JP7060441B2 - Radar device and target detection method - Google Patents

Description

The present invention relates to a radar device and a target detection method.

Conventionally, a radar device that performs tracking processing using a filter having a small processing load when the future speed prediction value is larger than the threshold value and the future position prediction value is in the area of interest is known (for example, Patent Document). 1).

Japanese Unexamined Patent Publication No. 2013-167476

However, since the radar device performs tracking processing using a filter having a large processing load when the future speed prediction value is smaller than the threshold value, it is improved in that the processing load when executing the tracking processing is reduced. There is room for.

One aspect of the embodiment is made in view of the above, and an object thereof is to provide a radar device and a target detection method for reducing a processing load.

The radar device according to one embodiment includes a detection unit, a selection unit, a setting unit, and a generation unit. The detection unit detects the observed value that becomes the peak in the frequency spectrum based on the frequency-modulated transmission wave and the reflected wave of the transmission wave by the target. The selection unit selects the corresponding observation value corresponding to the target from the observation value. The setting unit sets a tracking filter from a plurality of filters based on the difference between the corresponding observation value and the predicted value predicted from the previous target value. The generation unit generates the target value of this time by using the tracking filter.

According to one aspect of the embodiment, the processing load can be reduced.

FIG. 1 is a diagram showing an outline of a target detection method according to an embodiment. FIG. 2 is a block diagram showing a configuration of a radar device according to an embodiment. FIG. 3 is a diagram showing an example of the relationship between the transmission frequency, the reception frequency, and the beat frequency. FIG. 4 is a diagram showing the result of performing FFT processing on one beat signal. FIG. 5 is a diagram showing an example of the phase change of the peak between the FFT processing result of the beat signal that is continuous in time and the beat signal. FIG. 6 is a block diagram showing the configuration of the tracking processing unit. FIG. 7 is a block diagram showing the configuration of the first filter processing unit. FIG. 8 is a diagram showing changes in the relative speed of the target to be tracked. FIG. 9 is a diagram showing changes in the received power of the target to be tracked. FIG. 10 is a block diagram showing the configuration of the second filter processing unit. FIG. 11 is a block diagram showing the configuration of the third filter processing unit. FIG. 12 is a block diagram showing the configuration of the fourth filter processing unit. FIG. 13 is a flowchart illustrating the tracking state setting process. FIG. 14 is a flowchart illustrating the tracking filter setting process. FIG. 15 is a flowchart illustrating the first filter processing. FIG. 16 is a flowchart illustrating the fourth filter processing.

Hereinafter, embodiments of the radar device and the target detection method disclosed in the present application will be described in detail with reference to the accompanying drawings. The present invention is not limited to this embodiment. In the following, the case where the radar device 1 is of the FCM (Fast-Chirp Modulation) method will be described as an example, but the radar device 1 will be described by another method such as the FM-CW (Frequency Modulated Continuous Wave) method. There may be.

First, an outline of the target detection method according to the embodiment will be described with reference to FIG. FIG. 1 is a diagram showing an outline of a target detection method according to an embodiment.

As shown in FIG. 1, the radar device 1 is mounted on, for example, in the front grill of the own vehicle MC, and is a target existing in the traveling direction of the own vehicle MC (for example, a preceding vehicle, a bicycle, a person, a guardrail). (Still objects, etc.) are detected. The radar device 1 may be mounted on other places such as the windshield, the rear grille, and the left and right side portions (for example, the left and right door mirrors). The radar device 1 may be used for various purposes other than the in-vehicle radar device (for example, monitoring of an airplane or a ship).

The radar device 1 detects an observed value that becomes a peak in the frequency spectrum based on the transmitted wave SW and the received wave RW (S1).

The observed value is a value indicating the state vector of the target, and includes values such as the distance to the target and the relative velocity. The observed value may be called an instantaneous value. In addition, the observed values may include noise.

The radar device 1 selects a corresponding observation value corresponding to the target from the observation value (S2). For example, the radar device 1 selects the observation value having the shortest distance from the predicted value as the corresponding observation value among the observation values. The predicted value is the value predicted as the destination of the previous target value generated by the previous scan. Specifically, the radar device 1 calculates a predicted value using an αβ filter having a small processing load.

The radar device 1 sets a tracking filter from a plurality of filters based on the difference between the corresponding observed value and the predicted value (S3).

For example, in the radar device 1, when the difference between the corresponding observed value and the predicted value is large and the variation in the observed noise is large, a particle filter having excellent tracking performance for a target performing irregular operation is set as a tracking filter. ..

Further, when the difference between the corresponding observed value and the predicted value is small and the variation in the observed noise is small, the radar device 1 sets a Kalman filter having a processing load smaller than that of the particle filter as a tracking filter.

After setting the tracking filter, the radar device 1 generates a target value using the set tracking filter (S4).

The radar device 1 sets a tracking filter based on the difference between the corresponding observed value and the predicted value, and generates a target value using the set tracking filter. As a result, the radar device 1 can improve the tracking performance for the target and reduce the processing load.

Next, the configuration of the radar device 1 according to the embodiment will be described with reference to FIG. FIG. 2 is a block diagram showing the configuration of the radar device 1 according to the embodiment. In FIG. 2, only the components necessary for explaining the features of the present embodiment are represented by functional blocks, and the description of general components is omitted.

In other words, each component shown in FIG. 2 is a functional concept and does not necessarily have to be physically configured as shown. For example, the specific form of distribution / integration of each functional block is not limited to the one shown in the figure, and all or part of it is functionally or physically distributed in any unit according to various loads and usage conditions. -It is possible to integrate and configure.

The radar device 1 includes a transmission unit 10, a reception unit 20, a generation unit 30, and a processing unit 40. The transmission unit 10 includes a signal generation unit 11, an oscillator 12, and a transmission antenna 13.

The signal generation unit 11 generates a modulated signal whose voltage changes in a sawtooth pattern and supplies it to the oscillator 12. The oscillator 12 generates a transmission signal ST, which is a chirp signal whose frequency increases with the passage of time, for each predetermined period Tc (hereinafter referred to as a chirp period Tc) based on the modulated signal generated by the signal generation unit 11. Then, it is output to the transmitting antenna 13.

The transmission antenna 13 converts the transmission signal ST from the oscillator 12 into a transmission wave SW, and outputs the transmission wave SW to the outside of the own vehicle MC (see FIG. 1). The transmission wave SW output by the transmission antenna 13 is a chirp wave whose frequency increases with the passage of time for each chirp period Tc. The transmitted wave SW transmitted from the transmitting antenna 13 in front of the own vehicle MC is reflected by a target such as a preceding vehicle and becomes a reflected wave.

The receiving unit 20 includes a plurality of receiving antennas 21 forming an array antenna. Each receiving antenna 21 receives the reflected wave from the target as a received wave RW, converts the received wave RW into a received signal SR, and outputs the received wave RW to the generation unit 30. The number of receiving antennas 21 shown in FIG. 2 is 4, but may be 3 or less or 5 or more.

The generation unit 30 generates a beat signal SB from the transmission signal ST and the reception signal SR. The generation unit 30 includes a plurality of mixers 31 and a plurality of A / D conversion units 32. The mixer 31 and the A / D conversion unit 32 are provided for each receiving antenna 21.

The received signal SR output from the receiving antenna 21 is amplified by an amplifier (for example, a low noise amplifier) (not shown) and then input to the mixer 31. The mixer 31 mixes a part of the transmission signal ST and the reception signal SR, removes unnecessary signal components, generates a beat signal SB, and outputs the beat signal SB to the A / D conversion unit 32.

As a result, the beat frequency f, which is the difference between the frequency f ST of the transmission signal _ST (hereinafter referred to as the transmission frequency f _ST ) and the frequency f _{SR of the reception signal SR} (hereinafter referred to as the reception frequency f _SR ). A beat signal SB having _SB (= f _ST −f _SR ) is generated. The beat signal SB generated by the mixer 31 is converted into a digital signal by the A / D conversion unit 32 and then output to the processing unit 40.

FIG. 3 is a diagram showing an example of the relationship between the transmission frequency f _ST , the reception frequency f _SR , and the beat frequency f _SB . As shown in FIG. 3, the beat signal SB is generated for each chirp wave.

Further, in the example shown in FIG. 3, the transmission frequency f _ST increases with a slope θ (= (f1-f0) / Tm) from the reference frequency f0 with time for each chirp wave, and reaches the maximum frequency f1. It is a chirp wave that returns to the reference frequency f0 in a short time. The transmission frequency f _ST reaches the maximum frequency f1 from the reference frequency f0 in a short time for each chirp wave, and decreases with a gradient θ (= (f0-f1) / Tm) from the maximum frequency f1 with time. It may be wavy.

Returning to FIG. 2, the processing unit 40 will be described. The processing unit 40 includes a transmission control unit 41, a signal processing unit 42, and a storage unit 43.

The processing unit 40 includes, for example, a microcomputer including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an input / output port, and various circuits, and controls the entire radar device 1.

The CPU of the microcomputer functions as a transmission control unit 41 and a signal processing unit 42 of the processing unit 40, for example, by reading and executing a program stored in the ROM. At least one or all of the transmission control unit 41 and the signal processing unit 42 may be configured by hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

The storage unit 43 stores the history data 43a. The history data 43a is a history of processed data in a series of signal processing related to the detection of a target periodically executed by the signal processing unit 42.

Further, the storage unit 43 corresponds to, for example, a RAM or a data flash. The RAM and the data flash can store the history data 43a. The processing unit 40 may acquire the above-mentioned program and various information via another computer or a portable recording medium connected by a wired or wireless network.

The transmission control unit 41 controls the signal generation unit 11 of the transmission unit 10 to output a modulated signal whose voltage changes like a saw from the signal generation unit 11 to the oscillator 12. As a result, the transmission signal ST whose frequency changes with the passage of time is output from the oscillator 12 to the transmission antenna 13.

The signal processing unit 42 includes a conversion unit 44, a peak extraction unit 45, a distance / relative speed calculation unit 46, an angle estimation unit 47, a clustering processing unit 48, and a tracking processing unit 49. The conversion unit 44 to the clustering processing unit 48 constitute a detection unit.

The conversion unit 44 performs a two-dimensional Fast Fourier Transform (FFT) process (hereinafter referred to as FFT process) on the beat signal SB output from each A / D conversion unit 32, respectively.

The conversion unit 44 converts the beat signal SB into a frequency spectrum indicating the distance and the relative velocity by performing two-dimensional FFT processing on the beat signal SB generated by the generation unit 30 for each reception antenna 21 in a predetermined period. .. Then, the conversion unit 44 outputs the frequency spectrum to the peak extraction unit 45.

Here, although it is a well-known technique, the principle of detecting the distance and the relative velocity to the target in the FCM type radar device 1 will be briefly described. As described above, the transmission wave SW based on the transmission signal ST is transmitted from the transmission antenna 13, the transmission wave SW is reflected by the target to become a reflected wave, and the reflected wave is received by the reception antenna 21 as a reception wave RW. It is output as a received signal SR.

The period from the transmission of the transmission wave SW to the output of the reception signal SR increases or decreases in proportion to the distance between the target and the radar device 1, and the frequency of the beat signal SB is the object. It is proportional to the distance between the marker and the radar device 1 (hereinafter referred to as the distance to the target).

Therefore, by performing the FFT process on the beat signal SB, a peak appears in the frequency bin (hereinafter, may be referred to as a distance bin) corresponding to the distance from the target. By specifying the distance bin where the peak exists, the distance to the target can be detected.

FIG. 4 is a diagram showing the result of performing FFT processing on one beat signal SB. In FIG. 4, the horizontal axis is the frequency and the vertical axis is the magnitude of the power. In the example shown in FIG. 4, a peak appears in the distance bin fr10, indicating that the target exists at the distance corresponding to the distance bin fr10.

By the way, when the relative speed between the target and the radar device 1 is zero, the Doppler component does not occur in the received signal SR, and the phase is the same between the received signal SRs corresponding to each chirp wave, so that each beat. The phase of the signal SB is also the same.

On the other hand, when the relative speed between the target and the radar device 1 is not zero, a Doppler component is generated in the received signal SR, and the phase is different between the received signal SRs corresponding to each chirp wave, so that beats that are continuous in time. A phase change appears between the signal SBs according to the Doppler frequency.

As described above, when the relative velocity between the target and the radar device 1 is not zero, the phase change according to the Doppler frequency appears at the peak of the same target between the beat signals SB. Therefore, by arranging the frequency spectra obtained by FFT processing each beat signal SB in time series and performing the second FFT processing, it is possible to obtain a frequency spectrum in which a peak appears in the frequency bin with respect to the Doppler frequency. By detecting the frequency bin in which the peak appears (hereinafter, may be referred to as a velocity bin), the relative velocity with the target can be detected.

FIG. 5 is a diagram showing an example of the phase change of the peak between the FFT processing result of the beat signal SB and the beat signal SB which are continuous in time. In the example shown in FIG. 5, there is a peak in the distance bin fr10, indicating that the phase of the peak is changing.

In this way, by performing the FFT process twice on the beat signal SB and detecting the distance bin and the velocity bin in which the peak exists, the distance to the target and the relative velocity can be detected.

Returning to FIG. 2, the peak extraction unit 45 extracts a peak whose power is equal to or higher than a predetermined value from the frequency spectrum input from the conversion unit 44 as an observed value, and outputs information about the peak to the distance / relative velocity calculation unit 46. ..

The distance / relative velocity calculation unit 46 calculates the distance to the target and the relative velocity based on the combination of the distance bin and the velocity bin of the peak extracted by the peak extraction unit 45.

The angle estimation unit 47 separates information about a plurality of targets existing in the same distance bin from the signals of each distance bin of the peak extracted by the peak extraction unit 45 by a predetermined angle calculation process, and the plurality of them. Estimate the angle of each target.

The angle estimation unit 47 pays attention to the signals of the same peak frequency bin (hereinafter referred to as peak signals) in all the frequency spectra of the four beat signals SB based on the received signal SRs of the four receiving antennas 21, and those peaks. Estimate the angle of the target based on the phase information of the signal.

The directional estimation in the angle estimation unit 47 is performed by using a predetermined directional estimation method such as ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques), DBF (Digital Beam Forming), or MUSIC (Multiple Signal Classification). Will be.

The angle estimation unit 47 outputs the observed values in the latest scan to the clustering processing unit 48.

The clustering processing unit 48 detects, for example, an observation value satisfying predetermined conditions such as a short distance and a short relative velocity as an observation value corresponding to one target.

Next, the tracking processing unit 49 will be described with reference to FIG. FIG. 6 is a block diagram showing the configuration of the tracking processing unit 49.

The tracking processing unit 49 includes a tracking state setting unit 50, a tracking processing setting unit 51, a first filter processing unit 52, a second filter processing unit 53, a third filter processing unit 54, and a fourth filter processing unit 55. And prepare.

The tracking state setting unit 50 determines whether the tracking state for the target is the wake state during the test or the confirmed wake state. The wake state during the test is a state in which it is being tested whether or not the newly registered target value (observed value) should be tracked, and a temporary tracking is performed. In other words, the wake state during the test is the state before the target value is determined to be the tracking target. The confirmed wake state is a state in which the target value is determined to correspond to the target and formal tracking is performed. In other words, the confirmed wake state is the state in which the target value is determined to be the tracking target.

The tracking state setting unit 50 determines the tracking state based on the survival counter associated with the target value. The survival counter is a value set for each target value.

The survival counter is set to a predetermined initial value for the newly registered target value. Then, the survival counter is adjusted based on the processing result by the first filter processing unit 52.

For example, the survival counter of the target value that is correlated with the observed value by the processing result, in other words, the survival counter of the target value that is continuous is added, and the survival counter of the target value that is extrapolated without correlation is added. Is subtracted.

The tracking state setting unit 50 sets the tracking state of the newly registered target value to the wake state during verification. When the survival counter becomes equal to or higher than a preset predetermined counter value, the tracking state setting unit 50 sets the tracking state to the definite wake state. The predetermined counter value is a preset value, and is a value that can be determined that the target value corresponds to the target. For example, the predetermined counter value is set so that the survival counter becomes equal to or higher than the predetermined counter value when the correlation is obtained several times in succession. As a result, the tracking state setting unit 50 can quickly shift the tracking state of the highly correlated target value to the definite track state.

The tracking state setting unit 50 erases the target value whose survival counter is equal to or less than the erased value. That is, the tracking state setting unit 50 determines that the target value whose survival counter is equal to or less than the erased value is a missed search, and excludes it from the tracking target. The erase value is a preset value and is smaller than the initial value, for example, "0".

When the survival counter of the target value whose tracking state is the track state under verification becomes equal to or less than the erased value, the tracking state setting unit 50 erases the target value without setting the tracking state to the definite track state. As a result, the radar device 1 can erase the target value corresponding to the noise, and can reduce the processing load.

After setting the tracking state to the definite wake state, the tracking state setting unit 50 maintains the tracking state in the definite wake state even when the survival counter becomes smaller than the predetermined counter value. The tracking state setting unit 50 erases the target value even after the tracking state is set to the definite wake state when the survival counter becomes equal to or less than the erase value. As a result, the radar device 1 can eliminate the target value that cannot be correlated with the observed value and reduce the processing load.

The tracking processing setting unit 51 sets the tracking processing for the target value based on the difference between the corresponding observed value and the predicted value for the target value whose tracking state is changed from the in-testing track state to the confirmed track state. When the target value is determined to be the tracking target, the tracking processing setting unit 51 sets the tracking filter based on the difference between the corresponding observed value and the predicted value. The tracking process setting unit 51 constitutes a setting unit.

Specifically, the tracking processing setting unit 51 sets the tracking filter based on the average value of the difference between the corresponding observation value and the predicted value used in the first tracking processing in the first filter processing unit 52, which will be described in detail later. do. As a result, the tracking processing setting unit 51 can reduce the influence of the variation in the difference between the corresponding observed value and the predicted value, can set the tracking filter accurately, and can improve the tracking performance.

When the average value is equal to or less than the first predetermined value, the tracking process setting unit 51 sets the tracking process to the second tracking process. The second tracking process is a process of generating a target value by using a linear Kalman filter which is a linear filter. That is, the tracking processing setting unit 51 sets the tracking filter to the linear Kalman filter. The first predetermined value is a preset value, and is a value that can be determined to be a target with little variation in observed noise and few irregular movements.

When the average value is equal to or less than the first predetermined value, it is presumed that the target is a vehicle, for example. The linear filter is not limited to the linear Kalman filter, and may be, for example, an αβ filter, and may be a filter having a small processing load.

When the average value is larger than the first predetermined value and is equal to or less than the second predetermined value, the tracking processing setting unit 51 sets the tracking process to the third tracking process. The third tracking process is a process of generating a target value by using an unscented Kalman filter which is a non-linear filter. That is, the tracking processing setting unit 51 sets the tracking filter to the unscented Kalman filter. The second predetermined value is larger than the first predetermined value and is a preset value. The second predetermined value is a value that can be determined to be a target that has a large variation in observed noise and may cause irregular operation.

The second predetermined value is a preset value, and is a value larger than the first predetermined value. When the average value is larger than the first predetermined value and equal to or less than the second predetermined value, it is presumed that the target is, for example, a bicycle. The non-linear filter is not limited to the unscented Kalman filter. For example, the non-linear filter may be an extended Kalman filter, and may be a filter that has a smaller processing load than the particle filter and can track a target that may perform irregular operation.

When the average value is larger than the second predetermined value, the tracking process setting unit 51 sets the tracking process to the fourth tracking process. The fourth tracking process is a process of generating a target value by using a particle filter which is a non-linear filter. That is, the tracking processing setting unit 51 sets the tracking filter to the particle filter.

When the average value is larger than the second predetermined value, it is presumed that the target is a person, for example. A particle filter is a filter that can track targets that are likely to behave irregularly.

The tracking process setting unit 51 sets the tracking process to the first tracking process for the target value whose tracking state is the wake state during the verification. The first tracking process is a process of generating a target value using an αβ filter.

Next, the first filter processing unit 52 will be described with reference to FIG. 7. FIG. 7 is a block diagram showing the configuration of the first filter processing unit 52.

The first filter processing unit 52 includes a prediction processing unit 52a, a gating processing unit 52b, a connection processing unit 52c, and a target value generation unit 52d.

The first filter processing unit 52 performs the first tracking process on the target value whose tracking state is the track state during the verification. The first filter processing unit 52 generates a target value by filter processing using an αβ filter which is a linear filter. The filter used in the first tracking process is a linear filter having a smaller processing load than the non-linear filter. The filter used in the first tracking process is the filter having the smallest processing load among the filters used in the radar device 1. The linear filter may be an αβγ filter or a linear Kalman filter. The αβ filter is an example of a predictive filter during testing.

The prediction processing unit 52a performs prediction processing and predicts the predicted value this time. Specifically, the prediction processing unit 52a performs prediction processing during the test. The in-verification prediction process is a prediction process performed before the tracking state is set to the definite track state by the tracking state setting unit 50, that is, when the tracking state is the in-test track state. In other words, the in-test prediction process is a prediction process performed before the target value is determined to be the tracking target. The prediction processing unit 52a predicts the current predicted value by moving the previous target value by the inter-observation time (between the previous observation time and the current observation time) at the speed of the previous target value.

The gating processing unit 52b performs gating processing and excludes a part of the detected observed values from the candidates for the corresponding observed values. The gating processing unit 52b excludes observation values whose distance from the predicted value is larger than a predetermined distance from the candidates for the corresponding observation value. That is, the gating processing unit 52b uses an observed value whose distance from the predicted value is equal to or less than a predetermined distance as a candidate for the corresponding observed value. The predetermined distance is a preset distance, and is the maximum distance at which it can be determined that an observation value corresponding to the physical slip exists.

Further, the gating processing unit 52b calculates the relative speed difference between the previous corresponding observation value and the current observation value, and excludes the observation value whose relative speed difference is larger than the predetermined speed difference from the candidates for the corresponding observation value. .. That is, the gating processing unit 52b uses an observed value having a relative speed difference of less than or equal to a predetermined speed difference as a candidate for the corresponding observed value. The predetermined speed difference is a preset speed difference, and is a speed difference that can be determined that the observed value is an observed value corresponding to noise.

When the target to be tracked is moving at a constant relative speed, for example, as shown by the solid line in FIG. 8, the relative speed of the target changes within a predetermined speed range including an error. do. FIG. 8 is a diagram showing changes in the relative speed of the target to be tracked.

On the other hand, as shown by the broken line in FIG. 8, the observed value corresponding to the noise has a large change in the relative speed and changes outside the predetermined speed range.

Therefore, it can be determined that the observed value having a large relative velocity difference between the previous corresponding observation value and the current observation value is noise. Therefore, the gating processing unit 52b can exclude the observation value corresponding to the noise by excluding the observation value whose relative speed difference is larger than the predetermined speed difference from the candidates of the corresponding observation value, and can improve the tracking performance. Can be improved.

Further, the gating processing unit 52b calculates the received power difference between the previous corresponding observation value and the current observation value, that is, the power difference, and sets the observation value in which the received power difference is larger than the predetermined power difference as the corresponding observation value. Exclude from candidates. The gating processing unit 52b uses an observed value having a received power difference of less than or equal to a predetermined power difference as a candidate for a corresponding observed value. The predetermined power difference is a preset power difference, and is a power difference that can be determined that the observed value is an observed value corresponding to the clutter.

For example, when the target approaches the own vehicle MC, as shown by the solid line in FIG. 9, the received power of the target increases with the passage of time within a predetermined power range including an error. .. On the other hand, the received power of a clutter such as the ground changes within a predetermined clutter power range with the passage of time. FIG. 9 is a diagram showing changes in the received power of the target to be tracked.

Therefore, it can be determined that the observed value having a large difference in received power between the previous corresponding observation value and the current observation value is a clutter. Therefore, the gating processing unit 52b can exclude the observation value corresponding to the clutter by excluding the observation value whose received power difference is larger than the predetermined power difference from the candidates of the corresponding observation value, and can improve the tracking performance. Can be improved.

In this way, the gating processing unit 52b excludes the observed values that do not correspond to the target to be tracked from the observed values based on the above conditions.

The concatenation processing unit 52c performs concatenation processing and selects a corresponding observation value from the observation values not excluded by the gating processing unit 52b. Specifically, the connection processing unit 52c calculates the difference between the observed value and the predicted value. Then, the connection processing unit 52c selects the observation value having the smallest difference, that is, the observation value closest to the predicted value from the observation values, as the corresponding observation value. The connection processing unit 52c constitutes a selection unit.

The target value generation unit 52d performs a target value generation process, and adds a value obtained by multiplying the difference between the corresponding observed value and the predicted value by a predetermined filter constant to the predicted value to generate the target value. The target value generation unit 52d constitutes a generation unit. Since the difference between the corresponding observed value and the predicted value is used by the tracking processing setting unit 51 described above, it is stored in the storage unit 43.

Next, the second filter processing unit 53 will be described with reference to FIG. FIG. 10 is a block diagram showing the configuration of the second filter processing unit 53.

The second filter processing unit 53 includes a prediction processing unit 53a, a gating processing unit 53b, a connection processing unit 53c, and a target value generation unit 53d.

The second filter processing unit 53 performs the second tracking process on the target value whose tracking state is the definite track state. The second filter processing unit 53 generates a target value by filter processing using a linear Kalman filter.

When the tracking state is changed from the in-test track state to the definite track state, the second filter processing unit 53 takes over the target value generated by the first filter processing unit 52 and performs the tracking process. The same applies to the third filter processing unit 54 and the fourth filter processing unit 55.

The prediction processing unit 53a performs prediction processing and predicts the predicted value this time. Specifically, the prediction processing unit 53a performs definite prediction processing. The definite prediction process is a process performed after the tracking state is set to the definite track state by the tracking state setting unit 50. In other words, the definite prediction process is a process performed after it is determined that the target values are continuous. The prediction processing unit 53a predicts the predicted value this time by moving the previous target value according to a constant velocity linear model or the like which is a linear motion model.

The gating processing unit 53b performs gating processing, and similarly to the gating processing unit 52b of the first filter processing unit 52, a part of the detected observed values is excluded from the candidates for the corresponding observed values. Here, some of the observed values are excluded based on the Mahalanobis distance. The Mahalanobis distance is a distance expressed using a plurality of parameters of the observed value and the predicted value, and is a distance considering the statistical variation calculated from the correlation between the observed value and the predicted value.

The concatenation processing unit 53c performs concatenation processing and selects a corresponding observation value from the observation values not excluded by the gating processing unit 53b. Specifically, the connection processing unit 53c selects the observation value having the smallest Mahalanobis distance with respect to the predicted value as the corresponding observation value. The connection processing unit 53c constitutes a selection unit.

When the target value generation unit 53d performs target generation processing, assumes that both the current observation value and the previous target value contain an error, and moves the previous target value according to a constant velocity linear model. In addition, the optimum Kalman gain that minimizes the error is calculated. Then, the target value generation unit 53d adds the value obtained by multiplying the difference between the corresponding observed value and the predicted value by the optimum Kalman gain to the predicted value to generate the target value. The target target generation unit 53d constitutes a generation unit.

Next, the third filter processing unit 54 will be described with reference to FIG. FIG. 11 is a block diagram showing the configuration of the third filter processing unit 54.

The third filter processing unit 54 includes a prediction processing unit 54a, a gating processing unit 54b, a connection processing unit 54c, and a target value generation unit 54d.

The third filter processing unit 54 performs the third tracking process on the target value whose tracking state is the definite track state. The third filter processing unit 54 generates a target value by filter processing using an unscented Kalman filter.

The prediction processing unit 54a performs prediction processing, specifically, definite prediction processing, and predicts the predicted value this time. The prediction processing unit 54a predicts the predicted value this time by moving the previous target value as a non-linear model according to a motion model showing a state transition equation in sudden acceleration / deceleration / sharp turning.

The gating processing unit 54b performs gating processing, and similarly to the gating processing unit 52b of the first filter processing unit 52, a part of the detected observed values is excluded from the candidates for the corresponding observed values. The gating processing unit 54b excludes a part of the observed value based on the Mahalanobis distance, similarly to the gating processing unit 53b of the second filter processing unit 53.

The concatenation processing unit 54c performs concatenation processing and selects a corresponding observation value from the observation values not excluded by the gating processing unit 54b. Specifically, the connection processing unit 54c selects the observation value having the smallest Mahalanobis distance with respect to the predicted value as the corresponding observation value. The connection processing unit 54c constitutes a selection unit.

The target value generation unit 54d performs the target generation process, assumes that both the current observation value and the previous target value contain an error, and moves the previous target value according to the motion model. Calculate the optimum Kalman gain that minimizes the error. Then, the target value generation unit 54d adds the value obtained by multiplying the difference between the corresponding observed value and the predicted value by the optimum Kalman gain to the predicted value, and generates the current target value. The target target generation unit 54d constitutes a generation unit.

Next, the fourth filter processing unit 55 will be described with reference to FIG. FIG. 12 is a block diagram showing the configuration of the fourth filter processing unit 55.

The fourth filter processing unit 55 includes a prediction processing unit 55a, a gating processing unit 55b, an allocation unit 55c, a weighting unit 55d, a resampling unit 55e, and a target value generation unit 55f.

The fourth filter processing unit 55 performs the fourth tracking process on the target value whose tracking state is the definite track state. The fourth filter processing unit 55 generates a target value by filter processing using a particle filter.

The prediction processing unit 55a performs prediction processing, specifically, definite prediction processing, and predicts the predicted value this time. The prediction processing unit 55a predicts the distribution state in the current particle data from the distribution state in the previous particle data. Specifically, the prediction processing unit 55a samples particles based on the probability density function in the previous particle data, and moves each particle based on the velocity of the previous particle data.

The gating processing unit 55b performs gating processing, and similarly to the gating processing unit 52b of the first filter processing unit 52, a part of the detected observed values is excluded from the candidates for the corresponding observed values. The gating processing unit 55b excludes a part of the observed value based on the Mahalanobis distance, similarly to the gating processing unit 53b of the second filter processing unit 53.

The allocation unit 55c performs allocation processing, and allocates the observed values not excluded by the gating processing unit 55b to the predicted particle data of this time.

The weighting unit 55d performs weighting processing, and weights each of the current particle data based on the assigned observation value. The weighting unit 55d increases the weight of the particles close to the observed value in the particle data of this time, and decreases the weight of the particles far from the observed value of this time.

The resampling unit 55e performs resampling processing and rearranges (resampling) the particle data based on the weight of each of the particle data this time. The resampling unit 55e moves the particle data having a small weight closer to the observed value.

The target value generation unit 55f performs a target generation process and generates a target value based on the average of the current particle data rearranged by the resampling unit 55e. The target value generation unit 55f generates a target value based on the average of the probability density function. The target value generation unit 55f may generate a target value based on the median value of each parameter of the particle data, or may generate a target value based on the center of gravity of the probability density function. The target target generation unit 55f constitutes a generation unit.

Next, the tracking state setting process according to the embodiment will be described with reference to FIG. FIG. 13 is a flowchart illustrating the tracking state setting process. Here, it is assumed that the tracking state of the target is the wake state during the test. In addition, this process is executed for each target value.

The radar device 1 determines whether or not the survival counter of the target value is equal to or less than the erased value (S10). When the survival counter is equal to or less than the erased value (S10: Yes), the radar device 1 erases the target value (S11).

When the survival counter is larger than the erased value (S10: No), the radar device 1 determines whether or not the survival counter is equal to or greater than the predetermined counter value (S12).

When the survival counter is smaller than the predetermined counter value (S12: No), the radar device 1 ends the current process.

When the survival counter is equal to or higher than the predetermined counter value (S12: Yes), the radar device 1 sets the tracking state to the definite track state (S13). Then, the radar device 1 sets the tracking filter (S14).

Next, the tracking filter setting process according to the embodiment will be described with reference to FIG. FIG. 14 is a flowchart illustrating the tracking filter setting process.

The radar device 1 determines whether or not the average value of the difference between the corresponding observation value calculated by the processing in the first filter processing unit 52 and the predicted value is equal to or less than the first predetermined value (S20). When the average value is equal to or less than the first predetermined value (S20: Yes), the radar device 1 sets the tracking filter to the linear Kalman filter (S21).

When the average value is larger than the first predetermined value (S20: No), the radar device 1 determines whether or not the average value is equal to or less than the second predetermined value (S22). The radar device 1 sets a tracking filter when the average value is equal to or less than the second predetermined value (S22: Yes), that is, when the average value is larger than the first predetermined value and equal to or less than the second predetermined value. Set to the unscented Kalman filter (S23).

When the average value is larger than the second predetermined value (S22: No), the radar device 1 sets the tracking filter in the particle filter (S24).

Next, the first filter processing executed by the first filter processing unit 52 according to the embodiment will be described with reference to FIG. FIG. 15 is a flowchart illustrating the first filter processing.

The radar device 1 performs prediction processing and predicts the predicted value (S30). The radar device 1 performs a gating process and excludes a part of the observed values from the candidates for the corresponding observed values (S31).

The radar device 1 performs a connection process and selects a corresponding observation value (S32). The radar device 1 performs a target value generation process and generates a target value (S33).

In the second filter processing executed by the second filter processing unit 53 and the third filter processing executed by the third filter processing unit 54, the processing is performed in the same procedure. Therefore, the flowchart explaining the process is omitted.

Next, the fourth filter processing executed by the fourth filter processing unit 55 according to the embodiment will be described with reference to FIG. FIG. 16 is a flowchart illustrating the fourth filter processing.

The radar device 1 performs prediction processing and predicts the distribution state in the particle data this time (S40). The radar device 1 performs a gating process and excludes a part of the observed values from the candidates for the corresponding observed values (S41).

The radar device 1 performs the allocation process and allocates the observed values not excluded to the predicted particle data of the present time (S42). The radar device 1 performs weighting processing and weights each of the particle data this time (S43).

The radar device 1 performs resampling processing and rearranges the particle data (S44). The radar device 1 performs a target value generation process and generates a target value (S45).

In the above embodiment, when the tracking state for the target value is set to the definite tracking state, the tracking filter is set for the target, and the setting is made while the definite tracking state continues thereafter. The same tracking filter as the tracking filter was applied. However, even after the fixed tracking state is set, the tracking filter may be reset based on the difference between the predicted value and the corresponding observed value every time or for each predetermined number of scans. As a result, it is possible to set the tracking filter according to the target based on the difference between the predicted value and the corresponding observed value. Therefore, the tracking filter to be applied can be reviewed every time or every predetermined number of scans, and the tracking filter can be set according to the variation of the observed noise.

Next, the effect of the radar device 1 according to the embodiment will be described.

The radar device 1 sets a tracking filter based on the difference between the corresponding observed value and the predicted value predicted from the previous target value, specifically, the average value of the difference, and uses the set tracking filter. Generate a target value.

As a result, the radar device 1 can set the tracking filter according to the variation in the observed noise. Therefore, the radar device 1 improves the tracking performance of the target, suppresses the application of the tracking filter having a large processing load to the target having a small variation in the observed noise, and reduces the processing load. Can be done.

When the average value is equal to or less than the first predetermined value, the radar device 1 sets the tracking filter to the linear filter. Specifically, the radar device 1 sets the tracking filter to a linear Kalman filter.

As a result, the radar device 1 can perform tracking processing on a target with few irregular movements by using a linear Kalman filter having a smaller processing load than a non-linear filter, and can reduce the memory usage amount and the processing load. ..

Further, when the average value is larger than the first predetermined value and is equal to or less than the second predetermined value, the radar device 1 sets the tracking filter to the unscented Kalman filter which is a non-linear filter.

As a result, the radar device 1 can improve the tracking performance for targets that perform irregular operations, and can reduce the memory usage and the processing load as compared with the case of using the particle filter.

Further, when the average value is larger than the second predetermined value, the radar device 1 sets the tracking filter in the particle filter.

As a result, the radar device 1 can improve the tracking performance for a target having many irregular movements.

When the tracking state is the track state under verification, the radar device 1 has a smaller processing load than the tracking filter such as a nonlinear filter or a particle filter, and has the smallest processing load among the filters used in the radar device 1. Predict the predicted value using the αβ filter.

As a result, the radar device 1 can reduce the memory usage amount and the processing load. Further, the radar device 1 can ensure a certain tracking performance even for a target having a large observation error by presetting a gain coefficient (predetermined filter constant) used in the αβ filter as a parameter.

The radar device 1 excludes observation values whose relative speed difference from the previous corresponding observation value is larger than a predetermined speed difference from the candidates for the current observation value.

As a result, the radar device 1 can exclude the observed value corresponding to the noise from the candidates of the corresponding observed value, and can improve the tracking performance.

The radar device 1 excludes observation values whose received power difference from the previous corresponding observation value is larger than a predetermined power difference from the candidates for the current observation value.

As a result, the radar device 1 can exclude the observation value corresponding to the clutter from the candidates for the corresponding observation value, and can improve the tracking performance.

Further, the radar device 1 performs gating processing in each of the filter processing units 52 to 55 based on the relative speed difference from the previous corresponding observation value and the received power difference from the previous corresponding observation value.

As a result, the radar device 1 can improve the tracking performance in each filter processing.

The radar device 1 according to the modified example may erase the target value whose survival counter in the predetermined scan set in advance does not exceed the predetermined counter value when the tracking state is the wake state during the verification. As a result, the radar device 1 according to the modified example can quickly erase the target value corresponding to the noise, for example, when the observed value corresponding to the noise is generated as the target value.

In the radar device 1 according to the modified example, in each of the filter processing units 52 to 55, in addition to the gating process based on the distance (or Mahalanobis distance), the relative speed difference from the previous corresponding observation value and the previous corresponding observation. Only one of the received power differences from the value may be gated.

In the radar device 1 according to the embodiment, the filter processing units 51 to 55 have different configurations, but the filter processing units 51 to 55 may be integrated.

Further effects and variations can be easily derived by those skilled in the art. For this reason, the broader aspects of the invention are not limited to the particular details and representative embodiments described and described above. Accordingly, various modifications can be made without departing from the spirit or scope of the general concept of the invention as defined by the appended claims and their equivalents.

1 Radar device 44 Conversion unit 45 Peak extraction unit (detection unit)
46 Distance / Relative Velocity Calculation Unit (Detection Unit)
47 Angle estimation unit (detection unit)
48 Clustering processing unit (detection unit)
49 Tracking processing unit 51 Tracking processing setting unit (setting unit)
52 First filter processing unit 52b Gating processing unit (selection unit)
52d Target value generation unit 53 Second filter processing unit 53b Gating processing unit (selection unit)
53d Target value generation unit (generation unit)
54 Third filter processing unit 54b Gating processing unit (selection unit)
54d Target value generation unit (generation unit)
55 Fourth filter processing unit 55b Gating processing unit (selection unit)
55f Target value generation unit (generation unit)

Claims (7)

A detector that detects the observed value that peaks in the frequency spectrum based on the frequency-modulated transmitted wave and the reflected wave of the transmitted wave by the target.
A selection unit that selects the corresponding observation value corresponding to the target from the observation value, and
A setting unit that sets a tracking filter from a plurality of filters based on the difference between the corresponding observation value and the predicted value predicted from the previous target value.
A radar device including a generator that generates the target value of this time by using the tracking filter. The setting unit is
When the difference is larger than the first predetermined value, the tracking filter is set as a non-linear filter.
The radar device according to claim 1, wherein when the difference is equal to or less than the first predetermined value, the tracking filter is set as a linear filter. The setting unit is
When the difference is larger than the first predetermined value and equal to or less than the second predetermined value larger than the first predetermined value, the tracking filter is set in the nonlinear Kalman filter.
The radar device according to claim 2, wherein when the difference is larger than the second predetermined value, the tracking filter is set in the particle filter. The generator is
In the case of the in-testing wake state, which is the state before the target value is determined to be the tracking target, the above-mentioned current target value is generated using the in-testing prediction filter with the smallest processing load.
The setting unit is
When it is determined that the target value is the tracking target, the tracking filter is set based on the difference between the corresponding observation value and the predicted value predicted by using the prediction filter during the test. The radar device according to any one of claims 1 to 3, wherein the radar device is characterized. The selection unit is
Claims 1 to 4, wherein the corresponding observation value is selected by excluding the observation value whose relative speed difference from the previously selected corresponding observation value is larger than the predetermined speed difference from the observation value. The radar device described in any one. The selection unit is
Claims 1 to 5, wherein the corresponding observation value is selected by excluding the observation value in which the received power difference from the previously selected corresponding observation value is larger than the predetermined power difference from the observation value. The radar device described in any one. A detection step for detecting an observed value that peaks in a frequency spectrum based on a frequency-modulated transmitted wave and a reflected wave of the transmitted wave by a target, and a detection step.
A selection step of selecting a corresponding observation value corresponding to the target from the observation value, and
A setting process for setting a tracking filter from a plurality of filters based on the difference between the corresponding observation value and the predicted value predicted from the previous target value, and
A target detection method including a generation step of generating the target value of the present time using the tracking filter.

Images