JP7119628B2 - Target object detection method and device for vehicle - Google Patents

Description

The present invention relates to a technology for detecting targets around a vehicle using radar transmitted from the vehicle.

2. Description of the Related Art In recent years, from the viewpoint of improving vehicle safety, an increasing number of vehicles are equipped with a driving support function that recognizes obstacles around the vehicle using radar and takes avoidance measures such as automatic braking as necessary. In such a vehicle with a driving support function, if there are both pedestrians and other obstacles (for example, other vehicles, guardrails, etc.) around the vehicle, the two must be recognized separately ( In particular, it is desirable to recognize pedestrians by distinguishing them from other obstacles, and various techniques for that purpose have been proposed.

For example, in Patent Document 1 below, a signal acquisition means for radiating radar from the vehicle and receiving the reflected signal, and performing FFT processing on the received reflected signal, the frequency derived from the reflected signal from the pedestrian spectrum signal acquisition means for obtaining a discrete first spectrum signal containing a component and a frequency component derived from a reflected signal from a preceding vehicle; spectral signal calculating means for obtaining a discrete second spectral signal indicating the frequency component of the preceding vehicle by a predetermined calculation; spectral signal subtracting means for subtracting the second spectral signal from the first spectral signal; A radar apparatus is disclosed that includes a distance/velocity calculation means for detecting pedestrians (calculating the relative distance and relative velocity between the own vehicle and pedestrians) based on the subsequent spectral signals.

JP 2016-102675 A

In the above Patent Document 1, spectral data (second spectral signal) of another vehicle, which is theoretically (by calculation) obtained, is subtracted from spectral data (first spectral signal) obtained by subjecting a reflected signal to FFT processing. Therefore, even in a situation where other vehicles and pedestrians coexist around the own vehicle, pedestrians can be accurately detected based on the data after the subtraction.

In addition to other vehicles, structures such as guardrails, utility poles, and signs (hereinafter collectively referred to as road structures) exist on the road on which the host vehicle travels. Such road structures are smaller in size than vehicles and have less flat areas. Therefore, the reflected signal from the road structure tends to have a significantly lower signal strength than the reflected signal from the other vehicle. Reflected signals from road structures of this nature are similar to reflected signals from pedestrians in the sense that their intensity is relatively low. Therefore, even if the spectral data of other vehicles are subtracted as described above, it is difficult to determine whether the remaining frequency components are derived from pedestrians or road structures. be. Thus, it is considered that the method of Patent Document 1 does not improve the pedestrian detection accuracy as expected.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a vehicle target detection method and apparatus capable of detecting pedestrians separately from road structures.

In order to solve the above-mentioned problems, the present invention provides a transmitting unit that transmits radar, a receiving unit that receives a reflected radar signal from the transmitting unit, and a processing unit that analyzes the signal received by the receiving unit. A method of detecting a target around the own vehicle using the radar transmitted from the transmitting unit of the own vehicle, comprising: a receiving step in which the receiving unit receives a reflected signal of the radar; a data generating step in which the processing unit converts the signal received in the receiving step into polar format data and generates phase time-series data by extracting phase information in the polar format data over a predetermined period ; A processing unit, from the phase time series data generated in the data generation step, calculates the variance of the time series data and the dynamic range that is the difference between the maximum value and the minimum value of the time series data, and calculates When the calculated variance is larger than a predetermined first threshold and the calculated dynamic range is larger than a predetermined second threshold, it is determined that a pedestrian exists around the vehicle, and the variance is equal to or less than the first threshold value or the dynamic range is equal to or less than the second threshold value, a determination step of determining that a road structure exists around the vehicle. There is (claim 1).

Reflected signals from road structures such as guardrails, utility poles, and signs are similar to reflected signals from pedestrians in the sense that the signal strength is relatively low (e.g., much smaller than reflected signals from other vehicles). ing. For this reason, it is difficult to distinguish pedestrians from road structures by directly analyzing the reflected signals of the received radar .

On the other hand , when the radar reflection signal obtained when the target is a pedestrian is converted into polar format data, it has characteristics that are clearly different from the polar format data when the target is a road structure. I know that. Utilizing this, in the present invention, the received signal is converted into polar format data, phase time-series data is generated from the converted polar format data, and the target object is detected based on the generated time-series data. is determined. As a result, it is possible to clearly distinguish pedestrians and road structures, both of which tend to have weak reflected signals, and to effectively improve pedestrian detection accuracy.

Here, as more detailed knowledge about the time-series data of the phase, the variance of the time-series data obtained when the target is a pedestrian is larger than that when the target is a road structure. I know it will be Also, it is known that the dynamic range ( difference between the maximum value and the minimum value ) of the time-series data is larger than that when the target is a road structure. This is due to the characteristics of humans as pedestrians, that is, the surface shape is a complex uneven shape unique to living organisms, and the body is composed of various substances such as skin, muscles, internal organs, and bones. It is considered that

Therefore, in the present invention, it is determined that a pedestrian exists around the vehicle when the variance is greater than the first threshold and the dynamic range is greater than the second threshold. Thereby, a pedestrian can be detected accurately.

In the target detection method of the present invention, preferably, the processing unit performs predetermined processing including Fourier transform on the signal received in the receiving step to obtain the relationship between the distance from the own vehicle and the signal strength. a first preparation step of generating a distance spectrum representing the When a peak waveform having a peak falling within the predetermined intensity range is confirmed, the processing unit performs predetermined processing including inverse Fourier transform on distance spectrum data in a predetermined distance range including the peak position of the peak waveform. and a second preparation step of converting into time-series polar format data by, in the data generation step, based on the polar format data generated for the predetermined distance range by the second preparation step, the time of the phase Generate series data ( claim 2 ).

According to this configuration, conversion processing to polar format data is performed only on the portion of the peak waveform that is considered to be derived from pedestrians (or road structures) in the data of the distance spectrum, thus reducing the processing burden. can be effectively reduced. In addition, since phase time-series data can be examined only for polar format data suspected of originating from pedestrians, it is possible to eliminate the possibility of misjudgment caused by processing extra data, thereby eliminating the possibility of pedestrian presence or absence. can be determined more accurately.

Further, the present invention is a device for detecting a target around the own vehicle using radar transmitted from the own vehicle, comprising: a transmitter for transmitting the radar; and a reflected signal of the radar from the transmitter. and a processing unit for determining the presence or absence of a pedestrian by analyzing the signal received by the receiving unit, wherein the processing unit converts the signal received by the receiving unit into polar format data. a data generation process for generating phase time-series data obtained by extracting phase information in the polar format data over a predetermined period; and generating the phase time-series data generated by the data generation process . Calculate the variance and the dynamic range that is the difference between the maximum value and the minimum value of the time-series data, and the calculated variance is greater than a predetermined first threshold, and the calculated dynamic range is predetermined If it is larger than the second threshold value, it is determined that there is a pedestrian around the vehicle, and if the variance is the first threshold value or less or the dynamic range is the second threshold value or less, ( Claim 3 ).

According to this target detection device, similarly to the invention of the target detection method described above, pedestrians can be clearly distinguished from road structures, and the pedestrian detection accuracy can be improved.

As described above, according to the present invention, pedestrians can be detected separately from road structures.

1 is a schematic plan view showing a preferred embodiment of a vehicle to which a target object detection method (or device) of the present invention is applied; FIG. 3 is a block diagram showing a control system of the vehicle; FIG. It is a flowchart which shows the first half part of the pedestrian detection process performed while the said vehicle is driving|running|working. It is a flow chart which shows the second half part of the above-mentioned pedestrian detection processing. FIG. 4 is an explanatory diagram for explaining the concept of IQ data; It is a figure which shows an example of the distance spectrum obtained by the process of step S6 of FIG. FIG. 5 is a diagram showing an example of IQ data obtained by the process of step S11 in FIG. 4, where (a) shows the quadrature component of the IQ data and (b) shows the in-phase component of the IQ data. 5 is a diagram showing another example of IQ data obtained by the process of step S11 in FIG. 4, where (a) shows the quadrature component of the IQ data and (b) shows the in-phase component of the IQ data; FIG. FIG. 7A is a diagram showing phase time-series data generated through step S12 in FIG. 4 from the IQ data shown in FIGS. 7A and 7B, and the time-series data obtained when the target is a pedestrian It is a figure which shows an example of data. FIG. 8A is a diagram showing phase time-series data generated from the IQ data shown in FIGS. 8A and 8B through step S12 in FIG. It is a figure which shows an example of series data.

FIG. 1 is a schematic plan view showing a preferred embodiment of a vehicle to which the target object detection method (or device) of the present invention is applied. The vehicle shown in this figure comprises a plurality of radar units 1 (1A to 1E) and a plurality of camera units 2 (2A, 2B) for detecting obstacles around the vehicle.

The plurality of radar units 1 are a first radar unit 1A arranged at the center of the front end of the vehicle, a second radar unit 1B arranged at the left end of the front end of the vehicle, and a right end of the front end of the vehicle. a fourth radar unit 1D arranged at the left end of the rear end of the vehicle; and a fifth radar unit 1E arranged at the right end of the rear end of the vehicle. The first radar unit 1A is arranged, for example, on the front grille, the second and third radar units 1B, 1C are arranged, for example, on the rear surface of the front bumper face, and the fourth and fifth radar units 1D, 1E are arranged, for example, on the rear bumper. Located on the back of the face.

The first to fifth radar units 1A to 1E respectively include a transmitter 5 that transmits radar (for example, millimeter wave radar) toward the surroundings of the vehicle, and a reflected radar signal transmitted from the transmitter 5, that is, the surroundings. and a receiver 6 for receiving the radar reflected by an obstacle or the like and returned to the own vehicle (see FIG. 2 for both). Each of the radar units 1A to 1E is a radar detector whose detection range is a fan-shaped area centered on itself. By arranging them, almost the entire circumference of the vehicle is made the detection range of the radar. In the following description, the first to fifth radar units 1A to 1E are simply referred to as the radar unit 1 when they are referred to without particular distinction.

The plurality of camera units 2 include a first camera unit 2A that captures an image of the front of the vehicle through the windshield from inside the vehicle, and a second camera unit 2B that captures an image of the rear of the vehicle from the rear end of the vehicle. The first camera unit 2A is arranged, for example, at the upper end of the windshield on the inside of the vehicle interior, and the second camera unit 2B is arranged, for example, at the rear gate.

Although illustration is omitted, the first and second camera units 2A and 2B each have a lens portion, an image sensor that converts light incident through the lens portion into an electrical signal, and a signal that is processed from the image sensor. and a data processing unit for outputting as image data. Each of the camera units 2A and 2B is an imaging device whose imaging range is a fan-shaped area centered on itself. A relatively wide-angle range in front and rear is taken as an imaging range by the camera. In the following description, when referring to the first and second camera units 2A and 2B without distinguishing them, they will simply be referred to as the camera unit 2. As shown in FIG.

FIG. 2 is a block diagram showing the vehicle control system of this embodiment. A controller 10 shown in this figure is a microprocessor for controlling each part of the vehicle, and is composed of a well-known CPU, RAM, ROM and the like.

The controller 10 is electrically connected to the radar unit 1 and camera unit 2 and transmits and receives various signals to and from each unit 1 and 2 . For example, the controller 10 outputs a predetermined control signal to the transmitter 5 of the radar unit 1 and receives data representing a radar reflected signal from the receiver 6 of the radar unit 1 . The controller 10 also outputs a predetermined control signal to the camera unit 2 and receives image data from the camera unit 2 .

The controller 10 functionally has a processing unit 11 and a storage unit 12 . The processing section 11 controls each operation of the radar unit 1 and the camera unit 2 and performs various analyzes and judgments based on the data input from each unit 1 and 2 . The storage unit 12 stores various data necessary for processing by the processing unit 11 .

3 and 4 are flow charts showing specific procedures of pedestrian detection processing executed by the controller 10 as described above. Although this flowchart shows a process of detecting pedestrians using radar, in the vehicle of the present embodiment, in addition to the processes shown in this flowchart, a camera is used to detect pedestrians, other vehicles, or roads. It is also possible to execute processing for detecting structures (guardrails, utility poles, signs, etc.) and detecting other vehicles or road structures using radar. In other words, it can be said that this flowchart shows processing for detecting a pedestrian, which may be overlooked by the camera, using the radar. Further, the processing of this flowchart is performed individually for each of the first to fifth radar units 1A to 1E. , and the first to fifth radar units 1A to 1E will be collectively referred to simply as "radar unit 1".

When the process shown in FIG. 3 starts, the processing section 11 of the controller 10 causes the transmitting section 5 of the radar unit 1 to transmit radar in step S1.

Next, the processing unit 11 shifts to step S2 to transmit the reflected radar signal transmitted from the transmitting unit 5, that is, the radar signal reflected around the own vehicle and received by the receiving unit 6 of the radar unit 1. take in.

Next, the processing unit 11 proceeds to step S3 and performs AD conversion for converting the data fetched from the receiving unit 6 from analog data to digital data.

Next, the processing unit 11 proceeds to step S4 and generates IQ data based on the data after AD conversion in step S3. The IQ data is polar format (polar coordinate format) data in which a time-series signal is expressed as a time function of the amplitude, frequency, and phase of a sine wave, and can be expressed by the following formula (1).

Here,
A(t) is the amplitude of the sine wave,
f(t) is the frequency;
φ(t) is the phase, each a function of time t.

The sine wave expressed by Equation (1) above can be expressed in polar form using complex numbers (I+jQ), as shown in FIG. In FIG. 5, the projection value of the real axis (I-axis) A(t)·cos(φ(t)) is Vi, and the projection value of the imaginary axis (Q-axis) A(t)· Let sin(φ(t)) be Vq. Note that Vi is called an in-phase component and Vq is called a quadrature component.

In step S4, the processing unit 11 performs predetermined arithmetic processing on the data (digitized input signal) after the AD conversion in step S3, thereby obtaining the time functions of the above-described in-phase component Vi and quadrature component Vq. specified and retained as IQ data.

Next, the processing unit 11 proceeds to step S5, and Fourier transforms the IQ data obtained by the processing of step S4 to generate a frequency spectrum indicating the relationship between frequency and signal strength (amplitude) FFT Execute the process.

Next, the processing unit 11 proceeds to step S6, and generates a distance spectrum representing the relationship between the distance from the own vehicle and the signal intensity based on the frequency spectrum obtained by the process of step S5. That is, since there is a one-to-one correlation between the frequency and the distance that is determined according to the radar transmission/reception system (for example, the FMCW system), by performing a predetermined arithmetic processing based on this correlation, the above-mentioned The distance spectrum can be generated from the frequency spectrum.

FIG. 6 shows an example of the distance spectrum obtained by the process of step S6. As shown in this figure, the distance spectrum is obtained as data showing continuous changes in signal strength (received power) when the distance from the vehicle changes continuously.

Next, the processing unit 11 proceeds to step S7, and the distance spectrum obtained by the processing of step S6 includes a peak waveform having a peak within the predetermined intensity range W (FIG. 6) shown in FIG. Determine if it exists. The predetermined intensity range W is predetermined based on the assumption of a peak waveform obtained when the signal received by the radar unit 1 (that is, the input signal on which the distance spectrum is based) includes a reflected signal from a pedestrian. . That is, it is empirically known that the maximum value (peak) of the peak waveform falls within the predetermined intensity range W when the peak waveform based on the reflected signal from the pedestrian exists in the distance spectrum. In other words, the presence of a peak within the predetermined intensity range W means that pedestrians may be present around the vehicle. For example, the distance spectrum shown in FIG. 6 has two peak waveforms having peaks within a predetermined intensity range W (indicated by symbols A1 and A2, respectively). (Positions on the horizontal axis corresponding to peaks) are about 10 m and about 12 m. This indicates that there is a possibility that a pedestrian exists at a distance of about 10 m and a distance of about 12 m from the own vehicle. Since the intensity of the reflected signal decreases as the distance from the own vehicle increases, the upper limit value W1 and the lower limit value W2 of the predetermined intensity range W increase as the distance from the own vehicle increases (to the right in FIG. 6). ) is set to be small.

Here, the intensity of reflected signals from pedestrians tends to be significantly smaller than that of reflected signals from vehicles. The predetermined intensity range W is predetermined according to such characteristics of the reflected signal from the pedestrian. That is, the upper limit W1 of the predetermined intensity range W is set to a value sufficiently smaller than the peak value obtained when the reflector (target) is a vehicle. In the distance spectrum illustrated in FIG. 6, there are two peak waveforms having peaks within the predetermined intensity range W (marks A1 and A2), and the predetermined intensity range W is set to the above size. Therefore, it can be said that these two peak waveforms A1 and A2 are not derived from reflected signals from the vehicle. If there are other vehicles around the subject vehicle, the distance spectrum should include a peak larger than the predetermined intensity range W (its upper limit value W1). No such peak is seen. This means that there are no other vehicles around the own vehicle.

On the other hand, the intensity of reflected signals from road structures such as guardrails, utility poles, and signs tends to be much lower than reflected signals from other vehicles, as is the case with pedestrians. This is probably because, unlike a box-shaped vehicle, a road structure has a small flat area that strongly reflects radar. For this reason, it is difficult to distinguish between pedestrians and road structures simply by comparing signal intensities. For example, the distance spectrum shown in FIG. 6 has two peak waveforms having peaks within a predetermined intensity range W (marks A1 and A2). It cannot be determined whether the signal originates from the reflected signal or from the reflected signal from the road structure only by examining the signal intensity. Therefore, when it is determined YES in step S7, that is, when it is confirmed that there is at least one peak waveform having a peak within the predetermined intensity range W, the processing unit 11 determines that the peak waveform is a pedestrian. Alternatively, in order to determine which of the road structures it originates from, the processing shown in the flow chart of FIG. 4 follows.

4, first, in step S10, the processing unit 11 determines a target range (predetermined distance range) for inverse Fourier transform to be performed in step S11, which will be described later. Specifically, the processing unit 11 targets peak waveforms of a predetermined intensity (peak waveforms A1 and A2 whose peaks fall within a predetermined intensity range W) as shown in FIG. The upper position is specified as the peak position, and a relatively narrow distance range including at least this peak position is determined as the predetermined distance range. For example, the processing unit 11 determines the predetermined distance range from a position 30 cm before the peak position to a position 30 cm away from the peak position.

In the distance spectrum illustrated in FIG. 6, the peak waveform A1 having a peak position of about 10 m is defined as a first peak waveform, and the peak waveform A2 having a peak position of about 12 m is defined as a second peak waveform. In this case, the predetermined distance ranges are individually set for the first and second peak waveforms A1 and A2. For example, assuming that the predetermined distance range set for the first peak waveform A1 is Z1, this predetermined distance range Z1 is approximately 10±0.3 m. Similarly, assuming that the predetermined distance range set for the second peak waveform A2 is Z2, this predetermined distance range Z2 is approximately 12±0.3 m.

Next, the processing unit 11 proceeds to step S11, and transforms the distance spectrum data in the predetermined distance range (for example, Z1 and Z2 in FIG. 6) determined in step S10 into IQ data by inverse Fourier transform. Specifically, the processing unit 11 converts the distance spectrum data (that is, the data indicating the relationship between the distance from the host vehicle and the signal strength) in the predetermined distance range to the frequency spectrum data indicating the relationship between the frequency and the signal strength. , and subjecting the data after this transformation (frequency spectrum data) to an inverse Fourier transform to restore the IQ data (see FIG. 5 and formula (1) described above), which is time-series data in polar format. . Note that, as in the example shown in FIG. 6, when two predetermined distance ranges Z1 and Z2 corresponding to the two peak waveforms A1 and A2 are specified, the IQ data after the inverse Fourier transform is the predetermined distance range The first IQ data corresponding to Z1 and the second IQ data corresponding to the predetermined distance range Z2 are respectively restored.

FIGS. 7A and 7B are time-series data of the quadrature component Vq and the in-phase component Vi of the IQ data generated corresponding to the second peak waveform A2 (predetermined distance range Z2) shown in FIG. 8A and 8B show the quadrature component Vq and the in-phase component Vi of the IQ data generated corresponding to the first peak waveform A1 (predetermined distance range Z1) shown in FIG. Each time-series data is shown. As already described, the orthogonal component Vq of the IQ data is the projection value (=A(t)·sin(φ(t))) of the imaginary axis (Q axis) of the polar coordinates, and the in-phase component Vi is a projection value (=A(t)·cos(φ(t))) of the imaginary axis (I-axis) of the polar coordinates.

Next, the processing unit 11 proceeds to step S12 and generates phase time-series data based on the IQ data obtained in step S11. The phase time-series data is the phase time function φ(t) represented by the following equation (2).

FIG. 9 is generated corresponding to the phase φ(t) calculated from the IQ data shown in FIGS. 7A and 7B, that is, the second peak waveform A2 (predetermined distance range Z2) in FIG. 10 shows the time-series data of the phase φ(t) calculated by the above formula (2) from the IQ data (its Vq and Vi) obtained, and FIG. 10 shows the IQ data shown in FIGS. From the phase φ(t) calculated from the data, that is, the IQ data (its Vq and Vi) generated corresponding to the first peak waveform A1 (predetermined distance range Z1) in FIG. 4 shows time-series data of the phase φ(t) calculated by . The phase graphs in FIGS. 9 and 10 are shown with the horizontal axis (time axis) enlarged as compared to the quadrature/in-phase component graphs in FIGS. 7 and 8 .

Next, the calculation unit 11 proceeds to step S13 to calculate the dispersion and dynamic range of the phase time-series data obtained in step S12.

Here, the variance is an index indicating the degree of data scattering, and is synonymous with the general statistical definition of variance. That is, the variance is σ ² in the following formula (3), and the average of the square of the difference between the average value (μ) of the multiple data that constitutes the population and the numerical value (x _i ) of each data It is what I did.

Here,
n is the total number of data in the population,
x _i is the numerical value of each data μ is the average value of the data.

Also, the dynamic range is the difference between the maximum value and the minimum value of population data.

For example, when the data shown in the graph of FIG. 9 or 10 is obtained as phase time-series data, the calculation unit 11 calculates the range indicated by symbol T in the graph A large number of data values included in this effective range T are examined for the portion excluding S1 and S2), and the variance and dynamic range of the data are calculated.

Next, the calculation unit 11 proceeds to step S14 and executes a process of comparing the dispersion and dynamic range of the phase time-series data obtained in step S13 with a predetermined threshold value. Specifically, the calculation unit 11 determines whether the variance is greater than the first threshold value α and the dynamic range is greater than the second threshold value β.

The first threshold α and the second threshold β are phase time-series data typically obtained when the target (reflector) to be detected is a pedestrian (hereinafter referred to as pedestrian model data). ) is set in advance. That is, when comparing the phase time-series data obtained when the target is a pedestrian and the phase time-series data obtained when the target is a road structure, empirically, the target is It has been found that both phase and dynamic range are greater when the target is a pedestrian than when the target is a road structure. Using this tendency, the first threshold value α and the second threshold value β are set to values capable of distinguishing between pedestrians and road structures (that is, standard values for pedestrians and standard values for road structures). values) are set respectively. In other words, if the variance is greater than the first threshold value α and the dynamic range is greater than the second threshold value β, the obtained phase time-series data and the pedestrian model data have a high degree of agreement, and the target is a pedestrian. It means that the prediction that it is is established. The first threshold value α and the second threshold value β are set to smaller values as the distance from the host vehicle to the target (the distance specified from the peak position of the distance spectrum shown in FIG. 6) increases.

If it is determined as YES in step S14 and it is confirmed that the variance is greater than the first threshold value α and the dynamic range is greater than the second threshold value β, that is, the matching degree between the phase time-series data and the pedestrian model data is high, the processing unit 11 determines that there are pedestrians around the vehicle (step S15). For example, it is assumed that the data shown in FIG. 9, that is, the data distributed over a relatively wide range along the vertical axis, is obtained as the phase time-series data. In this case, the variance and the dynamic range are larger than the thresholds α and β, respectively, so the processing unit 11 accordingly determines that there are pedestrians around the vehicle. The data of FIG. 9 is based on the second peak waveform A2 shown in FIG. 6, that is, the data of the peak waveform A2 having a peak at a distance of about 12 m. Therefore, in this case, it is determined that the pedestrian exists at a position about 12 m away from the vehicle.

On the other hand, when it is determined to be NO in step S14 and at least one of the variance is equal to or less than the first threshold value α and the dynamic range is equal to or less than the second threshold value β, that is, the phase time series When it is confirmed that the degree of matching between the data and the pedestrian model data is low, the processing unit 11 determines that there is a road structure around the own vehicle (step S16). For example, it is assumed that the data shown in FIG. 10, that is, the data distributed in a relatively narrow range along the vertical axis, is obtained as the phase time-series data. In this case, the variance and the dynamic range are equal to or less than the thresholds α and β, respectively, and accordingly the processing unit 11 determines that there are road structures around the vehicle. The data of FIG. 10 is based on the first peak waveform A1 shown in FIG. 6, that is, the data of the peak waveform A1 having a peak at a distance of about 10 m. Therefore, in this case, it is determined that the road structure exists at a position about 10 m away from the host vehicle.

As described above, in this embodiment, the received radar reflection signal is converted into polar format IQ data, and phase information in the IQ data is extracted over a predetermined period to generate phase time-series data. (See FIGS. 9 and 10). Then, the variance and dynamic range (the difference between the maximum value and the minimum value) of the time-series data of this phase are examined, and if the variance is greater than the first threshold value α and the dynamic range is greater than the second threshold value β, It is determined that there are pedestrians around. According to such a configuration, pedestrians can be detected separately from road structures, and there is an advantage that pedestrian detection accuracy can be improved.

In other words, reflected signals from road structures such as guardrails, utility poles, and signs have relatively low signal strength (e.g., much smaller than reflected signals from other vehicles). Similar. For this reason, it is difficult to distinguish pedestrians from road structures by directly analyzing the received reflected radar signals. On the other hand, in the above-described embodiment, the received signal is converted into polar format IQ data, and phase time-series data is generated from the converted IQ data. By checking the range, pedestrians can be detected clearly distinguishable from road structures.

Here, if the radar reflection signal obtained when the target is a pedestrian is converted into IQ data, the dispersion and dynamic range in the time series data of the phase of the IQ data will be different if the target is a road structure. It is known to be larger than that in the case of This is due to the characteristics of humans as pedestrians, that is, the surface shape is a complex uneven shape unique to living organisms, and the body is composed of various substances such as skin, muscles, internal organs, and bones. It is considered that Thus, in consideration of the tendency of the phase time-series data when the target is a pedestrian, in the above embodiment, the variance and the dynamic range in the phase time-series data are examined, and the variance is the first threshold. When it is confirmed that the dynamic range is larger than α and the dynamic range is larger than the second threshold β, that is, when the target is a pedestrian, typical phase time-series data (pedestrian model data) and is high, it is determined that there is a pedestrian. Therefore, it is possible to clearly distinguish pedestrians and road structures, both of which tend to have weak reflected signals, and to effectively improve pedestrian detection accuracy.

In particular, in the above-described embodiment, the received radar reflection signal is subjected to predetermined processing including Fourier transform, resulting in a continuous distance spectrum (Fig. 6) is generated, and when a peak waveform (A1, A2) having a peak within a predetermined intensity range W is confirmed in this distance spectrum, a predetermined distance range (Z1, Z2) is converted into time-series IQ data by predetermined processing including inverse Fourier transform. Based on the IQ data generated for this predetermined distance range (Z1, Z2), the above-described phase time-series data is generated (FIGS. 9 and 10). According to such a configuration, the conversion processing to IQ data is performed only on the peak waveform part considered to be derived from the pedestrian (or road structure) in the distance spectrum data, so the processing load is reduced. can be effectively mitigated. In addition, since the phase time-series data can be examined only for IQ data suspected to be derived from pedestrians, the possibility of misjudgment due to processing extra data can be eliminated, and the presence or absence of pedestrians can be eliminated. More accurate determination can be made.

Although the preferred embodiments of the present invention have been described above, the present invention should not be construed as being limited to these embodiments, and various modifications can be made without departing from the scope of the present invention.

For example, in the above embodiment, phase time-series data (FIGS. 9 and 10) are generated based on polar format data (IQ data) obtained from radar reflected signals, and variance and dynamics in this phase time-series data are generated. The range (the difference between the maximum value and the minimum value) is examined, and when the variance is greater than the first threshold value α and the dynamic range is greater than the second threshold value β, it is determined that there are pedestrians around the vehicle. Pedestrians can be detected based on the degree of matching between time-series data and pedestrian model data (phase time-series data typically obtained when the target is a pedestrian). The method is not limited to the method of the above embodiment. For example, a pedestrian may be detected based on only one of the variance and the dynamic range, or another index that is neither the variance nor the dynamic range may be calculated, and the pedestrian may be detected based on the other index. may be detected. Additionally, pedestrians may be detected based on more than two metrics, including variance and dynamic range and such other metrics.

5 transmitter 6 receiver 11 processor W predetermined intensity range Z1 predetermined distance range Z2 predetermined distance range

Claims (3)

transmitted from the transmitting unit of the own vehicle, which includes a transmitting unit that transmits radar, a receiving unit that receives a reflected radar signal from the transmitting unit, and a processing unit that analyzes the signal received by the receiving unit A method of detecting a target around the own vehicle using a radar that
a receiving step in which the receiving unit receives the reflected signal of the radar;
a data generating step in which the processing unit converts the signal received in the receiving step into polar format data and generates phase time-series data by extracting phase information in the polar format data over a predetermined period;
The processing unit, from the phase time-series data generated in the data generation step, calculates the variance of the time-series data and the dynamic range, which is the difference between the maximum value and the minimum value of the time-series data, When the calculated variance is larger than a predetermined first threshold and the calculated dynamic range is larger than a predetermined second threshold, it is determined that a pedestrian exists around the vehicle, and and determining that a road structure exists around the vehicle when the variance is equal to or less than the first threshold value or the dynamic range is equal to or less than the second threshold value. target detection method. In the vehicle target object detection method according to claim 1 ,
The processing unit performs predetermined processing including Fourier transform on the signal received in the receiving step to generate a distance spectrum representing the relationship between the distance from the host vehicle and the signal strength, and the distance spectrum. a first preparation step for determining whether or not there is a peak waveform having a peak within a predetermined intensity range;
When a peak waveform having a peak falling within the predetermined intensity range is confirmed in the first preparation step, the processing unit converts distance spectrum data in a predetermined distance range including the peak position of the peak waveform into an inverse Fourier a second preparation step of converting to time-series polar format data by a predetermined process including conversion,
The vehicle target detection method, wherein in the data generation step, the phase time-series data is generated based on the polar format data generated for the predetermined distance range in the second preparation step. A device for detecting targets around the own vehicle using radar transmitted from the own vehicle,
a transmitter that transmits the radar;
a receiving unit that receives a reflected radar signal from the transmitting unit;
A processing unit that determines the presence or absence of a pedestrian by analyzing the signal received by the receiving unit,
The processing unit is
A data generation process of converting the signal received by the receiving unit into polar format data and generating phase time-series data by extracting phase information in the polar format data over a predetermined period;
From the phase time-series data generated in the data generation process, the variance of the time-series data and the dynamic range that is the difference between the maximum value and the minimum value of the time-series data are calculated, and the calculated variance is When it is larger than a predetermined first threshold and the calculated dynamic range is larger than a predetermined second threshold, it is determined that there are pedestrians around the vehicle, and the variance is the first and determining that a road structure exists around the vehicle when the dynamic range is equal to or less than the second threshold or the dynamic range is equal to or less than the second threshold. Device.

Images