JP2016148515A - Radar device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine the possibility of being a pedestrian or not with high accuracy.SOLUTION: A peak group of a first spectrum included in one peak of a second spectrum when the first spectrum and the second spectrum are stacked one on top of another is made one group. Each of the amount of peak count variation, the amount of peak amplitude variation, and the amount of peak position variation are calculated a prescribed number of times from the most recently acquired first and second spectra and the previously acquired first and second spectra. A flag is turned on when the amounts of peak count variation of a prescribed number of times are greater than or equal to a peak count variation threshold any once. Comparison is made for the amount of peak amplitude variation and the amount of peak position variation as well, with a peak amplitude variation threshold and a peak position variation threshold and a flag is turned on. The target is determined to be a pedestrian when all of three flags are on.SELECTED DRAWING: Figure 5

Description

  The present invention relates to a radar apparatus that can determine whether or not a target is a pedestrian. In the present invention, the pedestrian includes not only a pure pedestrian but also a person on a bicycle or the like.

  A radar device is known as a device for detecting a target. The radar device irradiates a target with radio waves, receives a reflected wave reflected by the target, and detects a distance, speed, and direction to the target. As a method for measuring distance and speed, an FM-CW method, a multi-frequency CW method, a pulse method, and the like are known. Further, as a method for measuring the azimuth, a digital beam forming method, a phase monopulse method, and the like are known. In recent years, radar devices are also mounted on automobiles and the like, which are used for vehicle interval control and collision avoidance.

  Further, as a method for determining whether or not the object detected by the radar device is a pedestrian, there is a method described in Patent Document 1. In Patent Document 1, the distance and azimuth of an object are detected by a radar apparatus, image data is acquired by a camera, and image data in the vicinity of the distance and azimuth of the object detected by the radar apparatus is analyzed. It is determined whether or not is a pedestrian.

  Patent Document 3 describes that the type of an object is determined as follows. Of a plurality of peaks in the frequency spectrum obtained by the radar device, a certain peak is compared with the previously measured peak, the amount of time fluctuation is calculated, and the fluctuation amount is compared with the probability distribution table. It is determined that the type of the object that is most likely to be the variation amount is the actual type of the object. The other peaks are similarly determined. For example, since a vehicle has a small amount of peak time fluctuation and a pedestrian is large, it can be determined whether the vehicle is a pedestrian or not by the above method.

  Also, when measuring the distance with a radar device, the target has a three-dimensional shape with a normal size, so there are multiple reflection points for one target, and as a result, multiple peaks are detected. There is a case. That is, although there is only one target, there is a risk of erroneous detection as if there were a plurality of targets. Therefore, it is necessary to group a plurality of detected peaks for each target (grouping) so that the number of groups matches the number of targets so that the target can be detected correctly.

  As a grouping method, there is a method described in Patent Document 2. Patent Document 2 describes that peak groups having the same or approximately the same distance at a plurality of azimuth angles and forming a maximum value are grouped as one group. Then, it is described that the azimuth angle of the target is detected by image recognition using a camera or the like and the grouping is corrected. For example, when two adjacent targets are detected as one group, division into two groups is performed by correction by image recognition.

  As another grouping method, there is the following method (Patent Document 4). First, the received wave is FFT to generate a first spectrum that is a frequency spectrum, and a part of the frequency band of the received wave is FFT to generate a second spectrum that is a frequency spectrum. By using a part of the frequency band, the resolution is lowered and the reflection point of one target cannot be distinguished, so that one peak of the second spectrum becomes the peak of one target. Then, the peak group of the first spectrum included in a certain peak of the second spectrum is grouped as a peak group from one target.

JP 2009-295184 A JP 2010-151621 A JP 2012-112653 A JP 2014-224794 A

  However, the determination method disclosed in Patent Document 1 is expensive because a camera is required separately from the radar apparatus. Therefore, a method for determining a pedestrian at a lower cost is desired.

  In the pedestrian determination method of Patent Document 3, for example, in a situation where a vehicle is present behind the pedestrian, the reflected wave from the vehicle is blocked by the pedestrian, and the peak value fluctuates greatly. When the pedestrian determination method of Patent Document 3 is executed with the fluctuation amount of the peak value, the vehicle is likely to be determined as a pedestrian. Therefore, when the pedestrian determination method of Patent Document 3 is used for an application such as an automatic driving system or a collision prevention system of a vehicle, it causes a malfunction.

  Therefore, an object of the present invention is to realize a radar apparatus that can determine whether or not each target object is a pedestrian.

  The present invention provides a first spectrum which is a function of a physical quantity used for distance measurement using a received wave in a radar device that transmits a transmitted wave to a target and receives a received wave reflected by the target. The first spectrum calculation means for calculating the second spectrum calculation and the second spectrum calculation for calculating the second spectrum as a function of the physical quantity used for the distance measurement using a part of the frequency band of the received wave used by the first spectrum calculation means Means, a grouping unit for grouping a peak group of the first spectrum at a position included in each peak of the second spectrum as from the same target, and a first spectrum for each group. A peak number fluctuation amount calculating means for calculating a peak number fluctuation amount, which is a fluctuation amount of the peak number of the peak group, and a peak spectrum of the second spectrum for each group. Peak amplitude fluctuation amount calculating means for calculating the peak amplitude fluctuation amount that is the fluctuation amount of the second spectrum, and peak position fluctuation amount calculating means for calculating the peak position fluctuation amount that is the fluctuation amount of the peak position of the second spectrum for each group And a storage unit for storing the peak number fluctuation amount, the peak amplitude fluctuation amount, and the peak position fluctuation amount for a predetermined number of times, and, for a certain group, at least one of the predetermined number of peak number fluctuation amounts is a predetermined value. The peak number fluctuation amount threshold value is exceeded, and at least one of the peak amplitude fluctuation amounts for a predetermined number of times exceeds the predetermined peak amplitude fluctuation amount threshold value, and The target corresponding to the group is a pedestrian when the peak position fluctuation amount exceeds a predetermined value, at least once among the predetermined number of peak position fluctuation amounts. Determining a pedestrian determination unit, a radar device, characterized in that it comprises a.

  The physical quantity used for the distance measurement is a physical quantity such as frequency, phase, time, and distance, and depends on the distance measuring method of the radar apparatus of the present invention. For example, frequency is used in the FM-CW system, and time is used in the pulse system.

  The peak number fluctuation amount, peak amplitude fluctuation amount, and peak position fluctuation amount may be any quantities that represent the peak number, peak amplitude value, and peak position fluctuation, for example, absolute value of difference, standard deviation, standard These are errors, absolute values of time differentiation, etc., and a plurality of these values may be considered. In particular, the absolute value of the difference is desirable because it is simple to calculate.

  The predetermined number of times to be stored in the peak number fluctuation amount, peak amplitude fluctuation amount, and peak position fluctuation amount storage unit is set to an appropriate value according to the radar device configuration, measurement environment, etc., and is changed as needed according to the measurement environment. May be.

  The peak fluctuation amount threshold value, the peak amplitude fluctuation amount threshold value, and the peak position fluctuation amount threshold value are set to appropriate values depending on the configuration of the radar apparatus, the measurement method, the measurement environment, and the like. Moreover, you may change at any time according to a measurement environment.

  The grouping method of the present invention may be used in combination with various conventionally known methods. For example, it is possible to use a method of detecting the azimuth and speed of an object with a radar device and grouping the azimuth or speed. As the azimuth measurement method, a digital beam forming (DBF) method, a phase monopulse method, a method of mechanically scanning an antenna, or the like can be used.

  The part of the frequency band of the received wave at the time of calculating the second spectrum may be not only a part of the received wave itself but also a part of the signal after being converted into a beat signal or a baseband signal. Further, not only the frequency band but also a section (for example, time) of a parameter directly or indirectly related to the frequency is meant.

  In the present invention, the pedestrian includes not only a person who is walking or running or a person who is stopped, but also a person who is riding a bicycle or the like.

  When the radar apparatus of the present invention is an FM-CW system, the first spectrum calculation means is a means for generating a beat signal having a difference frequency between a transmission wave and a reception wave and calculating a frequency spectrum of the beat signal. The second spectrum calculation means generates a beat signal having a difference frequency between the transmission wave and the reception wave, extracts a predetermined continuous section of the time waveform of the beat signal, and calculates a frequency spectrum of the extracted beat signal. It can be a means to do.

  When the radar apparatus of the present invention is a pulse system, the first spectrum calculating means is means for detecting the time waveform of the received wave, and the second spectrum calculating means passes the received wave through a filter of a predetermined frequency band, It can be a means for detecting the time waveform of the transmitted reception wave. Further, a pulse compression method can be used in the present invention.

  According to the present invention, it is possible to group a plurality of detection peaks with high accuracy for each target, and it is possible to accurately determine whether each target is a pedestrian.

1 is a diagram illustrating a configuration of a radar apparatus according to Embodiment 1. FIG. The graph which showed the time change of the frequency of a signal. The figure which showed the hardware constitutions of the computer. A graph showing an example of the first and second spectra. The figure which showed the flowchart of pedestrian determination. The graph which showed the relationship between the integration value of peak amplitude fluctuation amount, and integration rate. The graph which showed the relationship between the integrated value of peak position fluctuation amount, and an integration rate. FIG. 5 is a diagram illustrating a configuration of a radar apparatus according to a second embodiment. The graph which showed the time waveform of the transmission signal. The graph which showed one example of the 1st and 2nd time waveform.

  Hereinafter, specific examples of the present invention will be described with reference to the drawings. However, the present invention is not limited to the examples.

  FIG. 1 is a diagram illustrating a configuration of a radar apparatus 1 according to the first embodiment. The radar apparatus 1 according to the first embodiment is an apparatus that uses an FM-CW method for measuring distance and speed and a phase monopulse method for measuring azimuth. As shown in FIG. 1, the radar apparatus 1 according to the first embodiment includes an oscillator 10, a directional coupler 11, a transmission antenna 12, a reception antenna 13, a mixer 15, a low-pass filter (LPF) 16, and AD conversion. The device 17 and the computer 18 are configured.

  Hereinafter, each structure of the radar apparatus 1 of Example 1 is demonstrated, and the operation | movement is demonstrated.

The oscillator 10 oscillates a frequency-modulated signal (transmission signal) as shown in FIG. Specifically, the transmission signal is a triangular wave having a period T, a frequency displacement width ΔF, and a center frequency f _0, and the frequency increases ΔF linearly with respect to time in the interval ₀ to T / 2. In the interval from 2 to T, the frequency decreases by ΔF linearly with respect to time. The oscillator 10 is connected to the input port of the directional coupler 11.

  The transmission signal may be a sawtooth wave in which the frequency increases or decreases ΔF linearly with respect to time in the interval from 0 to T.

  A transmission signal from the oscillator 10 is input to the directional coupler 11. The directional coupler 11 has an output port connected to the transmission antenna 12 and a coupling port connected to the mixer 15. The transmission signal from the oscillator 10 is branched and output to the transmission antenna 12 and the mixer 15.

  A transmission signal is input to the transmission antenna 12 from the oscillator 10 via the directional coupler 11, and a transmission wave is radiated as an electromagnetic wave from the transmission antenna 12. The receiving antenna 13 receives the received wave reflected by the target. As the transmission antenna 12 and the reception antenna 13, a microstrip antenna or the like can be used.

  The mixer 15 has two input terminals respectively connected to the coupling port of the directional coupler 11 and the output side of the receiving antenna 13, and transmits the transmission signal from the directional coupler 11 and the reception signal from the receiving antenna 13. Mix and output. Then, the output signal is passed through the low-pass filter 16 to extract the beat signal. The beat signal is a signal having a frequency that is the difference between the frequency of the transmission signal and the frequency of the reception signal. The beat signal is sampled as a digital signal by the AD converter 17 and input to the computer 18.

  The computer 18 performs first signal calculation means 18a, second spectrum calculation means 18b, grouping unit 18c, peak number fluctuation amount calculation means 18d, peak amplitude fluctuation amount calculation means 18e, peak position fluctuation amount by digital signal processing by a program. The calculation means 18f and the pedestrian determination means 18g are realized. The storage of the computer 18 is a storage unit 18h that stores data by signal processing. The beat signal, which is a digital signal input from the AD converter 17, is input to the first spectrum calculation unit 18a and the second spectrum calculation unit 18b, respectively, and the frequency spectrum is calculated. And based on the frequency spectrum, a target is detected and it is determined whether a target is a pedestrian. The operation of the computer 18 will be described in detail later.

  FIG. 3 is a diagram showing a specific hardware configuration of the computer 18. As shown in FIG. 3, the computer 18 includes a CPU 100, a memory 101, and a storage 102, and an input / output device 103 is connected to the computer 18. The storage 102 stores an OS and a program. In addition, frequency spectrum data is stored. This storage 102 corresponds to the storage unit 18h. The memory 101 stores temporary information. The CPU 100 executes a program and processes a digital signal input from the AD converter 17. The input / output device 103 is a touch panel display, for example, and displays a detection result of a pedestrian on the display, or changes a program setting by touch input to the display.

  Note that the processing equivalent to the low-pass filter 16 may be realized as digital signal processing in the computer 18 without providing the low-pass filter 16.

  Next, the operation of the radar apparatus 1 according to the first embodiment will be described in the order of distance / speed measurement, grouping, and pedestrian determination.

[Distance speed measurement]
In the radar apparatus 1 according to the first embodiment, the computer 18 calculates a frequency spectrum by performing FFT (Fast Fourier Transform) on the beat signal input to the computer 18, and the distance L of the target from the peak position of the frequency spectrum. Velocity V is measured. Specifically, the frequency spectrum of the beat signal in the frequency increase interval (0 to T / 2 interval in FIG. 2) is calculated, the beat frequency fu is obtained from the peak position, and the frequency decrease interval (T / T in FIG. 2) is calculated. 2 to T), the beat frequency fd is calculated from the peak position, and L = cT (fu + fd) / (8ΔF), where c is the speed of light and the distance L is The speed V can be calculated by V = c (fd−fu) / (4f ₀ ).

  If it is desired to measure the orientation of the target, a plurality of receiving antennas 13 are provided, and a switch is provided between the receiving antenna 13 and the mixer 15 to selectively receive one of the plurality of receiving antennas 13. By receiving the received signals from the respective receiving antennas 13 in a time-sharing manner by switching the switches, the direction of the target can be measured from the difference in the phase of the received signals from the respective receiving antennas 13. it can.

[grouping]
The target is a three-dimensional shape having a normal size. Therefore, if the distance resolution is high, there may be a plurality of reflection points of the target. If ΔF is sufficiently large, that is, if the distance resolution is sufficiently high, there will be a plurality of peaks corresponding to each reflection point in the frequency spectrum of the beat signal. For example, in a vehicle-mounted radar, a vehicle, a pedestrian, or the like is a target. When the target is a pedestrian, a peak of about 1 to 5 exists. In order to correctly detect the target, it is necessary to group these multiple peaks for each target.

  Therefore, the radar apparatus 1 according to the first embodiment performs grouping by performing signal processing in the computer 18 as follows.

  The beat signal input to the computer 18 is processed as follows in each of the first spectrum calculating means 18a and the second spectrum calculating means 18b.

  The first spectrum calculation means 18a performs FFT (Fast Fourier Transform) using the beat signal of T / 2 in the frequency increasing section or the frequency decreasing section to perform the frequency spectrum of the beat signal (hereinafter referred to as the first frequency spectrum). Is calculated.

  Further, the second spectrum calculating means 18b extracts only the portion of the predetermined section T ′ smaller than T / 2 from the beat signal for T / 2 in the frequency increasing section or the frequency decreasing section, and the extracted beat FFT is performed using the signal to calculate a frequency spectrum (hereinafter referred to as a second spectrum). The section T ′ is set to an appropriate value depending on the radar device configuration, measurement environment, target type, and the like.

  For the frequency analysis in the first spectrum calculation means 18a and the second spectrum calculation means 18b, a conventionally known method other than FFT may be used.

  The grouping unit 18c detects and compares the peaks of the first spectrum and the second spectrum. Then, for each peak of the second spectrum, grouping is performed with the peak group of the first spectrum included in the peak as one group.

  Here, the beat signal in the section T ′ used to calculate the second spectrum corresponds to a beat signal when the frequency displacement width of the transmission signal is ΔF · 2T ′ / T (see FIG. 2). ). Since the frequency displacement width is smaller than when the frequency displacement width is ΔF, the frequency resolution is low, and the peak that can be separated in the first spectrum cannot be separated in the second spectrum. That is, in the first spectrum, a plurality of reflection points existed in one target, and thus a plurality of peaks existed. In the second spectrum with low resolution, a plurality of reflection points of the target are distinguished. It becomes impossible to have one peak for one target. From this, when the first spectrum and the second spectrum are overlapped, the peak group of the first spectrum included in one peak of the second spectrum is derived from one target. Can do. Therefore, grouping can be performed by dividing the peak of the first spectrum for each such peak group.

  Whether or not the peak of the first spectrum is included in the peak of a certain second spectrum is determined as follows, for example. First, two local minimum values existing immediately before and after a certain second spectrum peak are extracted. If there is no minimum value, a point that is equal to or less than a predetermined threshold is extracted. The peak of the first spectrum existing between the two minimum values is included in the peak of the first spectrum. In addition, the inclusion of the peak may be determined based on the half width of the peak.

  FIG. 4 is a diagram illustrating an example of the first spectrum and the second spectrum. The grouping method will be specifically described with reference to FIG. In the first spectrum shown in FIG. 4, there are eight peaks a to h in order from the low frequency side. Note that the peak existing in the vicinity of the frequency of 0 is due to the influence of the sneak of the transmission wave, and is ignored here. On the other hand, the second spectrum has two peaks A and B in order from the low frequency side. Among these two peaks, four peaks a to d of the first spectrum exist between the minimum values of the peak A on the low frequency side. Therefore, these four peaks a to d are set as one group. Further, four peaks e to h of the first spectrum exist between the minimum values of the peak B on the high frequency side among the peaks of the second spectrum. These four peaks e to h are set as one group. As described above, the eight peaks of the first spectrum are grouped into two peak groups.

  In the above grouping, it is also possible to use a grouping using a direction detection result. The grouping using the detection result of the azimuth can be performed, for example, by grouping peaks existing within a predetermined azimuth angle range as one group. By the grouping using the azimuth detection result, the grouping result can be corrected, or the group can be subdivided. Further, by performing grouping using the orientation detection result first and then performing the grouping later, the load of the grouping process can be reduced. Of course, only grouping using the direction detection result may be performed, or another grouping method may be used.

  Further, when measuring the speed, it is necessary to associate the peak of the frequency spectrum in the frequency increasing section with the peak of the frequency spectrum in the frequency decreasing section (referred to as pairing). If grouping is performed, a pairing combination may be searched in the same group, and the load of the pairing process can be reduced.

[Pedestrian judgment]
It is determined according to the flowchart of FIG. 5 whether each target corresponding to each group divided by grouping is a pedestrian.

[Step S1]
First, the first and second spectra are repeatedly acquired at predetermined intervals, and the peak number fluctuation amount and peak amplitude fluctuation are obtained for each group from the first and second spectra acquired last time and the first and second spectra acquired this time. The amount and the peak position fluctuation amount are calculated as follows (step S1). These fluctuation amounts may be calculated in parallel, or may be sequentially calculated. The first and second spectrum data acquired last time are stored in the storage unit 18h.

[Step S1-A]
The peak number fluctuation amount is an absolute value of the difference between the peak number of the peak group corresponding to a certain group of the first spectrum acquired this time and the peak number acquired last time. The first spectrum is repeatedly acquired at a predetermined interval, and the data of the first spectrum is stored in the storage unit 18. Then, the peak number fluctuation amount calculating unit 18d calculates the peak number fluctuation amount from the first spectrum data acquired last time and the first spectrum data acquired this time (step S1-A). The calculated peak number fluctuation amount is stored in the storage unit 18h.

[Step S1-B]
The peak amplitude fluctuation amount is the absolute value of the difference between the peak amplitude value of the peak of a certain group of the second spectrum acquired this time (the intensity of the second spectrum at the peak) and the previously acquired peak amplitude value. The second spectrum is repeatedly acquired at a predetermined interval, and the data of the second spectrum is stored in the storage unit 18. Then, the peak amplitude variation calculating means 18e calculates the peak amplitude variation from the previously acquired second spectrum data and the currently acquired second spectrum data (step S1-B). The calculated peak amplitude fluctuation amount is stored in the storage unit 18h.

[Step S1-C]
The peak position fluctuation amount is an absolute value of a difference between a peak position of a peak in a group of the second spectrum acquired this time (a value obtained by converting a peak frequency into a distance) and a previously acquired peak position value. The peak position variation calculation means 18f calculates the peak position variation from the previously acquired second spectrum data and the currently acquired second spectrum data (step S1-C). The calculated peak position fluctuation amount is stored in the storage unit 18h. In addition, although the peak position variation | change_quantity is taken as the value which converted the frequency into distance, you may use it as it is, without converting.

  When the target has a relative speed, the position of each peak in the peak group of the first spectrum and the peak position of the second spectrum are different between the previous time and the current time. Therefore, the movement distance of the target is estimated from the measurement result of the distance and speed of the target by the radar apparatus 1 of the first embodiment, the previous peak and the current peak are associated with each other, the peak number variation amount, The peak amplitude fluctuation amount and the peak position fluctuation amount are calculated.

  The peak number variation amount may be an amount representing the peak variation, and may be not only the absolute value of the peak number difference as in Example 1, but also a standard deviation, a standard error, and the like. The value may be taken into account. In particular, it is most desirable that the peak number fluctuation amount be an absolute value of the difference between the peak numbers because it can be easily calculated. The same applies to the peak amplitude fluctuation amount and the peak position fluctuation amount.

[Step S2]
The peak number fluctuation amount calculating means 18d, the peak amplitude fluctuation amount calculating means 18e, and the peak position fluctuation amount calculating means 18f repeatedly calculate the peak number fluctuation amount, the peak amplitude fluctuation amount, and the peak position fluctuation amount, respectively, for a predetermined number of times ( n times) is stored in the storage unit 18h (step S2). The number n of the predetermined number of times is determined in consideration of the target detection time, the time required for grouping, the calculation time of each variation amount, and the like. If the number of times n to be stored is small, there is a large variation and the estimation accuracy of the pedestrian may be deteriorated. If the number of times n is large, it will take time to estimate the pedestrian.

  Note that the predetermined number of times n to be stored may be appropriately changed depending on the measurement environment or the like. Further, the number of times of storage may be different for each of the peak number fluctuation amount, the peak amplitude fluctuation amount, and the peak position fluctuation amount, but it is simple and desirable that they are the same.

[Steps S3-A, S4-A]
Next, in the pedestrian determination means 18g, it is determined whether or not the stored peak number fluctuation amount for n times is equal to or greater than a peak number fluctuation amount threshold value (step S3-A). If at least one of the n peak number fluctuation amounts is equal to or greater than the peak number fluctuation amount threshold, the peak number flag is turned on and stored in the storage unit 18h (step S4-A). If all of the n peak number fluctuation amounts are less than the peak number fluctuation amount threshold value, the peak number flag remains off.

[Step S3-B, Step S4-B]
Similarly, regarding the peak amplitude fluctuation amount, the pedestrian determination unit 18g determines whether or not the peak amplitude fluctuation amount is equal to or greater than the peak amplitude fluctuation amount threshold value (step S3-B). If at least one of the n peak amplitude fluctuation amounts is equal to or greater than the peak amplitude fluctuation threshold value, the peak amplitude flag is turned on and stored in the storage unit 18h (step S4-B). If all of the n peak amplitude fluctuation amounts are less than the peak amplitude fluctuation amount threshold value, the peak amplitude flag remains off.

[Step S3-C, Step S4-C]
Similarly, the pedestrian determination unit 18g also determines whether or not the peak position fluctuation amount is equal to or greater than the peak position fluctuation amount threshold value (step S3-C). If at least one of the n peak position fluctuation amounts is equal to or greater than the peak position fluctuation amount threshold value, the peak position flag is turned on and stored in the storage unit 18h (step S4-C). If all of the n peak position fluctuation amounts are less than the peak position fluctuation amount threshold value, the peak position flag remains off.

  The peak number variation threshold, peak amplitude variation threshold, or peak position variation threshold may be appropriately varied depending on the measurement environment, application, and the like.

  For example, the peak number variation threshold is set as follows. Percentage of fluctuation in peak number of pedestrians measured by measuring the number of peak fluctuations of a known pedestrian and a known vehicle (other than the vehicle, which is a target to be distinguished from a pedestrian) a predetermined number of times in a few seconds The peak number variation threshold C is set so that A> C / n> B, where A is (how many times the number of peaks have changed during a predetermined number of times) and B is the ratio of the change in the number of vehicle peaks. To do. As an example, if the pedestrian peak number fluctuation ratio A is 0.5, the vehicle peak number fluctuation ratio B is 0.1 and the predetermined number of times is 4, the C / 4 is between 0.1 and 0.5. Is set to 0.3, and the peak number variation threshold C is set to 1.2. If the peak number variation threshold value C is set in this way, the setting is easy, and the difference between the pedestrian and the vehicle can be accurately determined.

  The peak amplitude fluctuation amount threshold value is set as follows, for example. Measure peak amplitude fluctuations for known pedestrians and known vehicles multiple times in a few seconds. Then, the obtained peak amplitude fluctuation amount is integrated in order from a small value. FIG. 6 is a graph showing the relationship between the integration ratio and the peak amplitude fluctuation amount. The vertical axis represents the integration rate. The integration ratio represents the ratio of the number of integrations integrated in ascending order of the peak amplitude fluctuation amount with respect to the total number of integrations. The horizontal axis represents values obtained by arranging the peak amplitude fluctuation amounts in ascending order.

  Since the actual number of peak amplitude fluctuations is n, the value G () between the pedestrian peak amplitude fluctuation amount E and the vehicle peak amplitude fluctuation amount F when the integration ratio is 1 / n. For example, the midpoint of E and F) is selected as the peak amplitude fluctuation amount threshold value. If the peak amplitude variation threshold G is set in this way, the setting is easy, and the difference between a pedestrian and a vehicle can be accurately determined.

  The peak position fluctuation amount threshold value is set as follows, for example. Measure peak position fluctuations of known pedestrians and known vehicles multiple times in a few seconds. Then, the obtained peak position fluctuation amount is integrated in order from a small value. FIG. 7 is a graph showing the relationship between the integration ratio and the peak position fluctuation amount. The horizontal axis represents values obtained by arranging the peak position fluctuation amounts in ascending order. The vertical axis represents the integration rate.

  Since the actual number of peak amplitude fluctuations is n, the value J between the pedestrian peak fluctuation amount H and the vehicle peak amplitude fluctuation amount I when the integration ratio is 1 / n is obtained. Select the peak amplitude fluctuation threshold. If the peak position variation threshold value J is set in this way, the setting is easy, and the difference between the pedestrian and the vehicle can be accurately determined.

  As shown in FIGS. 6 and 7, the variation in the peak amplitude fluctuation amount and the peak position fluctuation amount is small in the vehicle, and the fluctuation amount linearly changes as the integration ratio increases. This is because in the case of a vehicle, the reflection intensity at each reflection point and the change in position of each reflection point are substantially constant. On the other hand, pedestrians have large variations in peak amplitude fluctuation amount and peak position fluctuation amount, and the cumulative amount increases rapidly as the cumulative ratio increases. That is, the integration ratio changes rapidly in the region where the fluctuation amount is small, and the integration ratio changes slowly in the region where the fluctuation amount is large. This is because, in the case of a pedestrian, the reflection intensity at each reflection point and the position of each reflection point vary greatly depending on the pedestrian's body shape, posture, clothing, pedestrian breathing, and the like. Since the peak amplitude fluctuation threshold G and the peak position fluctuation threshold J are determined as shown in FIGS. 6 and 7 based on the difference between the characteristics of the pedestrian and the vehicle, This makes it possible to accurately discriminate between vehicles.

[Steps S5 to S7]
Next, the pedestrian determination means 18g determines whether or not all of the three flags, the peak number flag, peak amplitude flag, and peak position flag, are on (step S5). If all three flags are on, it is determined that the target is a pedestrian (step S6). If even one flag is off, it is determined that the target is not a pedestrian (step S7).

  The above pedestrian determination is performed for each group, and it is determined whether each target corresponding to each group is a pedestrian. In addition, although pedestrian determination may be performed with respect to all groups, you may perform only with respect to a specific group. For example, a group that is clearly not a pedestrian may be omitted without performing pedestrian determination.

  The reason why pedestrian determination can be performed with high accuracy by such a method will be described.

  When the target is a pedestrian, the reflection point of the electromagnetic wave from the radar device 1 changes due to slight movement such as breathing of the pedestrian, and the reflected power becomes weaker or stronger. Therefore, the amplitude of the received signal varies greatly with time, and the number of peaks from pedestrians also varies. Therefore, when the peak number fluctuation amount exceeds a predetermined value, there is a high possibility that the target is a pedestrian.

  The second spectrum is such that one peak corresponds to one target by lowering the resolution. That is, the peak is a combination of received signals from a plurality of reflection points of a target. Even if there is a slight movement due to the pedestrian's breathing, the reflection point changes or the reflection intensity changes, so the phase and amplitude of the signal reflected by the pedestrian varies, and the composite signal of these signals The amplitude also varies frequently. Therefore, when the peak amplitude fluctuation amount exceeds a predetermined value, there is a high possibility that the target is a pedestrian.

  Similarly, the peak position of the second spectrum is likely to fluctuate due to a change in reflection point or a change in reflection intensity due to pedestrian breathing. Therefore, when the peak position fluctuation amount of the second spectrum exceeds a predetermined value, there is a high possibility that the target is a pedestrian.

  Therefore, these three fluctuation amounts (peak number fluctuation amount, peak amplitude fluctuation amount, peak position fluctuation amount) are measured a predetermined number of times, and if any of the three fluctuation amounts exceeds a predetermined threshold value, If it is determined that the user is a pedestrian, the target pedestrian can be determined with high accuracy.

  As described above, according to the radar apparatus 1 of the first embodiment, it is possible to perform grouping with high accuracy and to accurately determine whether or not the target corresponding to each group is a pedestrian.

  FIG. 8 is a diagram illustrating a configuration of the radar apparatus 2 according to the second embodiment. The radar apparatus 2 according to the second embodiment measures a distance by a pulse method. That is, the target distance is measured from the time delay of the pulse. As shown in FIG. 8, the radar apparatus 2 according to the second embodiment is obtained by changing the configuration of the radar apparatus 1 according to the first embodiment as follows.

  In the radar apparatus 2 according to the second embodiment, the oscillator 10 in the radar apparatus 1 according to the first embodiment is replaced with an oscillator 20, and a switch 29 is provided between the directional coupler 11 and the transmission antenna 12. Further, mixers 25A and B, a 90 ° phase shifter 28 are provided, low pass filters (LPF) 26A to D, AD converters 27A to 27D, and a computer 28 are provided.

  The oscillator 20 oscillates a signal (transmission signal) that is a continuous wave of a predetermined frequency. The switch 29 conducts / cuts off the signal from the directional coupler 11 at a predetermined interval. The switch 29 adjusts the signal from the oscillator 20 into a pulse shape having a pulse width τ (see FIG. 9). Then, a pulsed transmission wave is transmitted from the transmission antenna 12.

  The coupling port side of the directional coupler 11 is divided into two parts, one being connected to the input terminal of the mixer 25A and the other being connected to the input terminal of the mixer 25B via the 90 ° phase shifter 28. The received signal from the receiving antenna 13 is divided into two and input to the input terminals of the mixers 25A and 25B, respectively. The mixer 25A mixes and outputs the signal from the directional coupler 11 and the received signal from the receiving antenna 13. The 90 ° phase shifter 28 shifts the signal from the directional coupler 11 by 90 °, and the mixer 25B mixes and outputs the signal from the 90 ° phase shifter 28 and the reception signal from the receiving antenna 13. To do.

  The output side of the mixer 25A is connected to the low-pass filters 26A and B, respectively. The output signal from the mixer 25A is passed through the low-pass filters 26A and B, respectively, and the beat signal is extracted. The output side of the mixer 25B is connected to the low-pass filters 26C and D, respectively. The output signal from the mixer 25B is passed through the low-pass filters 26C and D, respectively, and the beat signal is extracted. The beat signal is a signal having a frequency that is the difference between the frequency of the transmission signal and the frequency of the reception signal. The low-pass filters 26A and C are the same as the low-pass filter 16, whereby a signal in a predetermined band is transmitted and a signal is extracted. In addition, low-pass filters 26B and D having a lower cutoff frequency than the low-pass filters 26A and C are used.

  Note that the cutoff frequencies of the low-pass filters 26B and D are set to appropriate values depending on the configuration of the radar apparatus, the measurement environment, the type of the target, and the like. Further, the cut-off frequency may be varied as appropriate in accordance with changes in the measurement environment.

  Output sides of the low-pass filters 26A to 26D are connected to a computer 28 via AD converters 27A to 27D, respectively. The beat signals from the low-pass filters 26A to 26D are sampled as digital signals by the AD converters 27A to 27D and input to the computer 28.

  As shown in FIG. 8, the computer 28 includes a first spectrum calculation unit 28a, a second spectrum calculation unit 28b, a grouping unit 28c, a peak number variation calculation unit 28d, a peak amplitude variation calculation unit 28e, and a peak position variation calculation. It has means 28f, pedestrian determination means 18, and storage unit 18h. The same reference numerals are the same as those of the computer 18 of the first embodiment. Signals from the low-pass filters 26A and C are input to the first spectrum calculation unit 28a, and signals from the low-pass filters 26B and D are input to the second spectrum calculation unit 28b. About the pedestrian determination means 18g and the memory | storage part 18h, it is the same as that of Example 1. FIG.

  The hardware configuration of the computer 28 is the same as that of the computer 18 of the first embodiment, and there is a difference in the digital signal processing program stored in the storage unit 18h.

  Note that equivalent processing may be realized by digital signal processing in the computer 28 without providing the low-pass filters 26B and 26D. The same applies to the low-pass filters 26A and 26C.

  The radar apparatus 2 according to the second embodiment calculates the distance to the target from the difference in the peak position of the time waveform (first time waveform) of the beat signal obtained by the first spectrum calculating unit 28a, and calculates the beat frequency from the beat frequency of the beat signal. Detect the relative speed of the target. Here, when the relative speed of the target is zero, the beat signal has only a direct current component because the difference frequency is zero. Therefore, two mixers 25A, B and a 90 ° phase shifter 28 are provided to perform quadrature detection, and the relative speed is 0 due to the difference in the amplitude values of the DC components of the two output signals of the low pass filter 26A and the low pass filter 26C. The target can also be detected.

  Also in the radar apparatus 2 according to the second embodiment, if the pulse width τ is small, the target may have a plurality of reflection points. In such a case, the first time detected by the first spectrum calculation unit 28a. The waveform has a plurality of peaks. Therefore, the radar apparatus 2 according to the second embodiment performs grouping as follows.

  First, the first spectrum calculation means 28a detects the time waveform of the signals from the low-pass filters 26A and 26C (hereinafter referred to as the first time waveform). Similarly, the second spectrum calculation means 28b detects the time waveform of the signals from the low pass filters 26B and 26D (hereinafter referred to as the second time waveform). Since the cut-off frequency of the low-pass filters 26B and 26D is lower than that of the low-pass filters 26A and 26C, the pulse width spreads conversely.

  Next, the grouping unit 28c of the computer 28 extracts the peaks of the first time waveform and the second time waveform, respectively. The peak positions of the first time waveform and the second time waveform are compared, and the peaks of the first time waveform included in the peaks are grouped for each peak of the second time waveform. The method for determining whether or not the peak is included is the same as in the first embodiment.

  The grouping will be specifically described with reference to an example of first and second time waveforms shown in FIG. In the example shown in FIG. 10, there are two peaks A and B in the second time waveform. There are seven peaks a to g in the first time waveform. The peak A of the second time waveform includes the peaks a to c of the first time waveform, and the peak B of the second time waveform includes the peaks d to g of the first time waveform. Therefore, among the seven peaks of the first time waveform, the peaks a to c are grouped into one group, and the peaks d to g are grouped into one group.

  The reason why the grouping can be performed accurately by the above method is the same as in the first embodiment. In the distance measurement by the pulse method, the distance resolution is determined by the pulse width, and the distance resolution decreases as the pulse width increases.

  Here, the pulse width of the pulse and the frequency band are in an inversely proportional relationship. Further, the passbands of the low-pass filters 26B and D are lower than the cutoff frequency of the low-pass filters 26A and C. Therefore, the signals output from the low-pass filters 26B and D have a wider pulse width than the signals output from the low-pass filters 26A and C.

  Accordingly, the time waveform of the signals output from the low-pass filters 26B and D has a lower distance resolution (time resolution) than the signals output from the low-pass filters 26A and C. As a result, in the second time waveform, a plurality of reflection points of the target cannot be distinguished, and one peak exists for one target. From this, it can be said that the peak group of the first time waveform included in the peak of the second time waveform is derived from a plurality of reflection points of a certain target, and grouping can be performed by the above method.

  The association between the group obtained from the previously acquired reception signal and the group obtained from the reception signal acquired this time is performed as follows. For a target with a relative speed of 0, the beat signals from the low-pass filters 26A and B and the beat signal from the low-pass filters 26C and D are compared, and those located at the same distance are grouped with the same target. Associate. For a target having a relative speed, the relative speed and the moving distance are estimated, and the group located at the corresponding beat frequency and distance is associated as a group of the same target.

  Further, the pedestrian determination of the target in the radar device 2 of the second embodiment can be performed in the same manner as the radar device 1 of the first embodiment. That is, for a plurality of peaks of the first time waveform grouped for each target by the grouping unit 28c of the computer 28, the peak number variation calculation means 28d and the peak amplitude variation calculation means of the computer 28 according to the flowchart of FIG. It is possible to determine whether or not the target is a pedestrian by performing digital signal processing in 28e, the peak position variation calculation unit 28f, and the pedestrian determination unit 18g.

  When a target having a relative speed of 0 is also detected, in order to obtain a correct amplitude value of the signal, pedestrian determination is performed by combining beat signals of both the low-pass filters 26A and C and the low-pass filters 26B and D. There is a need. On the other hand, when only a target having a relative speed is to be detected, the pedestrian determination may be performed using only one of the beat signals from the low-pass filters 26A and 26B and the beat signals from the low-pass filters 26C and 26D. Good.

  As described above, the radar apparatus 2 according to the second embodiment can also perform grouping with high accuracy and perform pedestrian determination with respect to the target with high accuracy, similarly to the radar apparatus 1 according to the first embodiment.

  As the ranging method, the FM-CW method is used in the first embodiment and the pulse method is used in the second embodiment. However, the present invention can also be applied to other ranging methods. For example, the present invention can be applied to a conventionally known method such as a two-frequency CW method, a multi-frequency CW method, a pulse compression method, and the like, so that a pedestrian can be accurately determined. The first and second spectra in the present invention may be intensity distributions of parameters related to distance, and are based on a distance measuring method. For example, time is used in the pulse system, and frequency is used in the FMCW system.

  In addition, the radar apparatus according to the first and second embodiments measures the distance, speed, and direction of the target. However, the radar apparatus according to the present invention does not necessarily require the measurement, and the pedestrian determination of the target is performed. You may only do.

  The radar apparatus of the present invention can be used as an on-vehicle radar, for example.

DESCRIPTION OF SYMBOLS 1, 2: Radar apparatus 10, 20: Oscillator 11: Directional coupler 12: Transmission antenna 13: Reception antenna 15: Mixer 16, 26A-D: Low pass filter 17, 27A-D: AD converter 18, 28: Computer 18a, 28a: First spectrum calculation means 18b, 28b: Second spectrum calculation means 18c, 28c: Grouping units 18d, 28d: Peak number fluctuation amount calculation means 18e, 28e: Peak amplitude fluctuation amount calculation means 18f, 28f: Peak Position variation calculation means 18g: Pedestrian determination means 18h: Storage unit 28: 90 ° phase shifter 29: Switch 100: CPU
101: Memory 102: Storage 103: Input / output device

Claims (4)

In a radar apparatus that transmits a transmission wave to a target and receives a reception wave reflected by the target,
First spectrum calculation means for calculating a first spectrum that is a function of a physical quantity used for distance measurement using the received wave;
Second spectrum calculating means for calculating a second spectrum that is a function of a physical quantity used for distance measurement using a part of the frequency band of the received wave used in the first spectrum calculating means;
In each peak of the second spectrum, a grouping unit that groups the peak group of the first spectrum at a position included in each peak as from the same target,
For each group, a peak number fluctuation amount calculating means for calculating a peak number fluctuation amount that is a fluctuation amount of the peak number of the peak group of the first spectrum,
For each group, a peak amplitude fluctuation amount calculating means for calculating a peak amplitude fluctuation amount that is a fluctuation amount of the peak amplitude of the second spectrum;
For each group, a peak position fluctuation amount calculating means for calculating a peak position fluctuation amount that is a fluctuation amount of the peak position of the second spectrum;
A storage unit for storing the peak number variation, the peak amplitude variation, and the peak position variation for a predetermined number of times;
For a certain group, at least one of the peak number fluctuation amounts for a predetermined number of times exceeds a peak number fluctuation amount threshold value which is a predetermined value, and the peak amplitude fluctuation amount for the predetermined number of times. Among them, at least once, a peak amplitude fluctuation amount threshold value that is a predetermined value is exceeded, and at least one of the predetermined number of peak position fluctuation amounts is a predetermined value. Pedestrian determination means for determining that the target corresponding to the group is a pedestrian when the variation amount threshold is exceeded;
A radar apparatus comprising: The peak number fluctuation amount is an absolute value of a difference between the peak number acquired last time and the peak number acquired this time,
The peak amplitude fluctuation amount is an absolute value of a difference between the amplitude value of the peak acquired last time and the amplitude value of the peak acquired this time,
The peak position fluctuation amount is an absolute value of a difference between the peak position acquired last time and the peak position acquired this time.
The radar apparatus according to claim 1. The radar apparatus is an FM-CW system,
The first spectrum calculation means is a means for generating a beat signal having a difference frequency between the transmission wave and the reception wave, and calculating a frequency spectrum of the beat signal as a first spectrum,
The second spectrum calculation unit generates a beat signal having a difference frequency between the transmission wave and the reception wave, extracts a predetermined continuous section of the time waveform of the beat signal, and extracts the frequency spectrum of the extracted beat signal. Is a means for calculating as a second spectrum,
The radar apparatus according to claim 1 or 2, wherein The radar device is a pulse system,
The first spectrum calculating means is means for detecting a time waveform of the received wave as a first spectrum,
The second spectrum calculating means is means for passing the received wave through a filter of a predetermined frequency band and detecting a time waveform of the transmitted received wave as a second spectrum.
The radar apparatus according to claim 1 or 2, wherein

Images