JP2002156448A - Moving body tracking device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a moving body tracking device which can keep on tracking not only a straight moving body, but also a moving body even when the moving body moves curvedly. SOLUTION: Multiple movement paths where the moving body moves are previously set and it is assumed that the moving body on the paths; and the position and speed of the moving body at specific time are computed from the position and speed of the moving body, movement paths, and the position of an observation body and the precision of the continuation of the tracking of the moving body and a predicted position of the moving body is improved. This device is equipped with a movement path predicting unit which sets the movement paths, an identical-moving-body selector, using an assumed movement path, which decides whether or not the moving body is an identical target according to prediction information and observation information on the moving body at the time by using the set movement paths, and a selection path registration unit which registers the movement path determined by an identical-moving- body selector using the assumed movement path.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving object tracking apparatus for observing a moving object and tracking the moving object based on observation information of the moving object.

FIG. 17 is a diagram for explaining the background of the present invention. In FIG. 17, reference numeral 1 denotes an observation device of a mobile object tracking device which outputs a detection position / azimuth or the like from an unnecessary signal such as a mobile object and clutter as observation information; 46, a mobile object; 47, a moving path of the mobile object; Is a mobile object tracking device that tracks a mobile object.

In FIG. 17, when the observation apparatus 1 observes the moving body 46 at a predetermined time, and thereafter observes an object for a predetermined time near a position predicted by the motion state of the moving body 46, the observation is performed at this time. In order to determine whether the observed object is the same as the previously observed moving object, a correlation determination process is performed to determine whether the observed objects captured by the observation device 1 are the same or different. If it is determined that the captured object is the same moving object, a moving object tracking maintaining device that associates the determined moving object with the observed object, predicts its moving path, and maintains tracking is required. Was.

[0004] In particular, the present invention considers a plurality of moving paths that are considered to actually move, in order to more reliably track the moving object, and thereby makes it possible to determine the correlation of the moving object more realistically. The present invention also relates to mobile object tracking incorporating a device for performing tracking, integrating such motion information, and maintaining tracking.

[0005]

2. Description of the Related Art FIG. 18 is a block diagram showing a conventional moving object identifying apparatus, and FIG. 19 is a processing procedure showing a moving object identifying method of the conventional moving object identifying apparatus. In FIG. 18, reference numeral 1 denotes an observation device which outputs the position and orientation of a detection result from an unnecessary signal such as a moving object and clutter as observation information, and 2 converts the observation information into a coordinate system for performing a correlation process, and performs a correlation process. Observation information converter that performs pre-processing for 3 is a mobile object existence prediction range calculator that performs calculation processing of a range in which the moving object is predicted to exist, 4 is an observation information A moving body motion specification correlation determiner 5 for determining whether or not there is a correlation between the moving object and the moving body based on the correlation result from the moving body motion specification correlation determiner 4; The same moving object selector 6 selects a combination that is likely to be the same moving object. The same moving object selector 6 smoothes the position and velocity of the motion data at the current time based on the prediction information of the moving object and the observation information of the observed object. Mobile motion specification smoothing to calculate information 7 is then mobile motion specifications predictor to calculate the prediction information of the position and speed of movement specifications of the mobile body in time at which the moving object is observed.

In FIG. 19, in the conventional moving object tracking method, a smoothing value of the moving object position / velocity and a smoothing error covariance matrix such as a smoothed error covariance matrix based on the Kalman filter theory based on the observed position of the observed object in processing 19 are shown. An initial value is calculated, and the observation information of the new moving object is input in processing 20 and processing 2
In step 1, prediction information such as a prediction value of a position / velocity of the moving object at the current time and a prediction error covariance matrix obtained by estimating an error of the prediction value is calculated.

[0007] In a process 22, motion specifications such as an observation position are input as observation information, in a process 23, an existence range in which the moving object is expected to exist at the current time is calculated, and in a process 24, the observed object is moved to the moving object. In order to determine whether or not there is a relationship, the correlation determination as to whether or not the observation information falls within the presence prediction range centered on the predicted position of the moving object is performed using a prediction error covariance matrix.

[0008] With respect to the set of the observed object and the moving object determined to have a correlation in step 25, which observation object corresponds to which moving object is identified according to the degree of correlation. Using the prediction information of the moving object and the observation information of the observed object, calculate smoothing information such as a position / velocity smoothing value and a smoothing error covariance matrix based on Kalman filter theory,
This sequence is repeated until the tracking is completed in the process 27.

FIG. 20 shows a method of determining the relation between the observed object and the moving object in the process 21. In FIG. 20, reference numeral 49 denotes a smooth position of the moving object i, which is indicated by χi (+).
0 is the predicted position of the moving object i, indicated by χi (-), 51 is the observed position of the observed object i, indicated by Zi, and 52 is the predicted range of the moving object.

In the process 23, in order to determine whether or not the moving object and the observed object are related, first, a predicted existence range centered on χi (-) is created for each moving object. The predicted existence range is a prediction error range of the moving object, and indicates the existence range of the moving object at the predicted time. At this time, the prediction error covariance is used as the prediction error range. Therefore, if the observed object exists within this range, it means that the observed object may be the same moving object as the moving object that forms the existence prediction range.

Relevance means that the possibility exists. For example, as shown in FIG. 20, the observed object is within the predicted existence range of the moving object, and the observed object is related to the moving object. On the other hand, other observation objects do not exist within the predicted existence range of the moving object, and thus have no relation to the moving object. This operation is performed for all the observation objects and the moving objects, and finally, a combination of which observation objects and which moving objects are related is determined.

[0012]

In the conventional moving object tracking device, in order to determine which moving object is the observed moving object, a moving object tracked and maintained by the moving object tracking device is determined. Predictions have been made assuming a constant velocity linear motion, so if the observed moving object performs a curved motion to avoid obstacles, etc., within the existence prediction range centered on the predicted position of the moving object In some cases, the moving object may not be observed, making it difficult to keep track of the moving object.

Further, even if the curved moving body is observed within the existence prediction range and the moving body can be maintained, a tracking delay or the like occurs, and the tracking accuracy is not so good.
In some cases, tracking was maintained for the wrong observation object.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and is intended to be able to maintain a moving body not only in a straight moving body but also in a case where the moving body turns, so that the expected moving body can be maintained. Assuming a plurality of moving routes in advance or by calculation, incorporating a device that more accurately calculates the degree of correlation between the moving object and the observed object, and controlling them appropriately to improve the tracking maintenance performance of the moving object. The purpose is.

[0015]

According to a first aspect of the present invention, there is provided a mobile object tracking apparatus in which a plurality of moving paths on which a moving object moves are set in advance, and the moving object moves on the path. The purpose is to calculate the position and speed of a moving object at a predetermined time based on the position and speed of the moving object, the moving path, and the position of the observation object, to continue tracking of the moving object, and to improve the accuracy of the predicted position of the moving object. And a movement route predictor that sets the movement route, and an assumed movement that determines whether the target is the same based on the prediction information and the observation information of the moving body at the corresponding time by using the set movement route. This is achieved by providing the same moving object selector using a route and a selection route registering device that registers the moving route determined by the same moving object selector using the assumed moving route.

In the mobile object tracking device according to the second aspect of the present invention, information on the surrounding terrain where the mobile object moves and rules for the mobile object to avoid obstacles on the terrain are determined in advance, and the mobile object moves next. The purpose of the present invention is to sequentially predict the route to be taken, always correspond to the sequentially changing movement of the moving body, and to continue tracking the moving body and improve the accuracy of the predicted position of the moving body. A terrain information input device for inputting
A path selection rule setting device that sets an obstacle avoidance movement rule of the moving object, and predicts a predetermined number of paths to be moved next to the moving object based on the terrain information and the obstacle avoidance rule, and among the predetermined number, This is achieved by providing a mobile route determiner that sequentially determines one route.

According to a third aspect of the present invention, there is provided a mobile object tracking device in which topographical information on the area around which the mobile object moves and rules for the mobile object to move around obstacles on the terrain are determined in advance, and the mobile object moves next. By predicting a predetermined number of paths to be performed, by registering the plurality of movement paths, the reliability of the assumed path is calculated, and the reliability calculation result is used for identification determination between the moving object and the observation target, The purpose of the present invention is to more accurately predict the moving path of a moving object, always correspond to the sequentially changing movement of the moving object on a plurality of predicted paths, and continue tracking the moving object and improve the accuracy of the predicted position of the moving object. A terrain information input device for inputting terrain information around the movement of the moving object, a route selection rule setting device for setting an obstacle avoidance movement rule of the moving object, and a terrain information and the obstacle avoidance rule. A moving body route determiner for predicting a predetermined number of routes to be moved next to the moving body and sequentially determining one route from the predetermined number, and a predetermined number of predicted routes predicted by the moving body route determiner. A mobile route setting device to register,
The same movement using an assumed moving path for identifying a moving object and an observed object based on a moving object predicted path set by the moving object path setting device and a correlation result determined by a moving object motion specification correlation determiner. This is achieved by providing a body selector.

The moving object tracking apparatus according to the fourth aspect of the present invention calculates a correlation group in order to improve the processing speed and to make the correlation between the moving object and the observed object one-to-one, and to calculate a plurality of groups in which the moving object moves. The moving path is set in advance, and assuming that the moving object moves on the path, the position / speed of the moving object at a predetermined time is determined by the position / speed of the moving object, the moving path, and the position of the observation object. Calculate, calculate the degree of correlation between each moving object and the observed object using the position / velocity, calculate the hypothesis and hypothesis reliability of which moving object corresponds to which observed object from the degree of correlation. A moving body that calculates the presence or absence of a correlation between the moving body and the observed object, and calculates a correlation group from the presence or absence of the correlation, with the aim of continuing tracking of the moving body and improving the accuracy of the predicted position of the moving body. Exercise specification correlation A constant and correlation group determiner, a moving path predictor for setting the moving path, and using the set moving path, whether the predicted position of the moving body at the corresponding time and the observation information are the same target. Is determined for each correlation group, and a hypothesis generation and hypothesis reliability calculator using an assumed movement route for calculating a hypothesis and a hypothesis reliability between a moving object and an observed object belonging to the same correlation group for each correlation group And the same moving object selector using multiple hypotheses for selecting the same moving object from the hypotheses calculated and the hypothesis reliability for each correlation group, and the same moving object selector using the assumed moving route. This is achieved by providing a selected route register for registering a moving route.

The moving object tracking apparatus according to the fifth aspect of the present invention calculates a correlation group in order to improve the processing speed and make the correlation between the moving object and the observed object one-to-one, and calculates a correlation group for each correlation group. Hypothesis generation and calculation of hypothesis reliability of whether the moving object corresponds to the observation object with the observation object are performed, a smoothing target of the moving object is selected according to the hypothesis reliability, and the moving object moves. Preliminary information on the surrounding terrain and rules for the moving body to avoid obstacles on the terrain are determined in advance, and the route that the moving body will move next is predicted sequentially, always corresponding to the sequentially changing movement of the moving body, For the purpose of increasing the accuracy of the continuation of tracking with respect to the moving object and the predicted position of the moving object, calculate the correlation between the moving object and the observed object, and perform a moving object motion specification correlation determination and a correlation group determiner for performing grouping. , For each correlation group A hypothesis generation and hypothesis reliability calculator for performing hypothesis generation and reliability calculation between the moving object and the observed object, and selecting a hypothesis with high reliability from the hypotheses generated for each of the correlation groups, The same moving object selector using the multiple hypothesis for identifying the moving object and the observation object, the terrain information input device for inputting the terrain information around the moving object moves, and the obstacle avoiding movement rule of the moving object are set. A route selection rule setting device, and predicts a predetermined number of routes to be moved next to the moving object based on the terrain information and the obstacle avoidance rule,
This is achieved by providing a moving body route determiner that sequentially determines one route from the predetermined number.

The moving object tracking apparatus according to the sixth aspect of the present invention calculates a correlation group in order to improve the processing speed and make a one-to-one correspondence between the moving object and the observed object. Generates a hypothesis as to whether the moving object is equivalent to the observed object and calculates the reliability of the hypothesis between the object and the terrain information around the moving object and avoids obstacles on the terrain. By pre-determining the rules for movement, predicting a predetermined number of paths to which the moving object will move next, and registering the plurality of movement paths, the assumed path of each moving object included in the correlation group and the observed object By calculating the reliability and using the reliability calculation result for each correlation group for identification determination of the moving object and the observation target, the moving route of the moving object is more accurately predicted, and the movement for a plurality of predicted routes is performed. Sequential change of body Also always corresponds, aims to improve the accuracy of the predicted position of the continuation and the moving body tracking for the mobile calculates the correlation between the moving body and the observation object,
A moving body movement specification correlation determining and correlation group determining unit for performing grouping, a terrain information inputting unit for inputting terrain information around the moving body moving, and a path selection rule for setting an obstacle avoiding movement rule for the moving body A setting device, a moving body route determiner for predicting a predetermined number of routes to be moved next to the moving body based on the terrain information and the obstacle avoidance rule, and sequentially determining one route from the predetermined number; A moving body path setting unit that registers a predetermined number of predicted paths predicted by the moving body path determiner, and a moving body that is grouped and correlated by the moving body motion specification correlation determination and correlation group determination unit; For the observed object, based on the moving object predicted path set by the moving object path setting device and the correlation result determined by the moving object motion specification correlation determiner, which moving object is On the observation object A hypothesis generation and a hypothesis reliability calculator using an assumed movement route for calculating a hypothesis and a hypothesis reliability, and a hypothesis generated and calculated by the hypothesis generation and the hypothesis reliability calculator using the assumed movement route This is achieved by providing a same moving object selector using multiple hypotheses, which extracts a highly reliable hypothesis from, and identifies the moving object and the observed object.

[0021]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 1 and 2 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 1 of the present invention. In FIG. 1, reference numeral 1 denotes a predetermined observation device that outputs, as observation information, the position and orientation of a detection result from an unnecessary signal such as a moving object and clutter, 2 converts the observation information into a coordinate system for performing a correlation process, An observation information converter for performing pre-processing for correlation processing; a mobile object existence prediction range calculator for setting a range in which the mobile object is predicted to exist at the current time;
A moving body motion specification correlation determiner for determining whether or not the observation information is correlated with the prediction information of the moving body;
Is an assumed movement route setting device that sets a plurality of movement routes on which the moving object moves, 9 is a predicted movement position and observation of the moving object at the corresponding time using the set predetermined movement route and the utilization rate of the movement route. The same moving object selector using an assumed moving route that determines whether or not the information is the same target is used to register a moving route determined by the same moving object selector using a predetermined assumed moving route. A path register 6 is a moving object motion smoother for calculating smoothing information such as the position and velocity of the motion data at the current time from the prediction information of the moving object and the observation information of the observed object. This is a moving body motion specification predictor that calculates prediction information such as the position and speed of the movement specifications of the moving body at the time when the moving body is observed.

1 and 2, an assumed movement route of a moving object is set in advance (estimated movement route setting unit 8, processing 28), and observation information is input from any one of predetermined observation devices 1. Performs coordinate transformation to the coordinate system that performs correlation processing of the observation information, and based on the Kalman filter theory based on the observation information such as the position of the moving object, the smoothed value of the position and velocity of the moving object and the initial value of the smooth error covariance matrix (Moving object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is input as a new moving object (moving object existence prediction range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation determination of which observation object is included in the predicted range of the moving object (moving object specification correlator 4, processing 2)
4) is performed, and a moving object as a predetermined observation object is more likely to be generated by comparing it with assumed moving path information set in advance (the same moving object selector 9 using the assumed moving path). , Processing 30).

The reliability of the generated hypothesis is calculated in accordance with the correlation between the observed object and the moving object based on the utilization rate of the motion path and the correlation with the moving path (the same hypothesis using the assumed moving path). The moving object selector 9 and the processing 30), and registering which motion path the robot concluded to pass in the hypothesis created in the processing 30 (the selection path register 10 and the processing 31),
For the pair of the observation object and the moving object determined in that hypothesis,
The moving body smoothing information is calculated and registered for each motion model based on the Kalman filter theory (moving body motion specification smoother 6, processing 26).

In the above-mentioned hypothesis, it is registered which of the motion paths the user concluded to pass (selected path register 10, processing 27).
I do. This series of processing is repeated until the identification is completed (processing 27).

FIG. 3 shows the concept of the process 30. In FIG. 3, reference numeral 46 denotes a moving object, 53 denotes an obstacle, 54 denotes a first assumed moving path of the moving object, 55 denotes an n-th moving path of the moving object, 56 denotes an observation target position at a predetermined observation time,
57 is the position of the moving object on the first set movement path at the predetermined observation time, and 58 is the n-th position of the moving object at the predetermined observation time.
Indicates the position on the set movement route.

The assumed moving path of the moving object is given in advance by the processing 28, and it is assumed that the moving object moves substantially along any of the aforementioned assumed moving paths. Above the given location,
Based on the distance to the object observed at the predetermined observation time, calculate the degree of correlation P _{i, j} at the current observation time, and further calculate the usage rate Ψ _k of which path was used for the correlation degree up to the previous time. , The correlation β _{i, j, k} between the moving object and the route and the observed object is P
_i, given by the product of the _j and [psi _k, becomes its beta _{i, j,} observation object moving body highest of _k is considered route and the same to be moved.

Embodiment 2 4 and 5 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 2 of the present invention. In FIG. 4, 1 to 4 and 7 are the same as those in FIG. 5
Is the same moving object selector that selects a combination that is highly likely to be the same moving object between the observed object and the moving object that are correlated based on the correlation result from the moving object motion specification correlation determiner 4, Reference numeral 6 denotes a moving object motion smoother for calculating smoothing information such as the position and speed of the motion data at the current time from the prediction information of the moving object and the observation information of the observed object. A terrain information input device 12 to be input is a route selection rule setting device for setting a rule for determining a moving route of the moving object with respect to the terrain information input by the terrain information input device 11. A moving body route determiner that determines a moving route based on topographical information of the vehicle and a route selection rule from the route selection rule setting device.

Referring to FIGS. 4 and 5, topographic information around the moving object is set in advance (terrain information input unit 11, processing 32).
When the moving object approaches an obstacle on the terrain, rules for what kind of evasion action to take are set (route selection rule setting unit 12, processing 34), and the observation information is obtained from any of the predetermined observation devices. Is input, the observation information is subjected to coordinate transformation into a coordinate system for performing correlation processing, and based on the observation information such as the position of the moving object, the smoothed value and the smoothed error of the position and velocity of the moving object are shared based on the Kalman filter theory. The initial value of the variance matrix is calculated (mobile object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is input as a new moving object (moving object existence prediction range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation judgment (moving object specification correlation determiner 4, processing 24) is performed to determine which observed object is included in the predicted range of the moving object. Based on a hypothesis as to which moving object the predetermined observation object is likely to be, it is identified according to the degree (identical moving object selector 5, processing 25), and the observation object determined in the hypothesis is determined. Based on the Kalman filter theory, the moving body smoothing information is calculated and registered for each set of moving bodies based on the Kalman filter theory (moving body motion smoother 6, processing 26).

From the pre-set topographical information and the route selection rules, an expected route to which the moving object moves next is calculated (moving object route determiner 13, processing 36), and the turning acceleration of the moving object is calculated based on the expected route. Is calculated (moving object route determiner 13, processing 37). This series of processing is repeated until the identification is completed (processing 27).

FIG. 6 shows the concept of each processing in the processings 36 to 37. In FIG. 6, reference numeral 46 denotes a moving body, 59 denotes an obstacle existing in the traveling direction of the moving body,
0 indicates a moving trajectory when the moving body turns. When the moving body determines that an obstacle exists in the traveling direction, the moving body performs a turning motion to avoid the obstacle.
The turning motion involves turning acceleration. The turning acceleration that can be tolerated varies depending on the type of the moving object, and is set as one of the route selection rules in processing 34.

In the process 36, a route that can be turned with a turning acceleration equal to or less than the limit turning acceleration is searched for based on the traveling direction and speed of the moving body, the limit turning acceleration and the size of the obstacle set in the process 34, and At 37, a route close to the current altitude of the moving object is selected from the plurality of routes searched at the process 36.

In this embodiment, it is assumed that the moving body is moving at a speed V ₀ and a constant altitude in the direction of the obstacle. The limit turning acceleration at this time is α _limit , the initial position of the moving object is (x ₀ , y ₀ , z ₀ ), and the position of the moving object after time _t is (x _t , y _t ,
z _t ), and the longest point from the path of the obstacle that intersects the path at right angles to the path where the mobile body goes straight ahead is (x _b , y _b , z _b ),
Assuming that the intersecting position is (x ₁ , y ₁ , z ₁ ),

[0037]

(Equation 1)

Embodiment 3 7 and 8 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 3 of the present invention. In FIG. 7, reference numerals 1 to 4, 6 to 8, and 11 to 13 are the same as those in FIGS. Reference numeral 14 denotes a moving object route setting device that statistically totals the routes that the moving object moves on the basis of the route information determined by the moving object route determining device 13 so far.

In FIGS. 7 and 8, topographic information around the moving object is set in advance (terrain information input unit 11, processing 32).
When the moving object approaches an obstacle on the terrain, rules for what kind of evasion action to take are set (route selection rule setting unit 12, processing 34), and the observation information is obtained from any of the predetermined observation devices. Is input, the observation information is subjected to coordinate transformation into a coordinate system for performing correlation processing, and based on the observation information such as the position of the moving object, the smoothed value and the smoothed error of the position and velocity of the moving object are shared based on the Kalman filter theory. The initial value of the variance matrix is calculated (mobile object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is input as a new moving object (moving object existence prediction range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation determination of which observation object is included in the predicted range of the moving object (moving object specification correlation determining unit 4, processing 2)
4) is performed, and a moving object which is a predetermined observation object is more likely to be generated by comparing it with the set behavior pattern information (the same moving object selector 9 using the assumed moving path, processing 30). I do.

The reliability of the generated hypothesis is calculated in accordance with the strength of correlation between the observed object and the moving object based on the utilization rate of the motion path and the correlation with the moving path (the same using the assumed moving path). The moving object selector 9 and the processing 30) calculate and register moving object smoothing information for each motion model based on the Kalman filter theory for the set of the observed object and the moving object determined in the hypothesis (moving object motion information). The original smoother 6 and processing 26) are performed.

Based on the pre-set topographical information and the route selection rules, a predicted route to which the moving body will move next is calculated (moving body route determiner 13, processing 36), and the turning acceleration of the moving body is calculated based on the predicted route. Is calculated (moving object route determiner 13, processing 37), and the predicted route is compared with the action pattern information so far to create moving route statistics (moving object route setting device 14, processing 38). This series of processing is repeated until the identification is completed (processing 27).

FIG. 9 shows the concept of the processing 38. In FIG. 9, 59 is an obstacle existing in the traveling direction of the moving object, 61 is the moving path statistical result of the moving object with respect to the obstacle up to the previous time, 62 is the moving path determination result of the current moving object, and 63 is the current time. 5 shows the moving route statistical result of the moving object with respect to the obstacle.

Up to now, a predetermined moving object has performed an evasion motion on an obstacle on n paths, and γ _i (i is 1 to n) times selected on each path. And
If a route different from the previous route is calculated based on the current traveling route determination result in the process 37, the (n + 1) th route and γ _{n + 1} are set as the avoidance routes for the obstacle.
= 1 is set, and β _i is recalculated as the behavior pattern reliability for the route.

If the current traveling route judgment result in the process 37 corresponds to one of the past routes, the number of selections is added to the corresponding result, and the reliability of each pattern so far is added. β _i is recalculated and registered as action pattern information.

Embodiment 4 FIGS. 10 and 11 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 4 of the present invention. 10, 1 to 3 and 6 to 8 are the same as those in FIG. 15 determines whether or not the plurality of observation information is correlated with the prediction information of the plurality of moving objects,
The moving body motion specification correlation determination and correlation group determiner 16 that performs grouping of the observation object and the moving body based on the determination result, the 16 uses a predetermined moving path set and the usage rate of the moving path, For each group, a hypothesis generation is performed between the predicted position of the mobile unit at the corresponding time and the observation information at the corresponding time, the new target, or an unnecessary signal, and a hypothesis generation route for performing the reliability calculation of the hypothesis is used. A hypothesis generation and hypothesis reliability calculator, 17 is the same moving object selector using multiple hypotheses that perform the same moving object determination based on the hypothesis calculated for each group and the hypothesis reliability, and 10 is a multiple hypothesis. This is a selected moving route register that registers the moving route determined by the same moving object selector 17 used.

In FIG. 10 and FIG. 11, an assumed movement route of the moving object is set in advance (the assumed movement route setting unit 8, processing 2
8) If observation information is input from any of the predetermined observation devices, the observation information is subjected to coordinate transformation into a coordinate system for performing a correlation process, and a Kalman filter is performed based on the observation information such as the position of the moving body. Based on the theory, the smoothed value of the position and speed of the moving object and the initial value of the smooth error covariance matrix are calculated (moving object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is input as a new moving object (moving object existence prediction range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation judgment is made as to which observation object is included in the predicted range of the moving object (device 15, processing 40), and the correlation between the observation object and the moving object is determined. Group determination (moving object motion specification correlation determination and correlation group determination unit 15, processing 41) is performed to classify certain objects into one group.

[0051] By comparing the moving path information of a predetermined observation object which is likely to be a moving object with preset moving path information, it is determined whether the observation object is not the moving object for each group or a new moving object. Generation of a hypothesis as to whether the signal is an unnecessary signal or not and calculation of its reliability based on preset travel route information (hypothesis generation using assumed travel route and hypothesis reliability calculator 16, process 42), and processing In order to improve the speed, the hypotheses for each group generated so far are limited by the number of hypotheses or the reliability of the hypotheses, and the hypotheses are reduced (the hypothesis generation and hypothesis reliability calculator 16 using the assumed movement route, Processing 44) is performed.

Based on the above-mentioned hypothesis results, it is determined which mobile object corresponds to which observation object for each group (the same mobile object selector 17 using multiple hypotheses, processing 45), and processing 44
In the hypothesis created in step (1), it is registered which movement path the user concluded to pass (selected movement path register 10, processing 3
1) is performed, and for the set of the observed object and the moving object determined in the hypothesis, the moving object smoothing information is calculated and registered for each motion model based on the Kalman filter theory (moving object motion smoother 6, processing 26), and registers which motion path it concludes in the above-mentioned hypothesis (selected movement path register 1
0, process 27). This series of processing is repeated until the identification is completed (processing 27).

FIG. 12 shows the concept of the group judgment in the process 40. In FIG. 12, reference numeral 64 denotes the predicted position of the predetermined moving object at the predicted time, 65 denotes the observation position of the observed object at the predicted time, and 66 denotes the predicted existence range of the moving object at the predicted time.

The process 40 determines whether or not there is a correlation based on the positional relationship between the moving object and the observed object at the predicted time from the predicted existence range. If the observed object is within the predicted existence range of the moving object, there is a correlation. When it exists outside the predicted existence range, it is determined that there is no correlation, and in the determination result by the correlation determination, when a plurality of moving objects have a correlation with the same observed object, or when a plurality of observed objects have a correlation with the same moving object. If so,
Judge as the same correlation group.

In the example of FIG. 12, the observation objects 1 and 2 are present in the predicted existence range of the moving object 1, the observation objects 2 and 3 are present in the predicted existence range of the moving object 2, and the observation object 4 is present in the predicted existence range of the moving object 3. In this case, the moving bodies 1 and 2 and the observation objects 1, 2, and 3 are determined as one group, and the remaining moving body 3 and the observation object 4 are determined as another group.

In the present embodiment, after the grouping, as shown in FIG. 3, the degree of correlation between the moving object and the observation target is determined by comparison with a predetermined moving route of the moving object. After the calculation, the identification of the moving object and the observation target is determined based on the calculation result.

Embodiment 5 FIGS. 13 and 14 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 5 of the present invention. In FIG. 13, 1, 2, 3, 6, 7, 11,
Reference numerals 12 and 13 are the same as those in FIGS. 15 is a moving body motion specification correlation for determining whether or not a plurality of pieces of observation information are correlated with prediction information of a plurality of moving bodies, and performing grouping of the observed object and the moving body based on the determination result. The judgment and correlation group judgment unit 18 is the same moving object between the observed object and the moving object that are correlated for each group based on the moving object motion specification correlation judgment and the correlation result from the correlation group judgment unit 15. , A new mobile object, or an unnecessary signal, and a hypothesis generation and hypothesis reliability calculator 17 for calculating the reliability of the hypothesis and the hypothesis reliability calculated for each group. This is the same moving object selector using multiple hypotheses for determining the same moving object based on degrees.

In FIGS. 13 and 14, topographic information around the moving object is set in advance (terrain information input unit 11, processing 32).
When a mobile body approaches an obstacle on the terrain, a rule for what kind of evasion action to take is set (route selection rule setting unit 12, processing 34), and a predetermined observation device is used. When the observation information is input, the observation information is subjected to coordinate transformation into a coordinate system for performing a correlation process, and based on the observation information such as the position of the moving body, the smoothed value and the smoothed value of the position and velocity of the moving body based on the Kalman filter theory. The initial value of the error covariance matrix is calculated (mobile object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is inputted as a new moving object (moving object existence range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation determination of which observation object is included in the predicted range of the moving object (moving object specification correlation and correlation group determining unit 15, processing 40) ) And group determination of those having a correlation between the observed object and the moving body as one group (moving body motion specification correlation determining and correlation group determining unit 1)
5, processing 41) is performed.

By comparing the predetermined observation object which is likely to be a moving object with the moving path information set in advance, it is determined whether the observation object is not the moving object for each group. The generation of a hypothesis as to whether the signal is an unnecessary signal or not and the reliability thereof are calculated based on the movement path information set in advance (hypothesis generation and hypothesis reliability calculator 18, processing 42),
In order to improve the processing speed, the hypotheses generated so far for each group are limited by the number of hypotheses or the reliability of the hypotheses, and hypotheses are reduced (hypothesis generation and hypothesis reliability calculator 16, processing 44).

Based on the above-mentioned hypothesis results, it is determined for each group which mobile object corresponds to which observation object (same mobile object selector 17 using multiple hypotheses, processing 45), and it is determined within the hypothesis. Calculation and registration of moving object smoothing information for each motion model based on the Kalman filter theory for the set of the observed object and the moving object (moving object motion smoother 6, processing 26)
Then, based on the preset terrain information and route selection rules,
A predicted route to which the moving body moves next is calculated (moving body route determiner 13, processing 36), and the turning acceleration of the moving body is calculated based on the predicted path (moving body route determiner 13, processing 37). This series of processing is repeated until the identification is completed (processing 27).

In the present embodiment, as shown in FIG. 12, the moving object and the observation object are grouped according to their correlation, and the moving object and the observation object are grouped based on a hypothesis created in each group. When the obstacle is present in the traveling direction of the moving body, the obstacle avoidance course is determined by a preset limit turning acceleration of the moving body, as shown in FIG. Assuming that the moving object moves on the avoidance course, a prediction is made as to where the moving object is moving at the next predicted time.

Embodiment 6 FIG. 15 and 16 are a configuration diagram and a processing flow diagram, respectively, showing Embodiment 6 of the present invention. 15, 1, 2, 3, 6, and 7 are the same as in FIG. 1, 11, 12, and 13 are the same as in FIG. 4, 14 is the same as in FIG. 7, 15 is the same as in FIG. Table 14
By using the predetermined moving route set by the above and the usage rate of the moving route, whether the predicted position of the moving object at the corresponding time and the observation information at the corresponding time are the same target, a new target, or unnecessary A hypothesis generation and hypothesis reliability calculator using an assumed movement path for generating a hypothesis as a signal and calculating the reliability of the hypothesis, and 17 is a mobile unit based on the hypothesis and the hypothesis reliability calculated for each group. And a moving object route setting device that statistically totals the routes that the moving object moves, based on the same moving object selecting device that uses the multiple hypothesis for performing the determination.

Referring to FIGS. 15 and 16, topographic information around the moving object is set in advance (terrain information input unit 11, processing 32).
When a mobile body approaches an obstacle on the terrain, a rule for what kind of evasion action to take is set (route selection rule setting unit 12, processing 34), and a predetermined observation device is used. When the observation information is input, the observation information is subjected to coordinate transformation into a coordinate system for performing a correlation process, and based on the observation information such as the position of the moving body, the smoothed value and the smoothed value of the position and velocity of the moving body based on the Kalman filter theory. The initial value of the error covariance matrix is calculated (mobile object existence prediction range calculator 3, processing 19).

When a moving object is detected first, the observation information is input as a new moving object (moving object existence prediction range calculator 3, processing 20). The result of predicting what the moving object will be at the next observation time is calculated as prediction information (moving object motion specification predictor 7, processing 21), and the observation of the moving object captured at that time is performed. Information is input (observation device 1,
When the processing 22) is performed, a moving object predicted existence range centering on the predicted position of the moving object is calculated (moving object existence) in order to determine the relation of which observation object may correspond to which moving object. Predicted range calculator 3, process 23).

After calculating the predicted range of the moving object, a correlation judgment is made as to which observation object is included in the predicted range of the moving object of the moving object (moving object motion specification correlation judgment and correlation group judgment unit 15, processing 40). Then, a group determination (correspondence determination of moving body motion specifications and correlation group determiner 15, processing 41) is performed in which one having a correlation between the observed object and the moving body is regarded as one group.

By comparing the moving path information of a predetermined observation object which is likely to be a moving object with preset moving path information, it is determined whether the observation object is not the moving object for each group. Generation of a hypothesis as to whether the signal is an unnecessary signal or not and calculation of its reliability based on preset travel route information (hypothesis generation using assumed travel route and hypothesis reliability calculator 16, process 42), and processing In order to improve the speed, the hypotheses for each group generated so far are limited by the number of hypotheses or the reliability of the hypotheses, and the hypotheses are reduced (the hypothesis generation and hypothesis reliability calculator 16 using the assumed movement route, Processing 44) is performed.

Based on the above-mentioned hypothesis results, it is determined which mobile object corresponds to which observation object for each group (the same mobile object selector 17 using multiple hypotheses, processing 45), and is determined within the hypothesis. For the set of the observed object and the moving object, the moving object smoothing information is calculated and registered for each motion model based on the Kalman filter theory (moving object motion smoother 6, processing 26).

Based on the pre-set topographical information and the route selection rules, an expected route for the moving object to move next is calculated (moving object route determiner 13, processing 36), and the turning acceleration of the moving object is calculated based on the expected route. Is calculated (moving object route determiner 13, processing 37), and the predicted route is compared with the action pattern information so far to create moving route statistics (moving object route setting device 14, processing 38). This series of processing is repeated until the identification is completed (processing 27).

Further, in this embodiment, as shown in FIG. 12, the moving object and the observed object are grouped according to their correlation, and the hypotheses created in each group are shown in FIG. Thus, the reliability of the moving route calculated for each moving route is added to the reliability of each hypothesis, and the identification of the moving object and the observed object is determined based on the hypothesis generation and the hypothesis reliability results.

[0072]

According to the first aspect of the present invention, a moving route can be predicted in advance by providing the assumed moving route setting device, the same moving object selecting device using the assumed moving route, and the selected route registering device in the conventional device. For a moving object, it is possible to prevent deterioration of accuracy due to a delay in following the moving object, improve tracking maintenance performance, and reduce a processing load by designating a moving path in advance as a processing load. .

According to the second aspect of the present invention, by providing a terrain information input device, a route selection rule setting device, and a moving object route determining device in the conventional device, a moving route corresponding to the terrain where the moving object is moving can be determined. Calculated according to the route selection rule, the predicted route of the moving body can be analogized flexibly according to the movement of the moving body, and the following accuracy for the moving body can be improved.

According to the third aspect of the invention, the same moving object selector, terrain information input device, and terrestrial information input device using the assumed movement route are added to the conventional device.
By providing a route selection rule setting device, a moving object route determining device and a moving object route setting device, a moving route according to the terrain where the moving object is moving is calculated by the route selection rule,
The predicted route of the moving object can be inferred flexibly according to the motion of the moving object, and learning of the route is performed by taking in the route created here as past route information, and the Accordingly, it is possible to obtain a more reliable correlation between the moving object and the observation object, and as a result, it is possible to improve the route selection of the moving object and the accuracy of maintaining the tracking of the moving object.

According to the fourth aspect of the present invention, a hypothesis generation and hypothesis reliability calculator utilizing an assumed moving path, a hypothesis generation and a hypothesis reliability using an assumed moving body motion specification correlation judgment and correlation group judgment unit, a moving path setting unit, and a multiplex By providing the same moving object selecting device and the selected moving route registering device using the hypothesis, by grouping the moving object and the observed object, the amount of generated hypotheses is reduced, thereby improving the processing capacity, Also,
For a moving object whose movement route can be predicted in advance, it is possible to prevent deterioration in accuracy due to a delay in following the moving object, thereby improving the performance of tracking maintenance.

According to the fifth aspect of the present invention, the conventional apparatus is provided with a moving body motion specification correlation determination and correlation group determining unit, a hypothesis generation and a hypothesis reliability calculator, an identical moving body selector using multiple hypotheses, topographic information. By providing an input device, a route selection rule setting device and a moving object route determining device, by grouping the moving object and the observed object, the amount of generated hypotheses is reduced,
This makes it possible to improve the processing capacity, and also calculates a moving route according to the terrain where the moving object is moving by the route selection rules, and flexibly predicts the moving route of the moving object according to the motion of the moving object. By analogy, it is possible to improve the accuracy of following the moving object.

According to the sixth aspect of the present invention, there is provided a conventional apparatus in which a moving body motion specification correlation judgment and correlation group judgment unit, a hypothesis generation and hypothesis reliability calculator using an assumed movement path, and a same movement using multiple hypotheses are used. By providing a body selector, terrain information input device, route selection rule setting device, moving object route determining device, and moving object route setting device, by grouping the moving object and the observed object, the amount of generated hypotheses is reduced, This makes it possible to improve the processing capacity, and also calculates a moving route according to the terrain where the moving object is moving by the route selection rules, and flexibly predicts the moving route of the moving object according to the motion of the moving object. The route can be inferred, and the route created here is taken in as past route information to learn the route, and the moving object and the observed object are more certain according to the frequency of use of the route. Correlation it is possible to take, this result, it is possible to improve the tracking maintain accuracy for proper routing and movement of the moving body.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing Embodiment 1 of a mobile object tracking device according to the present invention.

FIG. 2 is a diagram showing a processing flow of the first embodiment.

FIG. 3 is a diagram showing a moving route selection method according to the first embodiment.

FIG. 4 is a configuration diagram showing Embodiment 2 of a mobile object tracking device according to the present invention.

FIG. 5 is a diagram illustrating a processing flow according to the second embodiment;

FIG. 6 is a diagram illustrating a moving route selection method according to a second embodiment.

FIG. 7 is a configuration diagram showing Embodiment 3 of a mobile object tracking device according to the present invention.

FIG. 8 is a diagram illustrating a processing flow according to the third embodiment;

FIG. 9 is a diagram illustrating a method of calculating the reliability of a moving route according to the third embodiment.

FIG. 10 is a configuration diagram showing a fourth embodiment of the moving object tracking device according to the present invention.

FIG. 11 is a diagram showing a processing flow of the fourth embodiment.

FIG. 12 is a diagram illustrating a method for creating a correlation group between a moving object and an observed object according to the fourth embodiment.

FIG. 13 is a configuration diagram showing Embodiment 5 of the moving object tracking device according to the present invention.

FIG. 14 is a diagram showing a processing flow of the sixth embodiment.

FIG. 15 is a configuration diagram showing a sixth embodiment of the moving object tracking device according to the present invention.

FIG. 16 is a diagram showing a processing flow of the sixth embodiment.

FIG. 17 is a diagram illustrating the background of the present invention.

FIG. 18 is a diagram showing a configuration of a conventional moving object identification device.

FIG. 19 is a diagram showing a processing flow of a conventional moving object identification device.

FIG. 20 is a diagram illustrating a method of determining a relation between an observation object and a management moving object.

[Explanation of symbols]

1 Observation device, 2 Observation information converter, 3 Moving object existence prediction range calculator, 4 Moving object motion specification correlation determiner, 5 Same moving object selector, 6 Moving object motion smoother, 7 Moving object motion Source predictor, 8 assumed moving route setting device, 9 same moving object selector using assumed moving route, 10 selected route register, 11 terrain information input device, 12 route selection rule setting device, 13 moving object route determiner, 14 mobile body path setting unit, 15 mobile body motion specification correlation determination and correlation group determination unit, 16 hypothesis generation and hypothesis reliability calculator using assumed movement path, 17 identical mobile body selector using multiple hypotheses, 18 Hypothesis generation and hypothesis reliability calculator, 46 mobile, 48 mobile tracking device.

Claims (6)

[Claims] 1. An observation apparatus for observing a moving object, an assumed movement path setting device for presetting a plurality of movement paths on which the moving object moves, and observation information for converting observation information from the observation apparatus into a correlated coordinate system. The current observation based on the converter and the smoothed parameters calculated by the moving body motion data smoother and the plurality of routes set by the assumed moving route setting unit, based on the route registered by the selected moving route register. A moving object motion predictor that calculates prediction data at a time, and a moving object presence prediction range at a predetermined time based on prediction information from the moving object motion data predictor and observation information from the observation information converter. And a moving body motion specification correlation determiner for determining whether or not an observed object can be present in the moving body presence predicted range calculated by the moving body presence predicted range calculator. And the mobile body motion specification correlation Based on the correlation determination result from the determiner and the plurality of pieces of route information registered in the travel route, the same movement using an assumed travel route that selects which route the mobile body follows and which observation object is the same. A body selector, a selection path register for registering a movement path selected by the same moving body selector using the assumed movement path, and a mobile body selected by the same moving body selector using the assumed movement path. A moving object tracking device, comprising: a moving object movement parameter smoothing device that calculates smooth parameters from each movement parameter of an observation target. 2. An observing device for observing a moving object, a terrain information input device for inputting terrain information, and a route selection rule setting for setting a rule indicating a selection criterion when a plurality of target routes are assumed. , An observation information converter for converting observation information from the observation device into a correlation coordinate system, a smoothing parameter calculated by the moving body motion specification smoothing unit, and a path determined by the moving body path determiner. A moving body motion specifier that calculates prediction data at the current observation time, and a moving body at a predetermined time based on prediction information from the moving body motion data predictor and observation information from the observation information converter. A moving body presence prediction range calculator for calculating a presence prediction range, and a moving body motion specification for performing a correlation determination as to whether an observed object can exist in the moving body presence prediction range calculated by the moving body presence prediction range calculator. A correlation determiner and the movement The same moving object selector for selecting which observation object is the same as the observed object based on the correlation determination result from the body motion specification correlation determining device, and the same moving object selector using the assumed moving path. A moving body motion parameter smoother that calculates smoothing parameters from the selected moving body and each motion parameter of the observation target; a terrain information from the terrain information input device and a smoothing from the moving body motion parameter smoother; Based on the information, predicting a future route of the moving body, and when a plurality of routes are assumed, a moving body route determiner that determines a route according to a selection rule from the route selection rule setting device. Characteristic mobile object tracking device. 3. A predetermined observation device for observing a moving object, a terrain information input device for inputting terrain information, and a route selection for setting a rule indicating a selection criterion when a plurality of target routes are assumed. A rule setter, an observation information converter for converting observation information from the observation device into a correlated coordinate system, and a path determined by a smoothing element calculated by a moving object movement element smoother and a moving object path determiner. At a predetermined time from the mobile motion specification predictor that calculates the prediction specifications at the current observation time based on the prediction information from the mobile motion specification predictor and the observation information from the observation information converter. A moving body presence prediction range calculator for calculating a moving body presence prediction range, and a moving body motion for determining whether an observed object can be present in the moving body presence prediction range calculated by the moving body presence prediction range calculator Specifications correlation determinator and before Based on the correlation judgment result from the moving object motion specification correlation judging device and the assumed route information set by the moving object route setting device, select which route the moving object follows and which observation object is the same. The same moving object selector using the assumed moving route, and the moving object motions for calculating the smoothing specifications from the respective moving items of the moving object and the observation target selected by the same moving object selector using the assumed moving route. Original smoother, based on the terrain information from the terrain information input device and the smoothing information from the moving body motion specification smoother, predict the future route of the moving body, if multiple routes are assumed, A moving object tracking device, comprising: a moving object route determining device that determines a route based on a selection rule from the route selection rule setting device. 4. An observing device for observing a moving object, an assumed moving route setting device for presetting a plurality of moving routes on which the moving object moves, and observing information for converting the observing information from the observing device into a correlated coordinate system. A converter and, among the plurality of routes set by the assumed movement route setting unit, the current observation time based on the route registered by the selected route register, out of the smoothed specifications calculated by the moving body motion specification smoother. The mobile object motion predictor that calculates the prediction parameters in, the mobile object presence prediction range at a predetermined time from the prediction information from the mobile object motion predictor and the observation information from the observation information converter A mobile object existence prediction range calculator to calculate and a correlation determination as to whether or not an observed object can be present in the mobile object existence prediction range calculated by the mobile object existence range prediction range calculator, and a plurality of mobile objects are subjected to the same observation. If there is a correlation with the object or multiple A moving body motion specification correlation determining unit and a correlation group determining unit that determine the moving body group and the observation target group to be one group when the observed object has a correlation with the same moving body; Generates a hypothesis as to which moving object corresponds to which observing object, or whether the observing object is a new moving object or an unnecessary signal, based on the correlation determination result from the detector and the plurality of route information registered in the moving route. A hypothesis generation and hypothesis reliability calculator using an assumed movement route for calculating the hypothesis reliability, and a hypothesis generation and hypothesis reliability calculator using the assumed movement route and the most reliable hypothesis from the hypothesis contents by the hypothesis reliability calculator. And selecting the same moving object selector using a multiple hypothesis that selects a combination of the moving object and a predetermined observation object for each correlation group, and the same moving object selector using the assumed moving path. Moving path parameters for calculating smoothed parameters from the moving parameters of the moving object and the observation target selected by the same moving object selector using the multiple moving hypothesis. A moving object tracking device comprising a smoothing device. 5. An observation device for observing a moving object, a terrain information input device for inputting terrain information, and a route selection rule setting for setting a rule indicating a selection criterion when a plurality of target routes are assumed. , An observation information converter for converting observation information from the observation device into a correlation coordinate system, a smoothing parameter calculated by the moving body motion specification smoothing unit, and a path determined by the moving body path determiner. A moving body motion specifier that calculates prediction data at the current observation time, and a moving body at a predetermined time based on prediction information from the moving body motion data predictor and observation information from the observation information converter. A moving body presence prediction range calculator for calculating a presence prediction range, and a correlation determination as to whether or not an observed object can be present in the moving body presence prediction range calculated by the moving body presence range prediction range calculator; The body is in phase with the same object A moving body motion specification correlation determining and correlation group determining unit that determines the moving body group and the observation target group to be one group when there is a relationship or when a plurality of observed objects are correlated with the same moving body; Based on the correlation judgment result from the body motion specification correlator, a hypothesis is generated as to which moving object corresponds to which observation object, or whether the observation object is a new moving object or an unnecessary signal, and its hypothesis reliability A hypothesis generation and hypothesis reliability calculator, which calculates the hypothesis with the highest reliability from the hypothesis contents by the hypothesis generation and hypothesis reliability calculator, and associates the combination of the moving object and the predetermined observation object with a correlation group. The same moving object selector using multiple hypotheses selected for each, and the moving object calculating smooth parameters from the moving objects selected by the same moving object selector using the multiple hypotheses and the respective motion parameters of the observation target Exercise specifications Slipper, based on the terrain information from the terrain information input device and the smoothing information from the moving body motion specification smoother, predict the future route of the moving body, if a plurality of routes are assumed, the A moving object tracking device, comprising: a moving object route determining device that determines a route based on a selection rule from a route selection rule setting device. 6. An observation device for observing a moving object, a terrain information input device for inputting terrain information, and a route selection rule setting for setting a rule indicating a selection criterion when a plurality of routes for a target move are assumed. , An observation information converter for converting observation information from the observation device into a correlation coordinate system, a smoothing parameter calculated by the moving body motion specification smoothing unit, and a path determined by the moving body path determiner. A moving body motion specifier that calculates prediction data at the current observation time, and a moving body at a predetermined time based on prediction information from the moving body motion data predictor and observation information from the observation information converter. A moving body presence prediction range calculator for calculating a presence prediction range, and a correlation determination as to whether or not an observed object can be present in the moving body presence prediction range calculated by the moving body presence range prediction range calculator; The body is in phase with the same object A moving body motion specification correlation determining and correlation group determining unit that determines the moving body group and the observation target group to be one group when there is a relationship or when a plurality of observed objects are correlated with the same moving body; Based on the correlation judgment result from the body motion specification correlation judging unit and the route information set by the moving route setting unit, which moving object corresponds to which observing object, or whether the observing object is a new moving object or an unnecessary signal A hypothesis generation and a hypothesis reliability calculator using an assumed movement route for calculating the hypothesis reliability, and a hypothesis generation and a hypothesis reliability calculation device using the assumed movement route to calculate the hypothesis reliability. The same moving object selector using a multiple hypothesis that extracts a reliable hypothesis and selects a combination of the moving object and a predetermined observation object for each correlation group, and the same moving object using the assumed moving path A moving body motion parameter smoother for calculating smoothing parameters from each of the motion parameters of the moving body and the observation target selected by the selector; a terrain information from the terrain information input device and the moving body motion parameter smoother; Based on the smoothed information from the mobile device, predicting a future route of the moving object, and, when a plurality of routes are assumed, a moving object route determiner that determines a route according to a selection rule from the route selection rule setting device. A moving object tracking device.