c1 to c3. In equation (6), “^” means that it is different from the previous x and y.

Formula (7) represents the observation equation of the group G1 which becomes object. Here, it is assumed that the observation of a certain loop k consists of a first term (HX _k ) depending on the state of the group G1 of the loop k, and a random disturbance term (w _k ).

  Equation (8) represents the first term coefficient matrix H.

Equation (9) represents the random disturbance w _k in paragraph. Here, it was assumed that random disturbances were normally distributed.

FIG. 7 is a diagram showing a second example of a broken line model, which is a basis of the target state estimation unit 112, executed in the control unit 110 of the first embodiment. As in FIG. 6, 1 indicates the state of the subgroup in each step, and arrow 2 indicates the propagation of the state x _{i, k} to the next step. The double-headed arrow 4 represents the constraint between subgroups g _i .

An example of the constraint conditions between the sub-groups g _i based on the model of FIG. 7 will be described using formulas.

Equation (10) represents the constraint between subgroups g _i . In equation (10), Di _{, i + 1} indicate the distance between the vehicles T1 to T6. In this restraint condition, it is indicated that the distance between the subgroups g _i of the vehicle train traveling in parallel is kept substantially constant. As a result, it is possible to impose a constraint such that the distances between the subgroups following with a temporal delay become substantially constant. Here, it is sufficient that the distance is substantially constant as long as an error in the distance between the centers of the subgroups is equal to or less than a predetermined threshold. The error of the distance is included in the random disturbance Gv _k of the above equation (2).

As described above, in the first embodiment, observation data from the sensor 160 is received every moment, the control unit 110 receives the received observation data, and detects a group consisting of a plurality of movement targets from the observation data. Then, the inside of the detected group is divided into a plurality of sub-groups, and the centers c1 to c3 of the respective sub-groups are connected by a broken line whose center is the bending point. Then, the control unit 110 delays the temporal delay of the center c _{i-1 of the} second and subsequent subgroups (g _i-1 ) of the broken line with respect to the center c _{i of the} immediately preceding subgroup g _i. The target state is estimated on the assumption that it is held and followed.

  According to the first embodiment, it is possible to express a train of vehicles (moving targets) having a behavior different from the overall average by the subgroups g1 to g3. In addition, even if the behavior of the entire group G1 and individual vehicles (moving targets) differ greatly, such as curves and intersections of the road 50, target states (positions, speeds) in the subgroups g1 to g3 are The prediction can be made accurately. That is, by estimating the target states (velocity vectors) of the vehicles T1 to T6 in the subgroups g1 to g3 with the average velocity vector for each of the subgroups g1 to g3, it is possible to prevent the difference from the average from diverging, It is possible to improve the accuracy of the prediction operation.

  Second Embodiment A group tracking device 100 according to a second embodiment of the present invention will be described with reference to FIGS. 8 to 11.

  FIG. 8 is a block diagram showing a functional configuration of the group tracking device 100 in the second embodiment of the present invention.

  The receiving unit 120 of the group tracking device 100 according to the second embodiment is connected to the plurality of sensors 160-a and 160-b to 160-n through the communication network 170, and the plurality of sensors 160-a to 160-n The observation data acquired from moment to moment is taken into the control unit 110. The configuration of the group tracking device 100 is the same as that of the first embodiment.

  FIG. 9 is a diagram illustrating an example of observing a target area Ar1 using a plurality of sensors 160-a to 160-n in the second embodiment. In the following, the whole of the sensors 160-a to 160-n is denoted by reference numeral 160. Also, the sensors 160 can be configured in different types.

  In the figure, 160-a indicates a sensor installed at a predetermined point Pt 1 of the road 50-a to detect an object flowing into (or entering into) the target area Ar 1 or an object flowing out (or leaving) from the target area Ar 1 Represent.

  When the group tracking device 100 detects a movement target passing the point Pt1 toward the road 50-b, the group tracking device 100 generates or updates the subgroup shown in the first embodiment. In the group tracking process shown in FIG. 5 of the first embodiment, the group detection in step 1004 and the subgroup generation in step 1005 may be performed based on observation data from the sensor 160-a.

  In the figure, 160-b represents a wide area monitoring sensor such as a radar installed to observe the entire target area Ar1. A sensor such as a radar can observe a wide area, but is susceptible to shielding due to topography, vegetation and the like, and it is thought that there are many missing observation data.

  In the figure, 160-n represents a mobile sensor such as a radar mounted on a UAV (Unmanned Air Vehicle) used to observe an important part (or area) Ar 2 of the target area Ar 1. The sensor 160-n mounted on the UAV is considered to have a narrower observation range than the sensor 160-a although the influence of the shielding due to the topography or the like is small.

  In the second embodiment, in the group tracking process shown in FIG. 5 of the first embodiment, observation data from the sensor 160-b and the sensor 160-n is used for data correlation in step 1002.

  The observation data from the sensor 160-b and the sensor 160-n respectively include defects, errors, and the like, so in the process of the conventional example, it is not always successful to associate observation data corresponding to the same target. Further, in the above-described conventional example, in consideration of all the possibilities with respect to the loss or error of the observation data, there arises a problem that the number of combinations of the observation data to be the target of the estimation operation explosively increases.

  However, as in the present invention, the group tracking device 100 performs data correlation on the observation data received from the sensors 160-a to 160-n in units of subgroups shown in the first embodiment, and the target state estimation unit 112 By performing the estimation operation in units of subgroups, it is possible to improve the estimation accuracy of the target state and to reduce the number of combinations of observation data for which the estimation operation is performed.

  For example, in FIG. 10, circle marks in the figure indicate moving targets detected by the sensor 160-b, and are observation data in which the group tracking device 100 estimates target states. A triangle mark in the figure indicates an example in which the group tracking device 100 estimates the target state but there is no corresponding observation data and a data loss occurs. Square marks in the figure indicate noise detected by the sensor 160-b, and are observation data that are not estimated in the previous target state of the group tracking device 100.

  Next, FIG. 11 shows an example in which a moving object (moving target) is detected by the sensor 160-n in the target area Ar2 of the ground surface from the UAV above the target area Ar1 of observation. In FIG. 11, circle marks in the figure indicate moving targets detected by the sensor 160-n, and are observation data in which the group tracking device 100 estimates the target state. A triangle mark in the figure indicates an example in which the group tracking device 100 estimates the target state but there is no corresponding observation data and a data loss occurs. The moving target detected by the sensor 160-n corresponds to the target state in which data loss has occurred in the sensor 160-b of FIG. 10.

  The group tracking apparatus 100 receives the observation data from the plurality of sensors 160 of the sensors 160-a to 160-n and executes the process shown in FIG. 5 of the first embodiment. Thereby, in the group tracking device 100, as shown in FIG. 10, even when data loss occurs in the sensor 160-b observing the entire target area Ar1, observation from 160-n observing a part of the target area Ar1 The data makes it possible to supplement the data.

  As a result, the group tracking device 100 can continue to track the moving target even if data loss occurs in part of the observation data. The group tracking device 100 may guide the UAV to the position where the data loss has occurred.

  FIG. 12 is a block diagram showing a modification of the second embodiment and showing a functional configuration of the group tracking device 100. As shown in FIG. Sensors 160-a to 160-n are connected to group tracking devices 100-a to 100-n, respectively. The group tracking devices 100-a to 100-n are connected to one another via the communication network 170 to complement the observation data from the plurality of sensors 160.

  The sensors 160-a to 160-n are the same as in FIG. 9. The group tracking devices 100-a to 100-n each have the same configuration as the group tracking device 100 of the first embodiment.

  The present invention is not limited to the embodiments described above, but includes various modifications. For example, the embodiments described above are described in detail in order to illustrate the present invention in an easy-to-understand manner, and are not necessarily limited to those having all the configurations described. Also, part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. In addition, addition, deletion, or replacement of other configurations may be applied singly or in combination with some of the configurations of the respective embodiments.

  Further, each of the configurations, functions, processing units, processing means, and the like described above may be realized by hardware, for example, by designing part or all of them with an integrated circuit. In addition, each configuration, function, and the like described above may be realized by software by a processor interpreting and executing a program that realizes each function. Information such as a program, a table, and a file for realizing each function can be placed in a memory, a hard disk, a recording device such as an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

  Further, control lines and information lines indicate what is considered to be necessary for the description, and not all control lines and information lines in the product are necessarily shown. In practice, almost all configurations may be considered to be mutually connected.

1 Subgroup state 2, 3 Subgroup state transition 4 Subgroup state restriction 10 Processor 20 Memory 30 Storage 100, 100-a, 100-b Group tracking device 110 Control part 111 Sensor information input part 112 Target state Estimation unit 113 Target state output unit 120 Reception unit 130 Display unit 140 Instruction unit 151 Parameter storage unit 152 Target state storage unit 160 Sensor 160-a, 160-b Sensor

Claims (12)

A group tracking device having a processor and a memory and connected to a sensor for detecting an object, the group tracking device comprising:
A sensor information input unit that receives observation data from the sensor;
A target state estimation unit that detects a moving object from the observation data and estimates a target state of the moving object;
A target information output unit that outputs the target state estimated by the target state estimation unit;
The target state estimation unit
A group composed of a plurality of moving objects is detected, the inside of the group is divided into a plurality of subgroups, the centers of these subgroups are calculated, and the segments are connected by a broken line with the centers of the subgroups as bending points. Estimate the target state of the center of the subgroup assuming that the center of the second and subsequent subgroups of the polygonal line follows the center of the previous subgroup with a temporal delay Group tracking device characterized by The group tracking device according to claim 1, wherein
The target state estimation unit
A group tracking device characterized in that subgroups following with a temporal delay have a constraint such that an error in distance between subgroups is less than or equal to a predetermined threshold. The group tracking device according to claim 1, wherein
The target state is
A group tracking device comprising a position of the center of the sub group and an average velocity vector of the center of the sub group. The group tracking device according to claim 3, wherein
The target state estimation unit
A group tracking device, which calculates a target state for each of the plurality of moving objects, and associates the sub group with the moving objects. A group tracking method for monitoring a moving object with a computer having a processor and a memory and connected to a sensor for detecting the moving object, comprising:
A first step of the computer receiving observation data from the sensor;
A second step of the computer detecting a moving object from the observation data and detecting a group consisting of a plurality of moving objects;
The computer divides the inside of the group into a plurality of subgroups, calculates the centers of these subgroups, connects them by a broken line with the center of the subgroup as a bending point, and the second and subsequent subgroups of the broken line A third step of estimating the target state of the center of the subgroup assuming that the center of the center follows the center of the previous subgroup with a temporal delay;
A fourth step of the computer outputting the estimated target state;
A group tracking method characterized in that it comprises: The group tracking method according to claim 5, wherein
The third step is
A group tracking method characterized in that subgroups following with a temporal delay have a constraint such that an error in distance between subgroups is less than or equal to a predetermined threshold. The group tracking method according to claim 5, wherein
The target state is
A group tracking method comprising the position of the center of the sub group and the average velocity vector of the center of the sub group. The group tracking method according to claim 7, wherein
The third step is
A group tracking method comprising: calculating a target state for each of the plurality of moving objects, and associating the sub group with the moving object. A program executed by a computer that has a processor and a memory and is connected to a sensor that detects a moving object,
A first step of receiving observation data from the sensor;
A second step of detecting a moving object from the observation data and detecting a group consisting of a plurality of moving objects;
The inside of the group is divided into a plurality of sub-groups, the centers of these sub-groups are calculated, connected by a broken line with the centers of the sub-groups as bending points, and the centers of the second and subsequent sub-groups of the broken line are one A third step of estimating a target state at the center of the subgroup assuming that the center of the previous subgroup is followed with a temporal delay;
A fourth step of outputting the estimated target state;
A program that causes the computer to execute the program. The program according to claim 9, wherein
The third step is
The program which is characterized in that the subgroups following with a temporal delay have a constraint condition such that an error in distance between subgroups is equal to or less than a predetermined threshold. The program according to claim 9, wherein
The target state is
A program comprising the position of the center of the subgroup and an average velocity vector of the center of the subgroup. The program according to claim 11, wherein
The third step is
A program comprising computing a target state for each of the plurality of moving objects, and associating the sub group with the moving object.

Images