JP5319207B2 - Compound sensor device - Google Patents

Description

  The present invention relates to a composite sensor device that acquires target information using a passive sensor that observes a target without emitting electromagnetic waves or sound waves and an active sensor that emits electromagnetic waves or sound waves to observe a target.

  2. Description of the Related Art Conventionally, an apparatus that observes a target using a passive sensor that does not emit electromagnetic waves or sound waves and an active sensor that emits electromagnetic waves or sound waves is known (see, for example, Patent Document 1). FIG. 6 is a block diagram showing an example of the configuration of a conventional composite sensor device that performs tracking of a target using target observation values of such passive sensors and active sensors. A conventional composite sensor device illustrated in FIG. 6 includes a passive sensor 61 that observes a target angle without emitting electromagnetic waves or sound waves, an active sensor 62 that observes a target distance and angle by emitting electromagnetic waves or sound waves, An angle tracking processing unit 63 that calculates a smooth value and a predicted value of a target angle using a tracking filter based on an observation value from the passive sensor 61, and a target distance using a tracking filter based on an observation value from the active sensor 62. And the distance / angle tracking processing unit 64 for calculating the angle smoothing value and the predicted value, and the processing results from the angle tracking processing unit 63 and the distance / angle tracking processing unit 64 are data-fused to achieve target smoothing. Value and predicted value, and schedule the observation timing of the passive sensor 61 and the active sensor 62 And a fused-controller 65.

  Next, the operation of the conventional composite sensor device illustrated in FIG. 6 will be described with reference to the flowchart of FIG. First, sensor selection control at the observation timing is performed by scheduling by the data fusion / control unit 65 (ST701). Here, when the passive sensor 61 is selected, the target is observed by the passive sensor 61 based on the control signal from the data fusion / control unit 65 (ST702), and the angle measurement data as the observation value is further obtained. Based on the processing using the tracking filter, the angle tracking processing unit 63 calculates a smooth value and a predicted value of the target angle, and sends them to the data fusion / control unit 65 (ST703). On the other hand, when the active sensor 62 is selected in the operation step of ST701, the target is observed by the active sensor 62 based on the control signal from the data fusion / control unit 65 (ST704), and further, the distance measurement as the observed value is performed. Based on the angle measurement data, the distance / angle tracking processing unit 64 calculates a smooth value and a predicted value of the target distance and angle by processing using a tracking filter, and sends them to the data fusion / control unit 65 ( ST705).

Next, in the data fusion / control unit 65, the smooth values from the angle tracking processing unit 63 and the distance / angle tracking processing unit 64 acquired at each observation timing are data-fused, and the target smooth value and predicted value are calculated. (ST706). Next, the sensor selection is scheduled at subsequent observation timings, and each sensor is notified (ST707). Then, the above-described series of operation steps is repeated until an operation end is instructed (ST708). The tracking filter and the like are described in detail in Non-Patent Document 1, for example.
Japanese Patent Laid-Open No. 7-128436 (page 4, FIG. 1) “Revised radar technology” supervised by Takashi Yoshida, IEICE, October 1, 1996, p. 264-267

  As described above, in the composite sensor device including the passive sensor and the active sensor, the sensor observation value is filtered by the tracking filter, and target information such as a target smooth value and a predicted value is acquired. Here, in order to reduce the detectability of the device, while continuously observing the target angle information with a passive sensor, to minimize the use of an active sensor that can observe the target angle and distance information, The observation timing of each sensor is scheduled.

  However, if the target moves with a sudden change, for example, different from the tracking filter motion model, the error in the target information after the filtering process may increase, which is not always sufficient. Tracking performance could not be obtained. In addition, if the time interval for using the active sensor is narrowed to improve the tracking performance, the detectability deteriorates, the power consumption required for radiation of electromagnetic waves and the like increases, and the accompanying device cooling and power supply There was a problem that the apparatus was increased in size.

  The present invention has been made in consideration of the above-described circumstances, and an object thereof is to provide a composite sensor device that is improved in tracking performance and reduced in size while reducing power consumption while reducing detectability. And

In order to achieve the above object, the composite sensor device of the present invention includes a passive sensor that observes a target angle without emission of electromagnetic waves or sound waves, a distance to the target by emitting electromagnetic waves or sound waves, or the target. An active sensor for observing the distance and angle to the target, and a tracking filter configured based on a predetermined motion model, and for the target track information held in the tracking filter, the target from the passive sensor Smoothing processing is performed regardless of whether the observed value of the target or the observed value of the target from the active sensor is input, and the smoothed value and smoothing error covariance matrix of the target and the predicted value and predictive error covariance matrix by calculating the target information including a tracking means for data fusion of observations from these two sensors with observations based on this tracking means Detecting the residual and motion of the target different from the predetermined motion model based on the residual covariance matrix of the observed and predicted values of the passive sensor and the active sensor when calculating the target information as maneuver of the target And a maneuver detecting means for classifying the detected state of the maneuver into a plurality of maneuver stage values associated with a plurality of preset threshold values, and the tracking filter based on the maneuver stage value from the maneuver detecting means. A system noise control means for controlling the level of system noise, and a smoothing error covariance matrix, a prediction error covariance matrix from the tracking means, and a maneuver detection means while continuing to observe the target by the passive sensor Based on at least one of the maneuver step values, the passive sensor and And having a sensor control unit for performing scheduling of observation timing of the active sensor.

  Further, the sensor control means continues to observe the target by the passive sensor, and an error covariance component calculated from the target smooth error covariance matrix or prediction error covariance matrix from the tracking means is predetermined. And when the target maneuver step value from the maneuver detection means is changed, scheduling is performed so that the target is observed by the active sensor. To do.

  Further, the sensor control means is based on the observation time interval of the active sensor set in advance in association with the target maneuver step value from the maneuver detection means while continuing to observe the target by the passive sensor. And scheduling the target to be observed by the active sensor.

  Further, the sensor control means controls the radiation direction of the electromagnetic wave or sound wave of the active sensor based on the target information from the tracking means, and the active sensor detects the electromagnetic wave or sound wave in the designated direction based on this control. It is characterized by radiating.

  ADVANTAGE OF THE INVENTION According to this invention, the small and low power consumption composite sensor apparatus which improved tracking performance while reducing to-be-detected property can be obtained.

  The best mode for carrying out the composite sensor device according to the present invention will be described below with reference to FIGS.

  FIG. 1 is a block diagram showing the configuration of a first embodiment of a composite sensor device according to the present invention. The composite sensor device illustrated in FIG. 1 includes a passive sensor 11, an active sensor 12, a data fusion / tracking processing unit 13, a maneuver detection unit 14, and a control unit 15. The passive sensor 11 includes, for example, an imaging sensor, a radio wave reception sensor, and the like, observes a target angle at an observation timing by control from the control unit 15 described below without accompanying electromagnetic wave or sound wave emission, and obtains the result as data. The data is sent to the fusion / tracking processing unit 13. The active sensor 12 includes, for example, a laser sensor, a radar sensor, and the like, emits electromagnetic waves or sound waves at the observation timing controlled by the control unit 15, observes the distance and angle to the target, and integrates and tracks the result. The data is sent to the processing unit 13.

The data fusion / tracking processing unit 13 has a tracking filter configured based on a predetermined motion model. The observation result from the passive sensor 11 and the active sensor 12 is filtered using the tracking filter, and the target Target information including a smooth value and a smooth error covariance matrix, and a target prediction value and a prediction error covariance matrix. In the present embodiment, the tracking filter is, for example, a linear Kalman filter, and the motion model thereof is, for example, constant velocity linear motion. Further, based on a maneuver step value from a maneuver detection unit 14 to be described later, the system noise level incorporated as an uncertain (error) component of the target motion in the tracking filter described above conforms to the target motion model. Control to do.

  The maneuver detection unit 14 detects a motion of a target different from the motion model of the tracking filter based on the residual between the target information and the observation value calculated by the data fusion / tracking processing unit 13 and the residual covariance matrix. The target maneuver is classified into maneuver step values associated with a plurality of preset threshold values and output. In this embodiment, the motion model of the tracking filter is a constant-velocity linear motion. Therefore, if the target maneuver to be detected is, for example, an aircraft, the motion is different from the motion model, that is, ascending, descending, and turning. Are classified according to their speed and output as maneuver step values.

  The control unit 15 schedules observation timings of the passive sensor 11 and the active sensor 12 based on at least one of the target information from the data fusion / tracking processing unit 13 and the maneuver step value from the maneuver detection unit 14. In the present embodiment, scheduling is performed while minimizing the use of the active sensor 12 while continuing to observe the target angle by the passive sensor 11 so as to reduce the detectability of the apparatus. In this embodiment, the error covariance component calculated from the target smoothing error covariance matrix or the prediction error covariance matrix calculated by the data fusion / tracking processing unit 13 exceeds a predetermined threshold value. And at least one of the cases where the target maneuver step value from the maneuver detector 14 changes, the active sensor 12 is scheduled to operate. That is, for the passive sensor 11, for example, when at least one of the above-described conditions occurs while scheduling to perform observation continuously at equal time intervals, the active sensor 12 is set at the immediately following observation timing. Schedule each sensor to perform the observations used.

  Next, the operation of the composite sensor device of the present embodiment configured as described above will be described with reference to the flowcharts of FIG. 1 and FIG. FIG. 2 is a flowchart for explaining the operation of the first embodiment of the composite sensor device according to the present invention illustrated in FIG.

  First, the overall operation flow will be described with reference to FIG. First, a sensor is selected at the observation timing by scheduling by the control unit 15 (ST201). Here, when the passive sensor 11 is selected, the target angle is observed by the passive sensor 11 according to the control from the control unit 15, and the observation result is sent to the data fusion / tracking processing unit 13 (ST202). . The data fusion / tracking processing unit 13 performs a filtering process using a tracking filter based on the observation result, and calculates a target smooth value and a smooth error covariance matrix (ST203). On the other hand, when the active sensor 12 is selected in the operation step of ST201, the target distance and angle are observed by the active sensor 12 according to the control from the control unit 15, and the observation result is sent to the data fusion / tracking processing unit 13. It is sent out (ST204). The data fusion / tracking processing unit 13 calculates a target smooth value and smoothing error covariance matrix by filtering processing based on these observation results (ST205).

  Next, the maneuver detection unit 14 detects the target maneuver based on the residual between the target information and the observed value calculated by the data fusion / tracking processing unit 13 and the residual covariance matrix. The detection result is output as a maneuver step value divided into a plurality of steps according to the maneuver level (ST206). Next, based on the maneuver step value, the data fusion / tracking processing unit 13 controls the internal tracking filter so that the system noise level is more suitable for the target motion model, and the filtering process is performed. Continuing, a target predicted value and a prediction error covariance matrix are calculated (ST207). Next, in the control unit 15, the error covariance component calculated from the target smoothing error covariance matrix or the prediction error covariance matrix calculated in the data fusion / tracking processing unit 13 in this way is a predetermined threshold value. Each of the sensors is scheduled at the subsequent observation timing so that the active sensor 12 is operated in at least one case when the target maneuver step value from the maneuver detecting unit 14 changes. FIG. 3 shows an example of this. In the example of FIG. 3, the active sensor 12 is scheduled to operate when the maneuver step value from the maneuver detection unit 14 changes, and the observation by the passive sensor 11 is performed at the observation timing set at equal time intervals. While continuing, when the maneuver step value from the maneuver detection unit 14 changes, the observation is performed by operating the active sensor 12 at the observation timing immediately thereafter. Even when the error covariance component exceeds the threshold value, the active sensor 12 is controlled to operate at the observation timing immediately after that, as in this case (ST208). Thereafter, the above-described series of operation steps is repeated until the operation ends (ST209).

  Next, the processing contents in each operation step described above will be described in detail. First, in the filtering process in the steps ST203 and ST205, the processing contents when the observation results from the passive sensor 11 for observing the target angle and the active sensor 12 for observing the distance and angle are fused on the basis of the observation value. explain.

The target motion model is defined as the following equations (1) to (5).

Here, x _k is a distance r _{k at the} observation time t _k, a state vector composed of the azimuth angle θ _k and its velocity component, and F _k and G _k are a transition matrix from the observation time t _k to the observation time t _{k + 1} . The drive matrix, w _k is the mean 0 of the observation time t _k , the system noise vector of the covariance matrix Q _k , and σ ^s _k is the standard deviation of the system noise at the observation time t _k . Also, ^{A T} is the transpose of vector or matrix A, _{I n} is a unit matrix of n rows and n columns, _{O n} denotes a zero matrix of n rows and n columns.

The observation model of the passive sensor 11 is defined as the following formulas (6) to (8).

^{Here, y} _{p k} is the observation ^{vector, H} _{p k} of the passive sensor 11 of the measurement time _{t k} is the observation observation matrix at time _{t k} of the passive sensor ^{11, v} _{p k} is the observation time _{t k} of the passive sensor 11 average 0, observation noise vector covariance matrix R ^{p _k,} σ θ _k is the standard deviation of the azimuthal component of the observation noise in the measurement time t _k.

The observation model of the active sensor 12 is defined as the following equations (9) to (11).

Here, y a k is the observation vector of active sensor 12 of the measurement time t k, H a k is the observation matrix of active sensor 12 of the measurement time t k, v a k is active sensor 12 of the measurement time t k , The observation noise vector of the covariance matrix R a k , and σ r k are standard deviations of the distance components of the observation noise at the observation time t k . When the active sensor 12 outputs only the distance measurement data instead of the distance measurement / angle measurement data, the observation matrix H a k and the covariance matrix R a k of the active sensor 12 are respectively expressed by the following formulas (12 ) And formula (13).

In the operation step of ST203, smoothing processing expressed by the following equations (14) to (18) is performed using the angle measurement data from the passive sensor 11.

Here, (y tilde) ^p _k is the residual vector of the passive sensor 11 at the observation time t _k , S ^p _k is the residual covariance matrix of the passive sensor at the observation time t _k , and K ^p _k is the observation time Kalman gain matrix of the passive sensor 11 at t _k , (x hat) _{k | k} is a smooth vector at observation time t _k , P _{k | k} is a smooth error covariance matrix at observation time t _k , and (x hat) _{k | k-1,} the prediction vector of the measurement time _{t k, P k | k-} 1 is the prediction error covariance matrix of the observation time _{t k.} A ^-1 indicates an inverse matrix of the matrix A.

In ST205, smoothing processing expressed by the following equations (19) to (23) is performed using the distance measurement / angle measurement data from the active sensor 12.

Here, (y tilde) ^a _k is the residual vector of the active sensor 12 at the observation time t _k , S ^a _k is the residual covariance matrix of the active sensor 12 at the observation time t _k , and K ^a _k is the observation it is a Kalman gain matrix of the active sensor 12 of time t _k.

Next, the contents of processing when maneuver is detected in the operation step of ST206 will be described. In this operation step, the maneuver state (maneuver presence / absence) is detected by the following equations (24) to (25), and the maneuver value is classified.

Here, epsilon _k, normalized residual observation time _{t k,} (y ^{tilde) p /} _{a k} is the residual vector of the passive sensor 11 or active sensor 12 of the measurement time _{t k,} f1 (·) is functions that extract azimuthal component of the residual vector of the passive sensor 11 or active sensors ^{12, S p /} _{a k} is the residual covariance matrix of the passive sensor 11 or active sensor 12 of the measurement time _{t k,} f2 (·) Is a function for extracting the azimuth angle component of the residual covariance matrix of the passive sensor 11 or the active sensor 12, m _k is a maneuver step value (the presence or absence of a maneuver) at the observation time t _k , and th _a and th _v are maneuvers. Threshold values of having and not having th _a > th _v . Note that the calculation of the normalized residual ε _k can be configured to perform weighted averaging of residual vectors of a plurality of observation times.

Next, the processing content of the prediction process in the operation step of ST207 will be described. In this operation step, prediction processing represented by the following formulas (26) to (27) is performed.

Here, (x hat) _{k + 1 | k} is a prediction vector at the observation time t _{k + 1} , and P _{k + 1 | k} is a prediction error covariance matrix at the observation time t _{k + 1} . Incidentally, the covariance matrix Q _k, since the true value r _k of the target distance is not obtained, smoothing distance r _k as shown in the following expression (28) _| with using a value calculated using the _k, the system noise The standard deviation σ ^s _k is changed according to the target maneuver step value m _k , and is a large value when the maneuver step value is 1, and a small value when the maneuver step value is zero.

Next, the contents of the scheduling process in the operation step of ST208 will be described. In this operation step, when any of the following formulas (29) to (30) is satisfied, the active sensor 12 is scheduled to be used.

Here, f3 (•) is a function for extracting the distance component of the error covariance matrix, and th _r is a threshold value. In addition, instead of the equation (30) using the smooth error covariance matrix P _{k | k} , the prediction error covariance matrix P _{k + 1 | k} is used as represented by the following equation (31). be able to.

  Further, when any one of the above formulas (29) to (31) is satisfied, the active sensor 12 may be used for scheduling.

  As described above, in the present embodiment, the target observation values by the passive sensor and the active sensor are merged, and this is subjected to a filtering process by a tracking filter configured based on a predetermined motion model such as a uniform linear motion. Thus, a residual between the processing result and the observed value, that is, a target motion different from a predetermined motion model is detected as a target maneuver and quantified as a maneuver step value. When the maneuver step value changes, and at least one error covariance component calculated from the target smoothing error covariance matrix or prediction error covariance matrix by the filtering process exceeds a predetermined threshold value In this case, the observation scheduling is controlled so that the observation is performed by the active sensor in addition to the passive sensor continuously performing the observation.

  As a result, for example, for a target that is moving relatively linearly (a low maneuver target), the time interval in which the active sensor is used is widened to reduce the detectability of the device itself, and low power consumption. In addition, the time interval in which the active sensor is used is narrowed to maintain and improve the tracking performance for a target (high maneuver target) that moves violently like a turning motion. Therefore, it is possible to obtain a small and low power consumption composite sensor device that improves the tracking performance while reducing the detectability.

  FIG. 4 is a block diagram showing the configuration of the second embodiment of the composite sensor device according to the present invention. FIG. 5 is a flowchart for explaining the operation of the second embodiment of the composite sensor device. The second embodiment will be described with reference to each part of the first embodiment shown in FIGS. The same parts are denoted by the same reference numerals, and the description thereof is omitted. The second embodiment is different from the first embodiment in that when the observation timing of the passive sensor and the active sensor is scheduled, in the first embodiment, the observation of the target by the passive sensor is continued. At least when the error covariance component calculated from the smoothing error covariance matrix or the prediction error covariance matrix of the filtering processing result by the tracking filter exceeds a predetermined threshold and when the target maneuver step value changes In one case, the target is scheduled to be observed by the active sensor, whereas the second embodiment has a table or the like that associates the maneuver step value with the observation time interval of the active sensor 12a in advance. , While continuing to observe the target with passive sensors, Based on observation time intervals of active sensor obtained by referring to serial table, a point which is adapted to schedule observation by active sensor. Further, for the active sensor, the radiation direction of electromagnetic waves or sound waves is controlled based on the target information. Further, in the first embodiment, the tracking filter is a linear Kalman filter, but in the second embodiment, this is an extended Kalman filter. Hereinafter, the difference will be mainly described with reference to FIGS. 1 to 3 and FIGS.

As illustrated in FIG. 4, the composite sensor device includes a passive sensor 11, an active sensor 12a, a data fusion / tracking processing unit 13a, a maneuver detection unit 14, and a control unit 15a. The passive sensor 11 and the maneuver detection unit 14 are configured in the same manner as in the first embodiment. The active sensor 12a emits an electromagnetic wave or a sound wave with a directivity direction and an observation timing controlled by the control unit 15a, which will be described later, observes the distance and angle to the target, and sends the result to the data fusion / tracking processing unit 13a. . Data fusion-tracking processing unit 13a, the inside of the tracking filter and extended Kalman filter, and off Irutaringu process observation results from the passive sensor 11 and the active sensor 12a by using the tracking filter, the target smoothed value and smoothing the error of The target information including the covariance matrix and the target prediction value and the prediction error covariance matrix is calculated. In this embodiment, the tracking filter motion model is, for example, constant velocity linear motion. Further, based on the maneuver step value from the maneuver detection unit 14, control is performed so that the system noise level matches the target motion model.

  The control unit 15a has a table in which the maneuver step value and the observation time interval of the active sensor 12a are associated in advance. Based on the observation time interval obtained by referring to this table, the control unit 15a observes the target by the passive sensor 11. While continuing, scheduling of observation timing by the active sensor 12a is performed. In this table, for example, by setting a long observation time interval of the active sensor 12a for a maneuver step value corresponding to a low maneuver, and setting a short observation time interval for a high maneuver, The tracking performance is improved while lowering the detectability of. In addition, based on the target information from the data fusion / tracking processing unit 13a, the active sensor 12a is controlled in the radiation direction of electromagnetic waves or sound waves.

  Next, the operation of the composite sensor device of the second embodiment will be described with reference to the flowchart of FIG. First, the overall operation flow will be described.

  First, a sensor is selected at the observation timing by scheduling by the control unit 15a (ST501). Here, when the passive sensor 11 is selected, the target angle is observed by the passive sensor 11 according to the control from the control unit 15a, and the observation result is sent to the data fusion / tracking processing unit 13a (ST502). . The data fusion / tracking processing unit 13a performs a filtering process using a tracking filter based on the observation result, and calculates a target smooth value and a smooth error covariance matrix (ST503). On the other hand, when the active sensor 12a is selected in the operation step of ST501, the target distance and angle are observed by the active sensor 12a according to the control from the control unit 15a including the control of the directivity / radiation direction of the sensor. The result is sent to the data fusion / tracking processing unit 13a (ST504). The data fusion / tracking processing unit 13a calculates a target smooth value and a smooth error covariance matrix by filtering processing based on these observation results (ST505).

  Next, based on the residual between the target information and the observed values calculated by the data fusion / tracking processing unit 13a and the residual covariance matrix, the maneuver detecting unit 14 detects the target maneuver. Then, the detection result is output as a maneuver step value divided into a plurality of steps according to the maneuver degree (ST506). Next, based on the maneuver step value, the data fusion / tracking processing unit 13a controls the internal tracking filter so that the system noise level is more suitable for the target motion model, and the filtering process is performed. Continuously, a target predicted value and a prediction error covariance matrix are calculated (ST507).

  At the same time, the maneuver step value from the maneuver detector 14 is also sent to the controller 15a. A table in which the maneuver step value and the observation time interval of the active sensor 12a are associated in advance is set in the control unit 15a, and the control unit 15a refers to this table and observes the target with the active sensor 12a. While acquiring the observation time interval and continuing the observation of the target by the passive sensor 11, scheduling of the subsequent observation timing is performed so that the active sensor 12a is operated at the observation time interval acquired with reference to this table. In the table, for example, a long observation time interval is set for the maneuver step value of the low maneuver, and a short observation time interval is set for the maneuver step value corresponding to the high maneuver. The frequency of observation by the active sensor 12a is controlled in accordance with the change in the maneuver step value detected in step 1, and the tracking performance is improved while lowering the detectability of the device itself. Furthermore, the control unit 15a receives target information from the data fusion / tracking processing unit 13a, and controls the radiation direction so that the active sensor 12a emits electromagnetic waves or sound waves in the direction of the target being observed based on the target information. Do. Then, these observation timing scheduling results and radiation direction control information are sent from the control unit 15a to the passive sensor 11 and the active sensor 12a (ST508). Thereafter, the above-described series of operation steps is repeated until the operation ends (ST509).

  Next, the processing contents in each operation step described above will be described in detail. First, in the filtering process in steps ST503 and ST505, the processing contents when the observation results from the passive sensor 11 for observing the target angle and the active sensor 12a for observing the distance and the angle are merged on an observation value basis. explain.

The target motion model is defined as the following equations (32) to (36).

Here, _{x k} is the position ^p _x k, the state vector of position ^p _{y k} and its velocity component in the y-axis of the x-axis at the measurement time _{t k, F k} and _{G k} is the observation time from the measurement time _{t k} t _{k + 1} transition matrix and the driving matrix onto, _{w k} is the average of the observation time _{t k} 0, the system noise vector covariance matrix _{Q k,} sigma ^s _k is the standard deviation of the system noise at the measurement time _{t k.} Also, ^{A T} is the transpose of vector or matrix A, _{I n} is a unit matrix of n rows and n columns, _{O n} denotes a zero matrix of n rows and n columns.

The observation model of the passive sensor 11 is defined as the following equations (37) to (39).

Here, y p k is, the observation vector of the passive sensor 11 of the observation time t k, h p k (· ) , the observation function of the passive sensor 11 of the observation time t k, v p k is, the observation time t k average 0 passive sensor 11, an observation noise vector covariance matrix R p k.

The observation model of the active sensor 12a is defined as the following equations (40) to (42).

Here, y a k is the observation vector for active sensor 12a of the observation time t k, h a k (· ) is the observation function of the active sensor 12a of the observation time t k, v a k is the observation time t k This is an observation noise vector having an average 0 of the active sensor 12a and a covariance matrix R a k .

In the operation step of ST503, smoothing processing expressed by the following formulas (43) to (48) is performed using the angle measurement data from the passive sensor 11.

Here, (y tilde) p k is the residual vector of the passive sensor 11 of the measurement time t k, H p k is the Jacobian matrix of the passive sensor 11 of the measurement time t k, S p k is the measurement time t k The residual covariance matrix of the passive sensor 11, K p k is the Kalman gain matrix of the passive sensor 11 at the observation time t k , (xhat) k | k is the smooth vector at the observation time t k , P k | k Is the smoothing error covariance matrix at the observation time t k , (xhat) k | k−1 is the prediction vector at the observation time t k , and P k | k−1 is the prediction error covariance matrix at the observation time t k. It is. A -1 indicates an inverse matrix of the matrix A.

In the operation step of ST505, smoothing processing expressed by the following formulas (49) to (54) is performed using the distance measurement / angle measurement data from the active sensor 12a.

Here, (y tilde) a k is the residual vector of the active sensor 12a of the observation time t k, H a k is the Jacobian matrix of the active sensor 12a of the observation time t k, S a k is the measurement time t k residual covariance matrix of the active sensor 12a of, K a k is a Kalman gain matrix of active sensor 12a of the observation time t k.

Next, the contents of processing when maneuver is detected in the operation step of ST506 will be described. In this operation step, the maneuver state is detected by the following formulas (55) to (56), and divided into maneuver step values.

Here, epsilon _k, normalized residual observation time _{t k,} (y ^{tilde) p /} _{a k} is the residual vector of the passive sensor 11 or active sensor 12a of the observation time _{t k,} f1 (·) is passive sensor 11 or function to extract the azimuth angle component of the residual vector of the active sensor ^{12a, S p /} _{a k} is the residual covariance matrix of the passive sensor 11 or active sensor 12a of the observation time _{t k,} f2 (·) Is a function for extracting the azimuth angle component of the residual covariance matrix of the passive sensor 11 or the active sensor 12a, m _k is a maneuver stage value (maneuver state) at the observation time t _k , and th ₁ and th ₂ are maneuver stages Each threshold of the value determination, and th ₂ > th ₁ . The number of divisions of the maneuver step value is not limited to the above three, and any value can be selected.

Next, the details of the prediction process in the operation step of ST507 will be described. In this operation step, prediction processing represented by the following formulas (57) to (58) is performed.

Here, (x hat) _{k + 1 | k} is a prediction vector at the observation time t _{k + 1} , and P _{k + 1 | k} is a prediction error covariance matrix at the observation time t _{k + 1} . The covariance matrix Q _k uses the following equation (59), and the standard deviation σ ^s _k of the system noise is changed according to the target maneuver state m _k .

Next, the contents of the scheduling process in the operation step of ST508 will be described. In this operation step, when the following equation (60) is satisfied, scheduling is performed so as to use the active sensor 12a.

Here, t ^a _k is the observation time when the active sensor 12a is last used, and Δt (·) is the observation time interval of the active sensor 12a set in advance corresponding to the maneuver step value.

  As described above, also in this embodiment, similarly to the first embodiment, the target observation values by the passive sensor and the active sensor are fused, and this is filtered by the tracking filter configured based on the predetermined motion model. The target maneuver is detected from the residual between the processing result and the observed value, and quantified as a maneuver step value. Then, based on a table in which the maneuver stage value and the observation time interval of the active sensor are associated in advance, the observation timing scheduling using the active sensor is performed while continuing the observation of the target by the passive sensor. The observation time intervals of the active sensors set in this table are wide observation time intervals for maneuver step values corresponding to low maneuvers, and narrow observation time intervals for maneuver step values corresponding to high maneuvers. Is set.

  As a result, as in the first embodiment, the time interval in which the active sensor is used is widened for the target that is moving relatively linearly (the target of the low maneuver), so that the self-detectability of the device itself is increased. In addition to lowering power consumption and reducing power consumption, the time interval for using the active sensor is narrowed for targets that are moving violently like turning motion (targets of high maneuvers) and the tracking performance is maintained.・ Improved. In addition, the radiation direction of the active sensor is controlled so that electromagnetic waves or sound waves are emitted in the direction of the target being observed, thereby further reducing the detectability and suppressing power consumption. Therefore, it is possible to obtain a small and low power consumption composite sensor device that improves the tracking performance while reducing the detectability.

  Note that the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, you may combine suitably the component covering different embodiment.

The block diagram which shows the structure of the 1st Example of the composite sensor apparatus which concerns on this invention. The flowchart for demonstrating operation | movement of the 1st Example of the composite sensor apparatus based on this invention illustrated in FIG. Explanatory drawing which modeled and illustrated the relationship between the sensor used at each observation timing, and a maneuver step value. The block diagram which shows the structure of the 2nd Example of the composite sensor apparatus which concerns on this invention. 6 is a flowchart for explaining the operation of the second embodiment of the composite sensor device according to the present invention illustrated in FIG. 4; The block diagram which shows an example of a structure of the conventional composite sensor apparatus. The flowchart for demonstrating operation | movement of the conventional composite sensor apparatus illustrated in FIG.

Explanation of symbols

11 Passive sensor 12, 12a Active sensor 13, 13a Data fusion / tracking processing unit 14 Maneuver detection unit 15, 15a Control unit

Claims (4)

A passive sensor that observes the target angle without the emission of electromagnetic waves or sound waves;
An active sensor that emits electromagnetic waves or sound waves to observe the distance to the target or the distance and angle to the target;
A tracking filter configured based on a predetermined motion model, and the target observation value from the passive sensor or the target observation from the active sensor with respect to the target track information held in the tracking filter. Smoothing process is performed regardless of which observation value is input, and target information including the target smoothing value and smoothing error covariance matrix and prediction value and prediction error covariance matrix is calculated. A tracking means for merging observation values from sensors on an observation value basis ,
The target motion different from the predetermined motion model based on the residual and residual covariance matrix of the observed value and the predicted value of the passive sensor and the active sensor at the time of calculating the target information in the tracking means And a maneuver detecting means for classifying the detected maneuver state into a plurality of maneuver step values associated with a plurality of preset threshold values,
System noise control means for controlling the system noise level of the tracking filter based on the maneuver step value from the maneuver detection means;
While observing the target by the passive sensor, based on at least one of the smoothing error covariance matrix from the tracking means, the prediction error covariance matrix, and the maneuver step value from the maneuver detecting means, the passive A composite sensor device comprising a sensor and sensor control means for scheduling observation timing of the active sensor.   While the sensor control means continues to observe the target by the passive sensor, an error covariance component calculated from the target smoothing error covariance matrix or prediction error covariance matrix from the tracking means is predetermined. The active sensor is scheduled to observe the target when at least one of a threshold value is exceeded and a target maneuver step value from the maneuver detecting means changes. Item 4. The composite sensor device according to Item 1.   The sensor control means, while continuing the observation of the target by the passive sensor, based on the observation time interval of the active sensor preset in association with the target maneuver step value from the maneuver detection means, The composite sensor device according to claim 1, wherein scheduling is performed so that the target is observed by the active sensor.   Further, the sensor control means controls the radiation direction of the electromagnetic wave or sound wave of the active sensor based on the target information from the tracking means, and the active sensor detects the electromagnetic wave or sound wave in the designated direction based on this control. The composite sensor device according to claim 1, wherein

Images