JP7274271B2 - Radar device and target object discrimination method - Google Patents

Description

The disclosed embodiments relate to a radar device and a target object discrimination method.

Conventionally, a radar apparatus receives a reflected wave of a transmission wave reflected by a target as a received signal, and analyzes the received signal to detect a target such as a pedestrian (see, for example, Patent Document 1). .

JP 2016-3873 A

However, with the conventional technology, it is difficult to distinguish between pedestrians and road clutter with high accuracy.

One aspect of the embodiments has been made in view of the above, and an object thereof is to provide a radar device and a target object discrimination method that can discriminate between pedestrians and road clutter with high accuracy.

A radar device according to an aspect of an embodiment includes a receiver, a calculator, and a determiner. The receiver receives a reflected wave of the transmitted wave reflected by the target as a received signal. The calculation unit obtains an instantaneous value of the target from the received signal, and calculates a plurality of feature values related to the target from the instantaneous value. The determination unit determines the object based on the plurality of feature values calculated by the calculation unit and a likelihood model representing the likelihood of whether or not the object is a pedestrian corresponding to the values of the feature values generated in advance. It is determined whether or not the mark is a pedestrian.

According to the radar device and the target object discrimination method according to one aspect of the embodiment, it is possible to discriminate between pedestrians and road clutter with high accuracy.

FIG. 1 is an explanatory diagram showing an outline of a target object discrimination method according to an embodiment. FIG. 2 is a block diagram of the radar device according to the embodiment. FIG. 3 is a diagram showing an example of the relationship between the transmission frequency, reception frequency, and beat frequency according to the embodiment. FIG. 4 is a diagram showing the result of performing distance FFT processing on the beat signal according to the embodiment. FIG. 5 is a diagram illustrating processing contents of a second processing unit according to the embodiment; FIG. 6 is an explanatory diagram illustrating an example of a score calculation method according to the embodiment. FIG. 7 is an explanatory diagram illustrating an example of a distance model according to the embodiment; FIG. 8 is an explanatory diagram showing an example of a ground speed model according to the embodiment. 9 is a flowchart illustrating an example of processing executed by the signal processing unit of the radar device according to the embodiment; FIG.

Hereinafter, embodiments of a radar device and a target object discrimination method will be described in detail with reference to the accompanying drawings. In addition, this invention is not limited by embodiment shown below. FIG. 1 is an explanatory diagram showing an outline of a target object discrimination method according to an embodiment.

A case where the radar device according to the embodiment detects a target existing around the own vehicle by the FCM (Fast Chirp Modulation) method will be described below as an example, but this is just an example. The radar device according to the embodiment may detect a target by, for example, the FM-CW (Frequency Modulated Continuous Wave) method.

The FCM method outputs a transmission wave in which a plurality of chirp waves with continuously changing frequencies are repeated, receives the reflected wave of the transmission wave reflected by the target as a reception signal, and analyzes the reception signal to identify the target. This method detects the distance to the vehicle, the relative speed, and the angle (azimuth) to the vehicle.

As shown in FIG. 1, when the radar device detects a target T1 based on the received signal (step S1), the received signal detects characteristics such as the distance to the target and the relative speed (sometimes called instantaneous values). ), and a plurality of feature values T2 relating to the target are calculated from the instantaneous values of the target (step S2).

Subsequently, the radar device calculates a score indicating the possibility that the target T1 is a pedestrian based on the feature amount T2 (step S3). At this time, the radar device refers to a likelihood model M that models the likelihood (hereinafter referred to as likelihood) as to whether or not the pedestrian is a pedestrian corresponding to the value of the feature value generated in advance (step S31). .

The likelihood model M includes a pedestrian model indicated by a solid line graph in FIG. 1 and a road surface model indicated by a dotted line graph. The pedestrian model expresses the relationship of the probability (that is, likelihood) that the target is a pedestrian with respect to the feature value indicated by the target.

Likewise, the road surface model expresses the relationship between the feature value indicated by the target and the likelihood that the target is the road surface. Such a likelihood model M is generated in advance for each type of a plurality of feature amounts T2 calculated in step S2.

Based on the feature value T2 calculated in step S2 and the likelihood model M, the radar device calculates a pedestrian likelihood indicating the likelihood that the target is a pedestrian and a road surface indicating the likelihood that the target is a road surface. A likelihood is obtained (step S32).

At this time, the radar device acquires the pedestrian likelihood and the road surface likelihood for each type of the plurality of feature amounts T2 calculated in step S2. In step S3, the radar device calculates a score indicating the likelihood that the target T1 is a pedestrian for each type of feature quantity T2 from these pedestrian likelihood and road surface likelihood.

Then, when the calculated total score of the multiple scores is equal to or greater than the threshold value, the radar device determines that the pedestrian is T3 (step S4). Also, when the total score is less than the threshold, the radar device determines that the road surface clutter is T4 (step S5).

Thus, the radar device calculates a plurality of feature quantities T2 regarding the target T1 from the instantaneous values of the target. Then, the radar device determines whether the target T1 is walking based on the calculated plurality of feature values T2 and the likelihood model M representing the likelihood of whether or not the target T1 is a pedestrian corresponding to the values of the feature values generated in advance. It is determined whether or not the person is T3. As a result, the radar device can discriminate between pedestrians and road clutter with high accuracy.

Next, an example of the configuration of the radar device 1 according to the embodiment will be described with reference to FIG. 2 . FIG. 1 is a block diagram of a radar device 1 according to an embodiment. A radar device 1 is mounted on a vehicle and detects an object by the FCM method.

Specifically, the radar apparatus 1 receives reflected waves, which are transmission waves reflected by targets, as received signals by a plurality of receiving antennas. Then, the radar device 1 performs two-dimensional Fast Fourier Transform processing (hereinafter sometimes referred to as two-dimensional FFT processing) on the beat signal generated from the received reflected wave and the transmitted wave. Go and detect the distance to the target, the relative velocity, and the accuracy (bearing) of the target.

Note that the two-dimensional FFT processing is performed by performing two FFT processing, a distance FFT processing in the distance direction corresponding to the distance to the target and a speed FFT processing in the speed direction corresponding to the relative speed with the target. be.

Such a radar device 1 is connected to a vehicle control device 2 as shown in FIG. The vehicle control device 2 performs vehicle control such as PCS (Pre-crash Safety System) and AEB (Advanced Emergency Braking System) based on the target detection result by the radar device 1 . Note that the radar device 1 may be used for various purposes other than the vehicle-mounted radar device (for example, monitoring of airplanes and ships, etc.).

The radar device 1 includes a transmitter 10 , a receiver 20 and a processor 30 . The transmitter 10 includes a signal generator 11 , an oscillator 12 and a transmission antenna 13 . The signal generator 11 generates a modulated signal whose voltage changes in a sawtooth waveform, and supplies the modulated signal to the oscillator 12 . Oscillator 12 generates a chirp signal based on the modulated signal generated by signal generator 11 and outputs the chirp signal to transmission antenna 13 .

The transmitting antenna 13 converts the chirp signal input from the oscillator 12 into a transmission wave and outputs the transmission wave to the outside of the vehicle. A transmission wave output from the transmission antenna 13 has a waveform in which a plurality of chirp waves are repeated. A transmission wave transmitted forward of the vehicle from the transmission antenna 13 is reflected by the target and becomes a reflected wave.

The receiving section 20 includes a plurality of receiving antennas 21 forming an array antenna, a mixer 22 provided for each receiving antenna 21 , and an A/D converter 23 provided for each mixer 22 . Each receiving antenna 21 receives a reflected wave from a target as a received wave, converts the received wave into a received signal, and outputs the received signal to the mixer 22 . Although the number of receiving antennas 21 shown in FIG. 2 is four, it may be three or less or five or more.

A received signal output from each receiving antenna 21 is input to the mixer 22 after being amplified by an amplifier (for example, a low noise amplifier) not shown. The mixer 22 mixes the chirp signal and part of the received signal to remove unnecessary signal components, generates a beat signal, and outputs the beat signal to the A/D converter 23 .

As a _result , _{the beat frequency f SB ( = f ST} −f _SR ) is generated. The beat signal generated by the mixer 22 is converted to a digital signal by the A/D converter 23 and then output to the processing section 30 .

FIG. 3 is a diagram showing an example of the relationship among the transmission frequency f _ST , the reception frequency f _SR , and the beat frequency f _SB . As shown in FIG. 3, a beat signal is generated for each chirp wave. Here, the beat signal obtained by the first chirp wave is defined as "B1", the beat signal obtained by the second chirp wave is defined as "B2", and the beat signal obtained by the n-th chirp wave is defined as "Bn". ”.

In the example shown in FIG. 3, the transmission frequency f _ST increases with time from the reference frequency f0 with a slope θ (=(f1−f0)/Tm) for each chirp wave, and reaches the maximum frequency f1. It is a sawtooth wave (so-called up-chirp) that returns to the reference frequency f0 in a short period of time.

The transmission frequency _fST reaches the maximum frequency f1 from the reference frequency f0 for each chirp wave in a short time, and decreases with time from the maximum frequency f1 with a slope θ (=(f0−f1)/Tm). A sawtooth wave (so-called down-chirp) may be used.

Returning to the description of FIG. 2, the processing unit 30 will be described. The processing unit 30 includes a transmission control unit 31 and a signal processing unit 32 . The signal processing unit 32 includes a first processing unit 33 , a second processing unit 34 , a calculation unit 35 , a determination unit 36 and an output unit 37 .

The processing unit 30 is, for example, a microcomputer including a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), input/output ports, etc., and controls the radar device 1 as a whole.

The processing unit 30 includes a transmission control unit 31 and a signal processing unit 32 that function by reading a program stored in the ROM by the CPU of the microcomputer and executing it using the RAM as a work area.

At least a part or all of the transmission control unit 31 and the signal processing unit 32 can be configured by hardware such as ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

The transmission control unit 31 controls the signal generation unit 11 of the transmission unit 10 and causes the signal generation unit 11 to output to the oscillator 12 a modulated signal whose voltage changes in a sawtooth shape. As a result, a chirp signal whose frequency changes over time is output from the oscillator 12 to the transmitting antenna 13 .

The signal processing unit 32 performs two-dimensional FFT processing (distance FFT processing and speed FFT processing) on the beat signals output from each A/D converter 23 . Then, the signal processing unit 32 calculates the distance, relative speed, and direction of the target based on the result of the two-dimensional FFT processing, and determines, for example, whether the target is a pedestrian or not based on the calculated distance and relative speed. discriminate. Processing of each unit of the signal processing unit 32 will be described below.

The first processing unit 33 of the signal processing unit 32 generates a frequency spectrum for each receiving antenna 21 by performing distance FFT processing on each beat signal input from each A/D converter 23 . Specifically, the first processing unit 33 performs distance FFT processing for each distance [bin] fr (fr1 to frm) for each beat signal. Here, the results of the distance FFT processing will be specifically described with reference to FIG.

FIG. 4 is a diagram showing the result of performing distance FFT processing on the beat signal. In the frequency spectrum shown in FIG. 4, the horizontal axis is the distance [bin] (frequency), and the vertical axis is the magnitude (peak magnitude) of the power spectrum (power [dB]). In the example shown in FIG. 4, it is assumed that a peak appears only at the distance [bin]fr10.

Here, the frequency of the beat signal increases or decreases in proportion to the distance between the target and the radar device 1 . For this reason, the first processing unit 33 performs distance FFT processing on the beat signal so that the peak (power is equal to or greater than a predetermined value) appearing at the distance [bin]fr corresponding to the distance to the target is detected in the distance direction. as the target peak of

The first processing unit 33 periodically performs distance FFT processing on each beat signal input from the four A/D converters 23 in a predetermined cycle. The first processing unit 33 outputs the result of distance FFT processing to the second processing unit 34 .

The second processing unit 34 performs velocity FFT processing on the result of the distance FFT processing in the first processing unit 33 . The speed FFT processing is to perform the second FFT processing for each speed [bin]fv for each distance [bin]fr of the frequency spectrum that is the result of the distance FFT processing. As a result of the velocity FFT processing, a peak appears in the velocity [bin]fv corresponding to the relative velocity of the target.

The second processing unit 34 uses the Doppler component of the received signal that occurs when the relative velocity of the target is not zero. Specifically, the second processing unit 34 detects a change in phase of peaks in the frequency spectrum of the beat signal. Here, the processing contents of the second processing unit 34 will be described with reference to FIG.

FIG. 5 is a diagram showing the processing contents of the second processing unit 34. As shown in FIG. In FIG. 5 , the frequency spectrum of any one receiving antenna 21 out of the plurality of receiving antennas 21 is arranged in chronological order. FIG. 5 also shows an example of the results of distance FFT processing of temporally continuous beat signals B1 to B8 and phase changes of peaks between the beat signals B1 to B8. In the example shown in FIG. 5, there is a peak in the distance [bin]fr10 of each of the beat signals B1 to B8, and the phase of the peak changes.

Here, when the relative velocity between the target and the radar device 1 is not zero, a phase change corresponding to the Doppler frequency appears at the peak of the distance [bin]fr10 corresponding to the same target among the beat signals B1 to B8.

The second processing unit 34 arranges the frequency spectrum obtained by periodically performing distance FFT processing in a predetermined cycle in time series and performs velocity FFT processing, so that a peak appears in the frequency (velocity [bin]) relative to the Doppler frequency. Acquire the frequency spectrum for the target peak in the velocity direction. The second processing unit 34 outputs the result of the speed FFT processing to the calculation unit 35 .

Returning to FIG. 2, the calculation unit 35 and the determination unit 36 will be described. The calculation unit 35 generates a power spectrum in a two-dimensional orthogonal coordinate system with the distance direction as the first axis and the speed direction as the second axis from the instantaneous value of the result of the speed FFT processing input from the second processing unit 34. do.

Then, the calculator 35 calculates the distance, relative velocity, and angle (azimuth) to the target based on the calculated power spectrum. Further, the calculation unit 35 calculates the ground speed of the target by removing the speed component of the own vehicle from the relative speed with respect to the target. That is, the ground speed is obtained by subtracting a component obtained by projecting the speed vector of the own vehicle in the direction of the target from the magnitude of the relative speed of the target.

Note that the accuracy estimation by the calculation unit 35 is performed using a predetermined 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 determination unit 36 determines, for example, whether the target is a pedestrian or not based on the distance, relative speed, ground speed, and angle calculated by the calculation unit. The determination unit 36 then outputs the calculation result and the determination result to the output unit 37 .

The output unit 37 outputs various information to the vehicle control device 2 . For example, the output unit 37 outputs target information regarding the detected target to the vehicle control device 2 . The target information includes, for example, the distance, relative speed, ground speed, and angle of the target determined as a pedestrian by the determination unit 36 .

Here, the discrimination unit of a general radar device discriminates whether or not the target is a pedestrian based on the target peak in the distance direction and the target peak in the speed direction. For example, regarding the target peak in the distance direction, pedestrians tend to have higher power than road surface clutter. discriminate.
However, sometimes the power of the road clutter in the range direction exceeds the threshold. In such a case, the determination unit may erroneously determine that the road surface is a pedestrian.

In addition, regarding the target peak in the speed direction, the pedestrian moves but the road surface does not. do. However, such a discrimination unit may erroneously discriminate a pedestrian moving at a ground speed below the threshold as a road surface.

In this way, a typical discriminating unit discriminates whether or not the target is a pedestrian based on the target peak in the distance direction and the target peak in the speed direction. It was difficult to distinguish by

It should be noted that a general determination unit can improve the accuracy of determination by determining whether a target is a pedestrian or not based on target peaks that are continuously acquired in a plurality of cycles of processing. However, this increases the time required to determine whether the object is a pedestrian or not.

Therefore, the radar apparatus 1 does not simply compare the instantaneous value of the received signal with a threshold value to determine whether the target is a pedestrian, but instead calculates the characteristic value of the target from the instantaneous value of the received signal. and the likelihood model described above, it is determined whether or not the target is a pedestrian.

Specifically, the calculation unit 35 of the radar device 1 calculates the distance to the target and the ground speed of the target as feature values related to the target from the instantaneous value of the received signal, and calculates the calculated distance and ground speed. Output to the determination unit 36 . The determination unit 36 is a probabilistic classifier that determines whether the target is a pedestrian or not, for example, using Bayesian theory from the distance and ground speed input from the calculation unit 35 .

The determining unit 36 calculates a score indicating the likelihood that the target is a pedestrian (hereinafter referred to as a target score ) is calculated. Here, an example of the score calculation method according to the embodiment will be described with reference to FIG. 6 .

FIG. 6 is an explanatory diagram illustrating an example of a score calculation method according to the embodiment. As shown in FIG. 6, the likelihood model includes a pedestrian model indicated by a solid line graph and a road surface model indicated by a broken line graph.

The pedestrian model describes in advance the relationship between the feature amount and the likelihood that the value of the feature amount indicates that the person is a pedestrian. The road surface model describes in advance the relationship between the feature amount and the likelihood that the value of the feature amount is the road surface.

In the pedestrian model, the higher the likelihood corresponding to the feature quantity, the higher the probability that the target is a pedestrian. In the road surface model, the higher the likelihood corresponding to the feature quantity, the higher the probability that the target is the road surface.

For example _, when a feature value having a value of d shown in FIG. , the likelihood value θ ₂ (pedestrian likelihood) corresponding to the value d of the feature amount in the pedestrian model is obtained.

Then, the determining unit 36 divides the pedestrian likelihood θ ₂ by the road surface likelihood θ ₁ , for example, so that the higher the pedestrian likelihood and the lower the road surface likelihood, the larger the value of the target. Calculate the score.

In this way, the discrimination unit 36 does not evaluate only how pedestrian-like one feature value is, but combines the evaluation result of how pedestrian-like and the road-surface-like evaluation result. Calculate the target score. As a result, the determination unit 36 can calculate a highly reliable score as a material for determining whether or not the target is a pedestrian.

In addition, the distance to the target and the ground speed of the target are input from the calculation unit 35 to the determination unit 36 as feature amounts related to the target. Therefore, the determination unit 36 calculates the score of the target individually for the distance to the target and the ground speed of the target.

At this time, the determination unit 36 calculates the score of the target based on the distance to the target and the distance model, and calculates the score of the target based on the ground speed of the target and the ground speed model. The distance model has the aforementioned pedestrian model and road surface model corresponding to the distance value, and is a likelihood model representing the likelihood of whether or not it is a pedestrian corresponding to the distance value to the target.

The ground speed model is a likelihood model that has the above-mentioned pedestrian model and road surface model corresponding to the ground speed, and expresses the likelihood of whether or not it is a pedestrian corresponding to the ground speed value of the target. The distance model and the ground speed model are generated by statistics or the like in advance based on feature values taken by the target when the target is a pedestrian.

The likelihood model is modeled by a predetermined distribution function such as Gaussian distribution, and is continuously expressed using mathematical formulas. In addition, when the distribution shape is complicated, there is also a method of directly using an occurrence frequency histogram or the like of the feature quantity, describing it discretely, and complementing it. There are various methods for describing the likelihood model, and one can be selected as appropriate according to requirements.

Next, examples of a distance model and a ground speed model according to the embodiment will be described with reference to FIGS. 7 and 8. FIG. FIG. 7 is an explanatory diagram illustrating an example of a distance model according to the embodiment; FIG. 8 is an explanatory diagram showing an example of a ground speed model according to the embodiment.

Regarding the distance to the target, pedestrians can exist in the entire range in the distance direction as long as they are within the maximum detectable distance based on the characteristics of the radar device 1 . On the other hand, the road surface clutter tends to be distributed in a predetermined range relatively close to the radar device 1 in the distance direction depending on the vertical beam width of the transmitted wave and the mounting state of the radar device 1 .

Therefore, as shown in FIG. 7, the road surface model indicated by the dashed line graph has a Gaussian distribution in a relatively short range, and the pedestrian model indicated by the solid line graph has a uniform distribution in the distance direction. is designed to

On the other hand, regarding the ground speed of the target, pedestrians, including hands and feet, are distributed over a wider range in the ground speed direction than the road surface clutter due to the influence of micro-Doppler. On the other hand, since the ground speed of the road surface is 0, the road surface clutter tends to be distributed near 0 in the ground speed direction.

Therefore, as shown in FIG. 8, the ground speed model has a narrow Gaussian distribution near 0 for the road surface model indicated by the broken line graph, and a Gaussian distribution wider than the road surface model for the pedestrian model indicated by the solid line graph. is designed to be

The determination unit 36 calculates the score of the target based on the distance to the target input from the calculation unit 35 and the distance model shown in FIG. 7 in the same manner as described with reference to FIG. . Further, the determining unit 36 performs the same method as described with reference to FIG. 6 based on the ground speed of the target input from the calculating unit 35 and the ground speed model shown in FIG. Calculate the score of

Subsequently, the determination unit 36 adds the score calculated based on the distance to the target and the distance model, and the score calculated based on the ground speed of the target and the ground speed model, to obtain the final target. Calculate the score.

Then, the determination unit 36 determines that the target is a pedestrian when the score of the target is equal to or greater than a predetermined threshold. Further, when the score of the target is less than the predetermined threshold value, the discrimination unit 36 discriminates it as a road surface.

In this way, the determination unit 36 does not simply compare the distance to the target or the ground speed with the threshold, but calculates the pedestrian likelihood and the road surface likelihood for each of the distance to the target and the ground speed. Then, by comprehensively evaluating these two evaluations, it is determined whether or not the target is a pedestrian. Thereby, the discrimination unit 36 can discriminate between pedestrians and road clutter with high accuracy.

Next, an example of processing executed by the signal processing unit 32 of the radar device 1 according to the embodiment will be described with reference to FIG. 9 . FIG. 9 is a flowchart showing an example of processing executed by the signal processing unit 32 of the radar device 1 according to the embodiment.

When the beat signal is input, the signal processing section 32 repeatedly executes the processing shown in FIG. 9 in a predetermined cycle. Specifically, when the beat signal is input, the signal processing unit 32 first performs distance FFT processing on the beat signal (step S101), and then performs velocity FFT processing on the result of the distance FFT processing. (Step S102).

Subsequently, the signal processing unit 32 calculates the distance to the target and the ground speed of the target based on the processing result of the speed FFT (step S103). Then, the signal processing unit 32 calculates the score of the target based on the distance to the target and the distance model (step S104).

Subsequently, the signal processing unit 32 calculates a score based on the ground speed of the target and the ground speed model (step S105). Then, the signal processing unit 32 adds up the scores calculated in the processes of steps S104 and S105 to calculate a total score (step S106).

After that, the signal processing unit 32 determines whether or not the total score is equal to or greater than the threshold (step S107). When the signal processing unit 32 determines that the total score is equal to or greater than the threshold (step S107, Yes), the signal processing unit 32 determines that the target is a pedestrian (step S108), and shifts the process to step S110.

When the signal processing unit 32 determines that the total score is less than the threshold (step S107, No), the signal processing unit 32 determines that the road surface is cluttered (step S109), and shifts the process to step S110.

In step S110, the signal processing unit 32 outputs target information including the distance, relative speed, angle, etc., of the target determined as a pedestrian to the vehicle control device 2, and terminates the process. Start processing.

As described above, the radar device 1 according to the embodiment includes the calculator 35 and the determiner 36 . Then, the calculation unit 35 obtains the instantaneous value of the target T1 from the received signal, and calculates a plurality of feature amounts regarding the target T1 from the instantaneous value. The determination unit 36 determines whether the target is a pedestrian based on the plurality of feature values calculated by the calculation unit 35 and a likelihood model representing the likelihood of whether or not the target is a pedestrian corresponding to the value of the feature value generated in advance. It is determined whether the object is a pedestrian or not. As a result, the radar device 1 can discriminate between pedestrians and road clutter with high accuracy.

Further effects and modifications can be easily derived by those skilled in the art. Therefore, the broader aspects of the invention are not limited to the specific details and representative embodiments so shown and described. Accordingly, various changes may be made without departing from the spirit or scope of the general inventive concept defined by the appended claims and equivalents thereof.

Reference Signs List 1 radar device 2 vehicle control device 10 transmission unit 11 signal generation unit 12 oscillator 13 transmission antenna 20 reception unit 21 reception antenna 22 mixer 23 A/D converter 30 processing unit 31 transmission control unit 32 signal processing unit 33 first processing unit 34 Second processing section 35 Calculation section 36 Discrimination section 37 Output section

Claims (5)

a receiver that receives, as a received signal, a reflected wave of the transmitted wave reflected by the target;
a calculation unit that obtains an instantaneous value of the target from the received signal, and calculates a plurality of feature values related to the target from the instantaneous value;
The plurality of feature values calculated by the calculating unit, a pedestrian model representing the probability of whether or not the pedestrian corresponds to the value of the feature value generated in advance , and the value of the feature value generated in advance. calculating the score of the target based on a likelihood model including a road surface model representing the likelihood of whether the road surface corresponds to the target, and determining whether the target is a pedestrian or the road surface A radar device comprising: a determination unit that determines whether or not. The determination unit is
2. The radar device according to claim 1, wherein the discrimination is performed using a distance to the target and a ground speed of the target as the plurality of feature quantities. The determination unit is
A distance model that represents the likelihood of a pedestrian or a road surface corresponding to the distance value, and a likelihood model that represents the likelihood of a pedestrian or the road surface corresponding to the ground speed value 3. The radar device according to claim 2, wherein said discrimination is performed based on a certain ground speed model. The determination unit is
The radar device according to any one of claims 1 to 3 , wherein the discrimination is performed by a probabilistic classifier that receives the plurality of feature quantities as inputs. a receiving step of receiving, as a received signal, a reflected wave of the transmitted wave reflected by the target;
a calculation step of obtaining an instantaneous value of the target from the received signal, and calculating a plurality of feature values related to the target from the instantaneous value;
The plurality of feature values calculated by the calculating step, a pedestrian model representing the probability of whether or not the pedestrian corresponds to the value of the feature value generated in advance, and the value of the feature value generated in advance. calculating the score of the target based on a likelihood model including a road surface model representing the likelihood of whether the road surface corresponds to the target, and determining whether the target is a pedestrian or the road surface A target discrimination method comprising: a discrimination step of discriminating whether or not.

Images