Radar signal processing device, radar signal processing method, radar signal processing program, and recording medium

Abstract

The radar signal processing apparatus includes a primary feature quantity extraction unit (21) that extracts, as primary feature quantities, information indicating the distance to an object, information indicating the azimuth of the object, and information indicating the reception intensity value of the reflected wave from the reception information by the reflected wave obtained by reflecting the transmitted wave from the object at each observation time; a data storage processing unit (22) that associates each of the primary feature quantities at a plurality of observation times extracted by the primary feature quantity extraction unit (21) with the object that reflected the transmitted wave and assigns an identification symbol of the object to the primary feature quantity; a secondary feature quantity extraction unit (24) that calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned in the data storage processing unit (22); a membership degree calculation unit (26) that obtains a membership degree score for the secondary feature quantity calculated by the secondary feature quantity extraction unit (24) using the distribution of the membership degree representing a score region in which a membership degree score with respect to the secondary feature quantity is given to each category of a plurality of stationary different objects, and calculates a cumulative score by adding the membership degree scores with respect to the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object; and an object determination unit (27) that performs identification determination of the category of the stationary object that reflected the transmitted wave based on the cumulative score for each category of the object calculated by the membership degree calculation unit (26) and obtains a determination result.

Description

Technical Field

[0001] The present disclosure relates to a radar signal processing apparatus, a radar signal processing method, a radar signal processing program, and a recording medium that are used in a radar apparatus mounted on a moving object such as an automobile or indoor mobility, and that identify the category of an object by using a reception signal based on a reflected wave from the object to determine the category (type) of the object.

Background Art

[0002] A radar signal processing apparatus for an in-vehicle radar apparatus for assisting in determining whether or not to perform braking control of a vehicle by identifying a reflecting object into a predetermined category while the mounted vehicle is traveling is disclosed in Patent Document 1. The radar signal processing apparatus disclosed in Patent Document 1 determines the type (category) of an object while the vehicle is traveling, and includes a primary feature amount extraction unit, a data storage processing unit, a secondary feature amount extraction unit, a membership degree calculation unit, and an object determination unit. The primary feature amount extraction unit extracts information related to the relative distance, relative speed, azimuth, and reflection intensity of the reflecting object as primary feature amounts.

[0003] The data storage processing unit stores the primary feature amounts and associates a plurality of primary feature amounts with the same object in time series over a plurality of cycles. The secondary feature amount extraction unit extracts secondary feature amounts including the reception intensity value for each relative distance and the amount of change in the reception intensity value from the primary feature amounts in the data storage processing unit. The membership degree calculation unit calculates the membership degree with respect to the distribution of the secondary feature amounts related to categories including a predetermined vehicle, pedestrian, and low-position object. The object determination unit determines to which category of object the object belongs, that is, determines the type (category) of the object based on the membership degree for each category of the object input from the membership degree calculation unit, and outputs the determination result.

Prior Art Documents

Patent Documents

[0004] Patent Document 1 WO2016 / 194036 Gazette Summary of the Invention Problems to be Solved by the Invention

[0005] The radar signal processing device shown in Patent Document 1 determines the category of an object from features related to the relative distance between the in-vehicle radar device and the reflecting object, that is, secondary feature quantities including the received intensity value for each distance and the amount of change in the received intensity value. While the vehicle is in motion, in addition to determining the category of an object from features related to the relative distance between the in-vehicle radar device and the reflecting object, in recent years, further, in a state where the in-vehicle radar device and the reflecting object do not move relatively, that is, the relative distance does not change, particularly when a moving object such as a vehicle on which the in-vehicle radar device is mounted stops, a radar signal processing device that can identify a stationary object when starting the moving object is desired.

[0006] The present disclosure has been made in view of the above points, and in a state where the relative distance between itself and the object does not change, particularly when itself stops, it is an object to obtain a radar signal processing device that can identify the category of a stationary object by using a received signal by a reflected wave (received wave) from the object with respect to the stationary object. Means for Solving the Problems

[0007] The radar signal processing apparatus according to the present disclosure includes a primary feature quantity extraction unit that extracts, as primary feature quantities, information indicating the distance to an object, information indicating the azimuth of the object, and information indicating the reception intensity value of the reflected wave from the reception information by the reflected wave obtained by reflecting the transmission wave from the object at each observation time; a data storage processing unit that associates each of the primary feature quantities at a plurality of observation times extracted by the primary feature quantity extraction unit with the object that reflected the transmission wave and assigns an identification symbol of the object to the primary feature quantity; a secondary feature quantity extraction unit that calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned in the data storage processing unit; a membership degree calculation unit that obtains a membership degree score using the distribution of the membership degree representing a score region in which a membership degree score with respect to the secondary feature quantity is given to each category of a plurality of stationary objects for the secondary feature quantity calculated by the secondary feature quantity extraction unit, and calculates a cumulative score by adding the membership degree scores with respect to the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object; and an object determination unit that performs identification determination of the category of the stationary object that reflected the transmission wave based on the cumulative score for each category of the object calculated by the membership degree calculation unit and obtains a determination result.

Effects of the Invention

[0008] According to the present disclosure, it is possible to determine the category of a stationary object in a state where the own device has stopped.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Embodiments for Carrying Out the Invention

[0010] Embodiment 1. The radar signal processing apparatus according to Embodiment 1 will be described with reference to FIGS. 1 to 9. FIG. 1 is a block diagram showing a radar apparatus including the radar signal processing apparatus according to Embodiment 1, mainly showing the radar signal processing apparatus. The radar apparatus including the radar signal processing apparatus according to Embodiment 1 is, for example, an in-vehicle radar apparatus mounted on a moving body such as an automobile or indoor mobility.

[0011] The radar signal processing apparatus according to Embodiment 1 has a feature that it can identify a stationary object when the moving body starts while the moving body is in a stopped state. Note that the radar signal processing apparatus according to Embodiment 1 is designed to function as the radar signal processing apparatus shown in Patent Document 1 when at least one of the moving body or the object moves and the relative distance between the moving body and the object changes.

[0012] In Embodiment 1, the target object is a person, another vehicle, an object that is difficult to overcome such as an obstacle, a low-position object that can be overcome such as a joint or grating on the road surface, and the like. In addition, in order to avoid complexity in the description in Embodiment 1, as an example of a stationary object, a person (hereinafter referred to as a stationary person) and one stationary object are taken as examples for explanation.

[0013] The radar device shown in FIG. 1 includes a transmission / reception device 1 and a radar signal processing device 2 according to Embodiment 1. Note that FIG. 2 shows an example of the relationship between a radar device mounted on a moving body in a state where the moving body has stopped and a stationary object, and shows a stationary person and a stationary object as the stationary objects. In FIG. 2, the distance (relative distance) between the radar device and the stationary person is denoted as D1, the angle indicating the azimuth is denoted as θ1, the distance (relative distance) between the radar device and the stationary object is denoted as D2, and the angle indicating the azimuth is denoted as θ2. The angle indicating the azimuth is the angle with respect to the central axis CA of the irradiation surface of the transmission wave in the radar device.

[0014] The transmission / reception device 1 radiates a transmission wave (radio wave), the transmission wave is reflected by an object, the reflected wave from the object is received, and reception information consisting of digital information is generated according to the received reflected wave (received wave). The reception information generated by the transmission / reception device 1 is information related to the relative distance to the object, the relative speed to the object, the azimuth of the object, and the reflection intensity, that is, the reception intensity. In addition, in Embodiment 1, in a state where the moving body, that is, the moving body itself mounted on the moving body has stopped, the category of the stationary object is determined, that is, the relatively stationary object is identified. Therefore, the relative distance to the stationary object is the distance to the object, and the relative speed to the object is 0.

[0015] The transmission / reception device 1 is a transmission / reception device used in a generally known in-vehicle radar device including a transmission antenna, a transmission signal source, a reception antenna, a reception information generation unit, and a control unit that performs overall control, and a detailed description thereof is omitted. For example, the transmission signal source outputs a transmission signal to the transmission antenna at a set period, for example, a period of 100 msec, and a transmission wave is radiated from the transmission antenna that has received the transmission signal. The receiving antenna receives the reflected wave of the transmitted wave radiated by the transmitting antenna reflected by an object, and the received information generation unit generates received information based on the received reflected wave (received wave).

[0016] The generation of information indicating the relative distance to the object and information indicating the relative velocity with the object, which are the received information generated by the received information generation unit, is performed using a frequency continuous modulation (FMCW: Frequency Modulated Continuous Wave) method, an FCM (Fast-Chirp Modulation) method, a pulse Doppler method, or the like. In addition, the generation of information indicating the orientation of the object is performed using a super-resolution processing technique such as MUSIC (Multiple Signal Classification) or ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques). The generation of information indicating the orientation of the object is not limited to super-resolution processing techniques such as MUSIC or ESPRIT, and processing techniques such as angle FFT (fast Fourier transform) or DBF (digital beam forming) may also be used. The information indicating the reception intensity is represented by the magnitude of the reception power of the received wave.

[0017] That is, the received information generation unit generates, for each observation time, information indicating the distance to the stationary object (hereinafter referred to as distance information), information indicating the velocity of the object (hereinafter referred to as velocity information), information indicating the orientation of the object (hereinafter referred to as orientation information), and information indicating the reflection intensity (reception intensity) (hereinafter referred to as reception intensity information) as received information.

[0018] The radar signal processing apparatus 2 includes a primary feature amount extraction unit 21, a data storage processing unit 22, a feature amount storage unit 23, a secondary feature amount extraction unit 24, an attribution database 25, an attribution calculation unit 26, and an object determination unit 27. The primary feature extraction unit 21 extracts distance information, azimuth information, and received signal strength information as information on primary features from the received information generated by the received information generation unit in the transmission / reception device 1 for each observation time. When there is received information due to reflected waves (received waves) from a plurality of objects, information on primary features regarding the plurality of objects is extracted.

[0019] The extraction of information on primary features is performed based on a period synchronized with a set period for outputting a transmission signal. The time when the primary features are extracted, the time when a transmission signal for a so-called transmission wave is output, or the time when a received wave is received is the observation time. The information on primary features based on the distance information, azimuth information, and received signal strength information is associated with the observation time. The observation time does not necessarily have to be an absolute time such as Coordinated Universal Time (UTC), and it may be a relative time by referring to a clock signal used within the radar device.

[0020] Also, the observation time associated with the primary features may be a time tag indicating the data update time or an order tag indicating the data input order. In short, the observation time only needs to represent the time when the information on primary features is extracted as a relative change in time, including the time indicated by the clock signal, the time tag, or the order tag, etc. In any case, it will be described as the observation time below.

[0021] The data storage processing unit 22 has a storage processing function unit that sequentially stores the information on primary features (hereinafter, simply referred to as primary features for the sake of simplicity) extracted by the primary feature extraction unit 21 and associated with the observation time in the feature storage unit 23 configured by a memory, and an object identification function unit that associates the primary features stored in the feature storage unit 23 associated with the observation time with the objects that reflected the transmission waves in time series within a set time, associates an identification number (ID) for the objects, and stores them in the feature storage unit 23 as primary features with the identification number assigned.

[0022] During the set time, the same primary feature quantity due to the reflected wave (received wave) reflected from the same object is associated with the same identification number by the object identification function unit in the data storage processing unit 22. Note that the observation based on the set time is sequentially repeated every set time until the moving body equipped with the radar device including the radar signal processing device according to the first embodiment starts.

[0023] The association between the primary feature quantity and the identification number by the object identification function unit in the data storage processing unit 22 is performed, for example, by associating the primary feature quantity stored in the feature quantity storage unit 23 at a plurality of observation times within the set time with the reflected wave from the same object that reflected the transmission wave in time series. At this time, the object identification by the object identification function unit in the data storage processing unit 22 identifies the same object by tracking processing using, for example, a Kalman filter together with the primary feature quantities at a plurality of observation times within the set time, and the object identification function unit in the data storage processing unit 22 performs the association between the primary feature quantities regarding the same object.

[0024] Also, as shown in FIG. 2, in the state where the moving body has stopped, since the reflected wave obtained by reflecting the transmission wave from the stationary object has the same observed reflection position, the relative position, that is, the distance information and the azimuth information are the same. Further, although the reflected wave obtained by reflecting the transmission wave from a stationary person has fluctuations in time such as breathing and body movement, by considering the fluctuations with respect to the distance information and the azimuth information, the distance information and the azimuth information can be regarded as the same. Therefore, it is possible to identify whether or not stationary objects are the same by using the distance information and the azimuth information, and the data storage processing unit 22 may perform the association with respect to the primary feature quantity as being due to the same object by using the distance information and the azimuth information.

[0025] For example, the primary feature quantity within the range where the distance information and the azimuth information at the previous observation time are the same or within the range considering the fluctuations is given the same identification number as the primary feature quantity at the previous observation time. For at least one of the distance information and the azimuth information at the previous observation time, particularly when the azimuth information is different, a different identification number is assigned to the primary feature amount that is different from the primary feature amount at the previous observation time. At this time, by using the information that the moving object is stopped, the identification of whether the objects are the same becomes more accurate.

[0026] That is, in the feature amount storage unit 23, distance information, azimuth information, and reception intensity information, which are primary feature amounts associated with the observation time by the data storage processing unit 22 for reflected waves (received waves) reflected from the same object and given the same identification number, are stored. In the feature amount storage unit 23, primary feature amounts having distance information, azimuth information, and reception intensity information associated with the observation time and given an identification number at each of a plurality of observation times within the set time are stored.

[0027] In a state where the moving object has stopped, it can be recognized that the reflected wave is a reflected wave of a transmitted wave from a stationary object or a stationary person because the information indicating the relative speed between the moving object and the object (hereinafter referred to as relative speed information), which is a primary feature amount, is 0. Therefore, in order to identify whether the objects are the same by the data storage processing unit 22 on the premise of relatively stationary objects, it is not necessary to identify whether the objects with relative speed are the same by using the relative speed information, and the calculation amount in the data storage processing unit 22 can be reduced.

[0028] Note that it may be recognized that the received wave is a reflected wave of a transmitted wave from a stationary object or a stationary person in a state where the moving object has stopped based on the information that the moving object has stopped and the information that there is no variation in the distance information, which is a primary feature amount. The fact that there is no variation in the distance information mentioned here means that there is no variation in the temporal distance fluctuation such as breathing and body movement by a stationary person.

[0029] The received wave may generally be superimposed with noise or a sharp change in received intensity due to multipath. In order to obtain a stable determination result in the subsequent stage, a filter for the received intensity based on the set time (set time) or the set observation point (observation time), for example, performing a moving average, and the data storage processing unit 22 may receive the received intensity information with the influence of the received intensity change suppressed as the primary feature amount and store it in the feature amount storage unit 23.

[0030] Also, in the data storage processing unit 22, for those in which the state where the primary feature amount of the same identification number is not updated or the primary feature amount of the same identification number is not added reaches the set number of observations, at the time when the set number of observations is reached, the data storage processing unit 22 may sequentially erase the primary feature amounts of the same identification number stored in the feature amount storage unit 23 in the order of oldest. By erasing in this way, the memory space and memory capacity of the feature amount storage unit 23 can be saved.

[0031] The secondary feature amount extraction unit 24 extracts the primary feature amount assigned with an identification number by the data storage processing unit 22, and calculates the amount of change in the received intensity value with respect to time from the extracted primary feature amounts assigned with the same identification number as the secondary feature amount assigned with the same identification number. That is, the secondary feature amount extraction unit 24 extracts the amount of change in the received intensity value with respect to time as the secondary feature amount assigned with the same identification number by using the received intensity value indicated by the received intensity information extracted as the primary feature amount assigned with the same identification number. The amount of change in the received intensity value is a feature amount defined by the difference or ratio of the received intensity values in the current and immediately previous observations of the received intensity value assigned with the same identification number. In short, the secondary feature amount extraction unit 24 calculates the amount of change in the received intensity value with respect to the reflected wave (received wave) reflected from the object assumed to be the same object as the secondary feature amount.

[0032] By the way, since the reflection characteristics of the transmitted wave with respect to the stationary object do not change, the received intensity value with respect to the stationary object is such that the stationary object reflects the transmitted wave with a constant power, so it shows the same value with respect to time and does not change. On the other hand, even when the person is stationary, the received intensity value for a stationary person varies with time due to minute displacements in the body surface shape and spatial position caused by breathing or body movements for unconscious balance, even when stationary. Note that body movement refers not only to the case of unconscious balance but also to a state where a person moves their body without moving, such as working while standing and movements without movement like squatting, with the same distance to a stationary person.

[0033] For example, the abdomen is displaced by several millimeters due to breathing, and body movement is a greater movement. When the frequency of the transmitted wave used is millimeter wave, the displacement due to breathing or body movement is a movement exceeding one wavelength of the transmitted wave, and temporal fluctuations in the received intensity value are observed due to changes in the simple reflection cross-section due to changes in the body surface or changes in the propagation path related to the position of body parts. Even when considering the average reference of the received intensity value, a change of several dB occurs, so the amount of change in the received intensity value changes moment by moment according to body movement.

[0034] On the other hand, even for received waves in which transmitted waves are reflected from the same object and received, the received intensity value for the object changes according to the distance between the radar device and the object and the directivity gain of the transmitting antenna and receiving antenna in the transmitting and receiving device 1. That is, the electromagnetic wave radiated from the transmitting antenna in the transmitting and receiving device 1, reflected from the object, and received by the receiving antenna has different received intensity values depending on the distance and azimuth to the object, even though it is a received wave from the same object, due to the propagation distance and the directivity gain of the antenna.

[0035] Fig. 3 shows an example of the received intensity value with respect to the relative distance to a stationary object, that is, the distance from the radar device to the stationary object. As shown in Fig. 3, the received intensity value decreases as it gets farther due to spatial attenuation according to the distance. In the case where the object is a stationary person, even if the object is at the same position due to temporal fluctuations such as breathing and body movement, the received signal strength value has a variation amount. However, the received signal strength value with respect to the relative distance to the stationary person exhibits a characteristic in which a variation amount with temporal fluctuations is superimposed along the curve shown as an example in FIG. 3. That is, in FIG. 3, it can be represented as a line having a width corresponding to the variation amount with fluctuations in the received signal strength value with respect to the relative distance.

[0036] Therefore, the variation amount of the received signal strength value extracted as the secondary feature amount by the secondary feature amount extraction unit 24 is calculated as follows. That is, first, for each observation time, the received signal strength value indicated by the received signal strength information, which is the primary feature amount extracted by the primary feature amount extraction unit 21 and stored in the data storage processing unit 22, is corrected using the distance indicated by the distance information extracted as the primary feature amount and the azimuth indicated by the azimuth information, so as to correct it to a value independent of the distance and azimuth to the stationary object.

[0037] The secondary feature amount extraction unit 24, as a second function following the first function of extracting the received signal strength value extracted from the primary feature amount extraction unit 21, normalizes the received signal strength value indicated by the received signal strength information extracted by the primary feature amount extraction unit so that the received signal strength value is independent of the relative distance and azimuth using the distance indicated by the distance information and the azimuth indicated by the azimuth information. In the following description, the primary feature amounts are the received signal strength information, distance information, and azimuth information at the same observation time.

[0038] The correction based on the distance to the stationary object is a correction considering the attenuation amount of the propagation distance based on the distance indicated by the distance information extracted as the primary feature amount. Basically, the attenuation amount of the propagation distance based on the distance uses the attenuation amount LOS of free space loss represented by the following equation (1). LOS = (4πr / λ) Λ 2 ···(1) In the above equation (1), r is the distance indicated by the distance information extracted as the primary feature amount, λ is the wavelength of the transmitted wave, and LOS is the square of (4πr / λ).

[0039] The correction based on the orientation with respect to a stationary object is a correction that takes into account the directivity gains of the transmission antenna and the reception antenna in the transmission and reception apparatus 1 with respect to the orientation indicated by the orientation information extracted as the primary feature amount. FIG. 4 shows an example of the reception intensity value with respect to the relative distance after correction based on the distance and orientation (angle) to a stationary object when the object is a stationary object. When the object is a stationary person, in FIG. 4, it can be represented as a line having a width corresponding to the amount of change in which the reception intensity value fluctuates with respect to the relative distance.

[0040] Therefore, the corrected reception intensity value is calculated by the secondary feature amount extraction unit 24 by, for example, subtracting the propagation attenuation amount based on the propagation distance from the reception intensity value indicated by the reception intensity information extracted as the primary feature amount, and adding the gain difference from the maximum value of the directivity gain of the antenna. As a result, the corrected reception intensity value becomes the same value regardless of the distance to the object and the orientation of the object as long as it is a reception wave reflected from the same object and received, and indicates a reception intensity value normalized by the distance and orientation.

[0041] In this way, in order to use a value normalized by distance and orientation as the reception intensity value, if the same calculation, that is, normalization by distance and orientation, is performed in the membership degree calculation unit 26 and the membership degree database 25 described later, in the identification of the category of the object, for example, it is not necessary to use the average of the reception intensity values or a value corrected by calculating the attenuation amount based on the distance and orientation as a reference value. Hereinafter, the normalized reception intensity value will be described as the corrected reception intensity value.

[0042] The secondary feature amount extraction unit 24 stores in the feature amount storage unit 23 the reception intensity information (hereinafter, simply abbreviated as the corrected reception intensity value) indicating the corrected reception intensity value associated with the observation time to which the identification number is assigned. Since the corrected received signal strength value is information corrected according to the distance indicated by the distance information and the direction indicated by the direction information, it is not necessary to store the distance information and the direction information in the feature quantity storage unit 23, and the memory space and memory capacity in the feature quantity storage unit 23 for storing the received signal strength value can be reduced.

[0043] Next, the secondary feature quantity extraction unit 24 calculates the amount of change with respect to time using the corrected received signal strength value. That is, the secondary feature quantity extraction unit 24, as a third function following the second function, calculates the amount of change in the received signal strength value with respect to time using the corrected received signal strength value having the same identification information, and extracts it as a secondary feature quantity.

[0044] The secondary feature quantity extraction unit 24 sequentially reads from the feature quantity storage unit 23 in time series according to the observation times associated with the corrected received signal strength values at a plurality of observation times to which the same identification number is assigned within the set time, and calculates the difference or ratio between the corrected received signal strength value at the immediately preceding observation time to which the same identification number is assigned and the corrected received signal strength value at the observation time. The calculated difference or ratio is extracted as a secondary feature quantity, and information indicating the secondary feature quantity with an identification number assigned to the extracted secondary feature quantity is stored in the feature quantity storage unit 23.

[0045] The secondary feature quantity calculated by the secondary feature quantity extraction unit 24 is the amount of change in the received signal strength value normalized by the distance and the direction. Since the feature quantity storage unit 23 stores the amount of change in the received signal strength value normalized by the distance and the direction as a secondary feature quantity, the memory space and memory capacity in the feature quantity storage unit 23 can be reduced.

[0046] Note that for each observation time, the extraction of the primary feature quantity by the primary feature quantity extraction unit 21 and the extraction of the secondary feature quantity by the secondary feature quantity extraction unit 24 may be processed as a series of processes, and the primary feature quantity extracted by the primary feature quantity extraction unit 21 may not be stored in the feature quantity storage unit 23, and the corrected received signal strength value and the secondary feature quantity may be stored in the feature quantity storage unit 23. That is, the information stored in the feature storage unit 23 is the corrected reception intensity value and secondary feature value that are linked to the observation time and assigned an identification number. In this way, by not storing the primary feature extracted by the primary feature extraction unit 21 in the feature storage unit 23, the memory space and memory capacity in the feature storage unit 23 can be reduced.

[0047] If the object is stationary, the time-corrected reception intensity value does not change, and although it is affected by sensor and environmental fluctuations and noise, the amount of change in the reception intensity value is almost zero regardless of the observation time, as shown by dashed line A in Figure 5. On the other hand, if the object is a stationary person, there will be temporal fluctuations due to breathing, and the change in the received signal strength will pulsate according to the observation time, as shown by solid line B in FIG.

[0048] In Figure 5, the solid line B shows the periodic change in the reception intensity value due to temporal fluctuations caused by breathing, but temporal fluctuations caused by body movement appear as a non-periodic change in the reception intensity value. In either case, if the object is a stationary person, the secondary feature amount fluctuates within the set time.

[0049] The attribution database 25 stores information representing the distribution of secondary features obtained by observing predetermined categories of stationary objects in advance, and assigning the distribution of secondary features to each category as a numerical score representing the attribution of the secondary features to each category. The distribution of the degrees of belonging may be a distribution of degrees of belonging scores given as scores that numerically represent the degrees of belonging to the secondary feature amounts in each category, the distribution of secondary feature amounts being derived based on the properties of each category.

[0050] In Embodiment 1, the secondary feature amount for creating the membership degree distribution is the amount of change with respect to time calculated by the secondary feature amount extraction unit 24 using the corrected reception intensity value, and is the amount of change in the reception intensity value calculated using the reception intensity value corrected (normalized) by distance and azimuth, similar to the case where it is used as the secondary feature amount. Thus, since the membership degree database 25 stores the distribution of the membership degree created using the amount of change in the reception intensity value corrected by distance and azimuth, the memory space and memory capacity in the membership degree database 25 can be reduced.

[0051] In the following description regarding the membership degree database 25, to avoid complexity, the secondary feature amount, which is the amount of change in the reception intensity value corrected by distance and azimuth, will simply be referred to as the secondary feature amount. The membership degree is given as a membership degree score, which is a numerical value representing the possible value of the amount of change in the reception intensity value, which is the secondary feature amount, for each category of stationary objects. The given membership degree score represents the distribution of the membership degree for each category of objects.

[0052] The distribution of the membership degree is represented as the score area of the category of stationary objects. In each of the categories of a plurality of different stationary objects, the membership degree score is large when the degree of belonging to the category is large, the membership degree score is small when the degree of belonging to the category is small, and the value of the membership degree score is 0 when the degree of belonging to the category is 0.

[0053] For example, if the object is a stationary object, since the reception intensity value corrected in terms of time hardly changes, the membership degree score based on the secondary feature amount becomes large when the amount of change in the reception intensity value is small, as shown by the broken line C in FIG. 6. That is, the distribution of the membership degree based on the secondary feature amount for a stationary object is created as a distribution with a small width of the amount of change centered around almost 0 in the amount of change in the reception intensity value, because there is almost no change in the reception intensity value with respect to time based on the properties of the reception intensity by the received wave from the stationary object.

[0054] On the other hand, if the object is a stationary person, since the reception intensity value corrected in terms of time changes due to breathing and body movement, the membership score based on the secondary feature amount becomes larger when the amount of change in the reception intensity value is large, as indicated by the solid line D in FIG. 6. That is, the distribution of the membership degree based on the secondary feature amount for a stationary person is based on the property of the reception intensity in the received wave from the stationary person, and the change in the reception intensity value with respect to time is due to the temporal fluctuation caused by breathing or the temporal fluctuation caused by breathing and body movement. Based on the data obtained by observing the amount of change in the reception intensity value, which is the secondary feature amount, a distribution with a large width of the amount of change is created centered on a value larger than the amount of change in the reception intensity value for a stationary object.

[0055] The creation of the distribution of the membership degree is performed by performing processing by machine learning using the amount of change in the reception intensity value (sample) obtained by observation for each category as teacher data. For example, the frequency for each amount of change in the reception intensity value is obtained from the teacher data in which the amount of change in the reception intensity value is used as a sample for each category, and the distribution of the membership degree is created. Alternatively, a representative value, for example, an average value, a median value, a standard deviation value, and a distribution shape are obtained from the teacher data in which the amount of change in the reception intensity value is used as a sample for each category, and the distribution of the membership degree is created. When identifying the category of a stationary object, the amount of change in the reception intensity value for the category of the object obtained may be sequentially used as teacher data to update the data indicating the distribution of the membership degree in the membership degree database 25.

[0056] Alternatively, the distribution of the membership degree may be created by obtaining the reception intensity value as a sample for each category, creating a histogram, and then normalizing it. Note that when the amount of change in the reception intensity value cannot be obtained by observation, a distribution of the membership degree with respect to the amount of change in the reception intensity value may be created based on empirical values or values obtained from literature or data.

[0057] In FIG. 6, the horizontal axis represents the secondary feature amount, that is, the change amount of the received signal strength value corrected by the distance and azimuth, and the vertical axis represents the degree of membership, that is, the membership score. In FIG. 6, the degree of membership indicated by the vertical axis shows, as the membership score, the degree to which it is estimated that the change amount of the received signal strength value belongs to the category of any stationary object. In FIG. 6, the broken line C shows, as a distribution curve, the region of the change amount of the received signal strength value assigned to the stationary object and the magnitude of the degree of membership (membership score) with respect to the change amount, and the region surrounded by the distribution curve represents the score region indicating the category of the stationary object.

[0058] Also, in FIG. 6, the solid line D shows, as a distribution curve, the region of the change amount of the received signal strength value assigned to the stationary person and the magnitude of the degree of membership (membership score) with respect to the change amount, and the region surrounded by the distribution curve represents the score region indicating the category of the stationary person. The score region indicating the category of the stationary person is created by giving, as the membership score, the change amount of the received signal strength value that can occur for the stationary person. Therefore, the change amount of the received signal strength value that is not possible for a stationary person, such as the vibration of a heavy machine, does not exist within the score region indicating the category of the stationary person.

[0059] The membership database 25 stores the membership score with respect to the change amount of the received signal strength value indicated by the broken line C for the category of the stationary object, and stores the membership score with respect to the change amount of the received signal strength value indicated by the solid line D for the category of the stationary person. Note that, although it shows the one assigned to one category for the stationary object, a distribution of the degree of membership representing the score regions indicating the categories for each of a plurality of stationary objects such as other vehicles, objects difficult to cross over, and low-position objects that can be crossed over may be created.

[0060] In FIG. 6, there is an overlapping region between the score region indicated by the dashed line C for the category of stationary objects and the score region indicated by the solid line D for the category of a stationary person. Based on the empirical finding that the amount of change in the received signal strength value of stationary objects is small, a threshold is set for the overlapping region. The region where the amount of change in the received signal strength value is smaller than the threshold may be set as the category of stationary objects, and the larger region may be set as the category of a stationary person. Also, in FIG. 6, the region where the amount of change in the received signal strength value is smaller than the threshold may be set as the category of stationary objects, represented by the distribution of the degree of attribution shown by the rectangular score region indicated by the long dashed-dotted line C´, and the larger region may be set as the category of a stationary person, represented by the distribution of the degree of attribution shown by the rectangular score region indicated by the long dashed line D´.

[0061] The degree-of-attribution calculation unit 26 obtains a degree-of-attribution score for each category of stationary objects using the distribution of the degree of attribution stored in the degree-of-attribution database 25 for the secondary feature amounts assigned with identification numbers in the secondary feature extraction unit 24. For example, when the distribution of the degree of attribution for the category of stationary objects shown in FIG. 6 and the distribution of the degree of attribution for the category of a stationary person are stored in the degree-of-attribution database 25, the degree-of-attribution scores for each category for the secondary feature amounts obtained by the degree-of-attribution calculation unit 26 are as follows.

[0062] In the region where the amount of change in the received signal strength value is smaller than the overlapping region between the score region indicated by the dashed line C for the category of stationary objects and the score region indicated by the solid line D for the category of a stationary person in the degree-of-attribution database 25, the degree-of-attribution score for the category of stationary objects is given corresponding to the secondary feature amount calculated in the secondary feature extraction unit 24, which is the degree-of-attribution score for the category of stationary objects in the degree-of-attribution database 25 obtained by the degree-of-attribution calculation unit 26, and the degree-of-attribution score for the category of a stationary person is obtained as 0 from the degree-of-attribution database 25 by the degree-of-attribution calculation unit 26.

[0063] In the overlapping region, the membership score for the category of stationary objects is obtained by the membership calculation unit 26 based on the membership score for the category of stationary objects in the membership database 25 provided corresponding to the secondary feature amount calculated by the secondary feature extraction unit 24, and the membership score for the category of a stationary person is obtained by the membership calculation unit 26 based on the membership score for the category of a stationary person in the membership database 25 provided corresponding to the secondary feature amount calculated by the secondary feature extraction unit 24.

[0064] In a region where the amount of change in the received signal strength value is larger than that in the overlapping region, the membership score for the category of stationary objects is obtained as 0 from the membership database 25 by the membership calculation unit 26, and the membership score for the category of a stationary person is obtained by the membership calculation unit 26 based on the membership score for the category of a stationary person in the membership database 25 provided corresponding to the secondary feature amount calculated by the secondary feature extraction unit 24.

[0065] The membership score obtained by the membership calculation unit 26 is obtained using the secondary feature amount calculated by the secondary feature extraction unit 24 for each observation time, and the obtained membership score is given an identification number associated with the observation time for each category of stationary objects and stored in the memory. When obtaining the membership score for each observation time, by directly receiving the information indicating the secondary feature amount from the secondary feature extraction unit 24, it is not necessary to store the information indicating the secondary feature amount in the feature amount storage unit 23, and the memory space and memory capacity in the feature amount storage unit 23 can be reduced.

[0066] Alternatively, the membership calculation unit 26 may obtain the membership score by reading out the secondary feature amount stored in the feature amount storage unit 23. The membership calculation unit 26 may collectively read out the secondary feature amounts for a plurality of observation times within the set time stored in the feature amount storage unit 23, and the membership calculation unit 26 may obtain the membership score for each category for each of the read-out plurality of secondary feature amounts.

[0067] The membership degree calculation unit 26, as a second function following the first function of obtaining a membership degree score for the secondary feature amounts calculated by the secondary feature amount extraction unit 24 for each category of stationary objects, adds the membership degree scores for the secondary feature amounts to which the same identification number is assigned at a plurality of observation times for each category of stationary objects to calculate a cumulative score.

[0068] Now, assuming a person who is stationary as an object, the secondary feature amounts calculated by the secondary feature amount extraction unit 24 to which the same identification number is assigned for each observation time are large, and due to the distribution of the membership degrees in the category of stationary persons stored in the membership degree database 25, the membership degree scores for each observation time obtained by the membership degree calculation unit 26 in the category of stationary persons are large. As a result, by sequentially adding the membership degree scores for the secondary feature amounts to which the same identification number is assigned in a time series for each observation time, for example, as shown by the broken line F in FIG. 7, the growth rate of the score with respect to the observation time, that is, the slope, is large, and moreover, the cumulative score becomes large.

[0069] On the other hand, assuming a person who is stationary as an object, due to the distribution of the membership degrees in the category of stationary objects stored in the membership degree database 25, the membership degree scores for each observation time obtained by the membership degree calculation unit 26 in the category of stationary objects are small. As a result, even if the membership degree scores for the secondary feature amounts to which the same identification number is assigned are sequentially added in a time series for each observation time, for example, as shown by the solid line E in FIG. 7, the growth rate of the score with respect to the observation time, that is, the slope, is small, and moreover, the cumulative score is small. That is, assuming a person who is stationary as an object, the cumulative score for the category of stationary persons shows a value that is sufficiently large to be distinguishable compared to the cumulative score for the category of stationary objects.

[0070] Also, when assuming a stationary object as the object, the secondary feature amounts calculated by the secondary feature amount extraction unit 24 to which the same identification number is assigned for each observation time are small, and the membership score for each observation time obtained by the membership calculation unit 26 in the category of the stationary object stored in the membership database 25 is large according to the distribution of membership degrees in the category of the stationary object. As a result, by sequentially adding the membership scores for the secondary feature amounts to which the same identification number is assigned in time series for each observation time, the growth rate of the score with respect to the observation time, that is, the slope, is large, and moreover, the cumulative score becomes large.

[0071] On the other hand, when assuming a stationary object as the object, the membership score for each observation time obtained by the membership calculation unit 26 in the category of the stationary person stored in the membership database 25 is small according to the distribution of membership degrees in the category of the stationary person. As a result, even if the membership scores for the secondary feature amounts to which the same identification number is assigned are sequentially added in time series for each observation time, the growth rate of the score with respect to the time, that is, the slope, is small, and moreover, the cumulative score is small. That is, when assuming a stationary object as the object, the cumulative score for the category of the stationary object shows a value large enough to distinguish it from the cumulative score for the category of the stationary person.

[0072] [[ID=⑨]] Based on the cumulative scores for each category to which the same identification number is assigned by the membership calculation unit 26, the object determination unit 27 determines whether the object to which the same identification number is assigned belongs to the category of the stationary object or the category of the stationary person in the first embodiment, or whether it is difficult to make an identification determination. When it is determined that the object belongs to the category of the stationary object, the object that reflected the transmission wave is determined to be a stationary object. When it is determined that the object belongs to the category of the stationary person, the object that reflected the transmission wave is determined to be a stationary person, and the determination result is output. The determination result obtained by the object determination unit 27 is used by a control unit (not shown) for, for example, braking control of a moving body, display of danger prediction, or notification by voice.

[0073] The object discrimination determination by the object determination unit 27 determines the object that reflects the transmission wave as the object indicated by the category of the stationary object with the largest cumulative score among the cumulative scores for each category of a plurality of stationary objects, in Embodiment 1, the category of the stationary object and the category of the stationary person, respectively. Alternatively, the object that reflects the transmission wave is determined as the object indicated by the category of the stationary object with the largest rate of increase (differential processing) of the cumulative score among the cumulative scores for each of the plurality of categories.

[0074] For example, in the example shown in FIG. 7, the object determination unit 27 determines that the object that reflects the transmission wave is a stationary person because the cumulative score for the category of the stationary person shows a value sufficiently larger than the cumulative score for the category of the stationary object. Also, in the example shown in FIG. 7, the object that reflects the transmission wave can also be determined as a stationary person because the rate of increase (differential value) of the score with respect to time in the cumulative score for the category of the stationary person shows a value sufficiently larger than the rate of increase (differential value) of the score with respect to time in the cumulative score for the category of the stationary object.

[0075] Note that as the determination result, instead of specifying one object, it may be a determination result represented by the ratio for the objects in the category where the cumulative score appears. For example, in the example shown in FIG. 7, when the ratio of the cumulative score for the category of the stationary person to the cumulative score for the category of the stationary object is 8 to 2, the object determination unit 27 performs discrimination determination with a determination result indicating numerically how much attribution each category of the stationary object has, such as a probability of 80% for a stationary person and 20% for a stationary object for the object that reflects the transmission wave.

[0076] Also, when the cumulative score for all categories of objects is less than the set threshold value, the object determination unit 27 may output a determination result that the object that reflects the transmission wave is unknown, that is, difficult to discriminate. The radar signal processing device 2 may erroneously estimate the received intensity value indicated by the received intensity information, which is a primary feature amount extracted by the primary feature amount extraction unit 21, and the distance indicated by the distance information immediately after the start of the discrimination determination of the category of the object. An error occurs in the secondary feature amount calculated based on the distance and the received intensity value, and high accuracy cannot be obtained as the secondary feature amount. As a result, due to the fluctuation of the membership score caused by an error between the membership score obtained by the membership score calculation unit 26 and the true membership score, the object determination unit 27 outputs a determination result of the wrong category. This can be prevented by making the discrimination determination difficult when the cumulative score is less than the threshold value.

[0077] The object determination unit 27 may obtain a determination result by using the membership scores at all the observation times within the set time and adding the membership scores for the secondary feature amounts to which the same identification number is assigned, or may update the determination result by using the membership scores at all the observation times within a certain time by dividing the set time at regular intervals and adding the membership scores for the secondary feature amounts to which the same identification number is assigned. The more membership scores are added to obtain the cumulative score for obtaining the determination result, that is, the longer the set time or the certain time, the more the influence of the fluctuation of the membership score is eliminated. Therefore, the reliability can be improved by increasing the set time or the certain time.

[0078] The object determination unit 27 uses the membership scores at all the observation times within a certain time by dividing the set time at regular intervals, obtains a cumulative score by adding the membership scores for the secondary feature amounts to which the same identification number is assigned, and may perform the discrimination determination of the object that reflected the transmission wave based on the cumulative tendency of the obtained multiple cumulative scores. In this way, by performing the discrimination determination of the object based on the cumulative tendency of the cumulative scores obtained at regular intervals, the radar signal processing device has an error in the secondary feature amount immediately after the start of the discrimination determination of the category of the object, and the possibility of misidentification due to the fluctuation of the membership score can be suppressed.

[0079] The object determination unit 27 may obtain a cumulative score by multiplying a weight coefficient by the membership score according to the distance indicated by the distance information, which is the primary feature amount extracted by the primary feature amount extraction unit 21, and adding the membership scores. For example, when the distance indicated by the distance information, which is the primary feature amount, exceeds the set distance, a value obtained by multiplying the membership score by a weight coefficient smaller than 1 is used as the membership score.

[0080] When the distance from a moving body equipped with a radar device including a radar signal processing device to a stationary object is long and the reception intensity value in the reflected wave (received wave) reflected by the transmission wave from the object is small, an error in the membership score due to fluctuations in the reception intensity value indicated by the reception intensity information, which is the primary feature amount, caused by noise and environmental influences is assumed. Therefore, when the distance to the object is long, misidentification of the object category can be reduced by reducing the membership score. Note that although a weight coefficient smaller than 1 is used when the set distance is exceeded based on the set distance, the distance may be divided into a plurality of sections, and the weight coefficient may be made smaller for the farther sections.

[0081] Next, the operation of the radar signal processing device 2 according to Embodiment 1 will be described with reference to FIG. 8. In the following description, the observation time within the set time is t k (k = 0, 1 to K), and it is assumed that the initial observation time is t0 and sequential observations are made (K + 1) times up to t K First, in step ST1, the primary feature amount extraction unit 21 sequentially extracts distance information, azimuth information, and reception intensity information as primary feature amounts in time series for each observation time t k (k = 0, 1 to K) from the transmission / reception device 1. The extracted primary feature amounts are associated with the observation time t k . Step ST1 is a primary feature amount extraction step in which the primary feature amount extraction unit 21 extracts the primary feature amount.

[0082] In step ST2, the data storage processing unit 22 stores the primary feature amounts extracted by the primary feature amount extraction unit 21 at the observation time t k ​The primary feature quantities associated with [ID] are sequentially stored in chronological order in the feature quantity storage unit 23, which is a storage unit. In step ST3, the data storage processing unit 22 assigns an identification number to the primary feature quantities sequentially stored in chronological order in the feature quantity storage unit 23 in association with the object that reflected the transmission wave. In step ST4, the data storage processing unit 22 observes the time t k The primary feature quantities associated with [ID] and assigned with identification numbers are sequentially stored in chronological order in the feature quantity storage unit 23, which is a storage unit.

[0083] Steps ST2 to ST4 are not divided processing steps. Instead, the data storage processing unit 22 stores the primary feature quantities extracted by the primary feature quantity extraction unit 21 at the observation time t k in the feature quantity storage unit 23, stores the primary feature quantities extracted by the primary feature quantity extraction unit 21 at the observation time t k+1 in the feature quantity storage unit 23, and at the same time as storing the primary feature quantities extracted at the observation time t k+1 performs association of the objects that reflected the transmission wave by tracking processing using, for example, a Kalman filter with the primary feature quantities extracted at the observation time t k and stores them in the feature quantity storage unit 23 after assigning the same identification number to the same object.

[0084] Or, since it is premised on identifying an object that is stationary when the moving body is in a stopped state, the distance information and azimuth information in the primary feature quantities can be compared to perform association of the objects that reflected the transmission wave. Therefore, without storing the primary feature quantities not assigned with identification numbers in the feature quantity storage unit 23, the data storage processing unit 22 k compares the distance information and azimuth information in the primary feature quantities extracted at the observation time t k with the primary feature quantities extracted previously at the observation time t to assign an identification number, and k it may be sufficient to store the primary feature quantities associated with [ID] and assigned with identification numbers in the feature quantity storage unit 23, which is a storage unit.

[0085] For example, an identification number 1 is assigned to the primary feature amount extracted at the initial observation time t0 and stored in the feature amount storage unit 23. If it is estimated that the primary feature amount extracted at the observation time t1 is the same as the object that reflected the transmission wave at the initial observation time t0, the identification number 1 is assigned. If it is estimated to be different, the identification number 2 is assigned and stored in the feature amount storage unit 23. Similarly, in chronological order, for the primary feature amount extracted at the observation time t k if it is estimated that the object that reflected the transmission wave is the same as the primary feature amount extracted before the observation time t k from which the primary feature amount is extracted, the same identification number is assigned and stored in the feature amount storage unit 23.

[0086] In short, from step ST2 to step ST4, the data storage processing unit 22 is a step of assigning an identification number to the object that reflected the transmission wave for the primary feature amount extracted at the observation time t k is a step of assigning an identification number to the object that reflected the transmission wave for the primary feature amount extracted at the observation time t Also, from step ST2 to step ST4, the data storage processing unit 22 is a step of storing the primary feature amount associated with the observation time t k and assigned an identification number in the feature amount storage unit 23, which is a storage unit.

[0087] In step ST5, the secondary feature amount extraction unit 24 corrects the received intensity value indicated by the received intensity information in the primary feature amount extracted at the observation time t k by the distance indicated by the distance information and the azimuth indicated by the azimuth information, and obtains a normalized received intensity value that does not depend on the distance and azimuth. Specifically, in step ST5, the secondary feature amount extraction unit 24 subtracts the propagation attenuation amount based on the propagation distance indicated by the distance information extracted as the primary feature amount from the received intensity value indicated by the received intensity information in the primary feature amount extracted at the observation time t k and adds the gain difference from the maximum value of the directivity gain of the antenna that radiates the transmission wave based on the azimuth indicated by the azimuth information extracted as the primary feature amount, to obtain a normalized received intensity value that does not depend on the distance and azimuth.

[0088] Step ST5 is that in step ST1, the primary feature amount extraction unit 21 observes the time t kThis is the step in which the secondary feature extraction unit 24 normalizes the primary feature amounts extracted at [observation time] t, and sequentially stores the normalized primary feature amounts in the feature amount storage unit 23 in time series. In this case, step ST5 and step ST1 are not divided processing steps.

[0089] Alternatively, step ST5 is the step in which the secondary feature extraction unit 24 normalizes the primary feature amounts to which the data storage processing unit 22 has assigned identification numbers in step ST3, and sequentially stores the normalized primary feature amounts in the feature amount storage unit 23 in time series. In this case, step ST5 and step ST3 are not divided processing steps.

[0090] In short, step ST5 is a step in which the secondary feature extraction unit 24 corrects the received signal strength value indicated by the received signal strength information, which is the primary feature amount extracted at [observation time] t, according to the distance indicated by the distance information and the azimuth indicated by the azimuth information in the primary feature amount extracted at [observation time] t, and obtains a normalized received signal strength value that is independent of the distance and azimuth. k In the primary feature amount extracted at [observation time] t, and sets it as a normalized received signal strength value that is independent of the distance and azimuth. k In step ST6, the secondary feature extraction unit 24 calculates the amount of change in the received signal strength value with respect to time using the received signal strength value indicated by the received signal strength information in the primary feature amount extracted at [observation time] t, and the received signal strength value indicated by the received signal strength information in the primary feature amount extracted at the observation time immediately before [observation time] t that has the same identification information, and extracts it as a secondary feature.

[0091] In step ST6, the received signal strength value used is the received signal strength value indicated by the received signal strength information in the primary feature amount extracted at [observation time] t, which the secondary feature extraction unit 24 corrects according to the distance indicated by the distance information and the azimuth indicated by the azimuth information in the primary feature amount extracted at [observation time] t, and is a normalized received signal strength value that is independent of the distance and azimuth. k In the primary feature amount extracted at [observation time] t, and uses it as a normalized received signal strength value that is independent of the distance and azimuth. k In the primary feature amount extracted at the observation time immediately before [observation time] t that has the same identification information, and extracts it as a secondary feature. In step ST6, the received signal strength value used is the received signal strength value indicated by the received signal strength information in the primary feature amount extracted at [observation time] t, which the secondary feature extraction unit 24 corrects according to the distance indicated by the distance information and the azimuth indicated by the azimuth information in the primary feature amount extracted at [observation time] t, and is a normalized received signal strength value that is independent of the distance and azimuth. k In the primary feature amount extracted at [observation time] t, and uses it as a normalized received signal strength value that is independent of the distance and azimuth. k In the primary feature amount extracted at [observation time] t, and is a normalized received signal strength value that is independent of the distance and azimuth.

[0092] Step ST6 is for the observation time t within the set time kIt is not necessary to perform the processing after steps ST2 to ST5 for all (k = 0, 1 to K). That is, for the primary feature amount extracted by the primary feature amount extraction unit 21 at the observation time t k the processing of associating (step ST2), assigning an identification number (step ST3), correcting (normalizing) (step ST5), and extracting a secondary feature amount (change amount of received signal strength value) (step ST6) with respect to the observation time t k may be performed in time series order from the initial observation time t 0、 to the observation time t K up to the observation time t1. Steps ST1 to ST6 are steps for obtaining a secondary feature amount (change amount of received signal strength value) from the primary feature amount.

[0093] Steps ST11 and subsequent steps are steps for identifying and determining the category of a stationary object using the secondary feature amount (change amount of received signal strength value). In step ST11, the membership degree calculation unit 26 obtains, from the feature amount storage unit 23, the secondary feature amount to which an identification number was assigned in the secondary feature amount extraction unit 24 at the initial observation time t0 at the set time, and in step ST12, extracts a membership degree score as the membership degree from each of the distributions of the membership degrees of a plurality of stationary objects stored in the membership degree database 25 corresponding to the change amount (secondary feature amount) of the received signal strength value for the object with the identification number i.

[0094] Step ST12 is a step in which the membership degree calculation unit 26 obtains a membership degree score using the distribution of the membership degrees for each category of the objects stored in the membership degree database 25 for each category of the stationary object with respect to the secondary feature amount calculated by the secondary feature amount extraction unit 24. In step ST12, the secondary feature amount calculated by the secondary feature amount extraction unit 24 is the received signal strength value indicated by the received signal strength information, which is the primary feature amount extracted for each observation time t k by steps ST5 and ST6. kIt is corrected according to the distance indicated by the distance information in the primary feature amount extracted and the direction indicated by the direction information, and is the amount of change in the normalized received intensity value that does not depend on the distance and direction.

[0095] Note that the membership degree calculation unit 26 may directly obtain the secondary feature amount from the secondary feature amount extraction unit 24. In the identification number i, i is a natural number. When the received wave reflected from the transmission wave of the object includes the received waves from a plurality of stationary objects, i is a plurality corresponding to the plurality of stationary objects, and processing is performed for each of the plurality of identification numbers i. The same applies to the following steps.

[0096] In step ST13, the membership degree calculation unit 26 calculates the cumulative sum of the membership degree scores of each category up to the observation time t corresponding to the object with the identification number i, that is, the secondary feature amount to which the identification number i is assigned. k to obtain a cumulative score. The cumulative score at the observation initial time t0 corresponding to the secondary feature amount to which the identification number i is assigned is the membership degree score at the observation initial time t0.

[0097] Step ST13 is for the observation time t after the observation time t1 k at which, at the observation time t k the membership degree score for the secondary feature amount to which the same identification number i was assigned up to the observation time t k-1 before is added to the cumulative score, and then the membership degree score at the observation time t k is added to obtain the cumulative score at the observation time t k Step ST13 is a step in which the membership degree calculation unit 26 calculates the cumulative score by adding the membership degree scores for the secondary feature amounts to which the same identification number is assigned at a plurality of observation times for each category of a plurality of stationary objects.

[0098] ​Also, when the secondary feature amount extraction unit 24 detects, as a secondary feature amount, the fluctuation of an object having a large reception intensity value in the received wave due to the reflection of the transmission wave, such as a heavy machine, it is a change in the reception intensity value of a magnitude that cannot occur in a human. Therefore, in step ST12, since the attribution degree distribution of the human category stored in the attribution degree database 25 is a distribution in which a change in the reception intensity value due to the movement of a stationary human can occur, even if the secondary feature amount extraction unit 24 detects the fluctuation of an object such as a heavy machine as a secondary feature amount, it will not be used as an attribution degree score for the human category, and in step ST13, the cumulative score in the human category will not be increased.

[0099] In step ST14, the object determination unit 27 determines whether it is possible or difficult to perform the identification determination based on the cumulative score for each category to which the same identification number has been assigned by the attribution degree calculation unit 26. If the identification determination is difficult, the process proceeds to step ST15. In step ST15, it is determined whether the observation time T k is the observation time T K or not. If the observation time T k is not the observation time T K , that is, if k is less than K, then k is incremented by 1 by step ST16, that is, the observation time T k is set to the observation time T k+1 , and the process returns to step ST12. Steps ST12 to ST14 are repeated until the identification determination is possible in step ST14 or the observation time T k becomes the observation time T K in step ST15.

[0100] If the identification determination is possible in step ST14, the object determination unit 27 selects the category of the object to which the same identification number has been assigned, and outputs, in step ST17, a determination result that the corresponding category is the category of the object that reflected the transmission wave. If the identification determination is difficult in step ST14 and the observation time T k is the observation time T KIf so, the object determination unit 27 outputs a determination result indicating that the category of the object that reflected the transmission wave is unknown in step ST18.

[0101] The determination result by the object determination unit 27 is as follows: 1) Among the cumulative scores for each category of the object, the object indicated by the category of the object with the largest cumulative score is set as the stationary object that reflected the transmission wave; 2) Among the cumulative scores for each category of the object, the object indicated by the category of the object with the largest growth rate of the score of the cumulative score is set as the stationary object that reflected the transmission wave; 3) It is represented by the ratio to the object indicated by the category of the object according to the cumulative score for each category of the object; 4) When the cumulative scores of all categories of the object are less than the threshold value, it is determined that the category of the object that reflected the transmission wave is unknown, or 5) The set time is divided into a plurality of sections at regular intervals in chronological order, and the identification determination of the category of the object that reflected the transmission wave is performed using the cumulative trend of the plurality of cumulative scores obtained by adding the membership scores at each observation time within the plurality of fixed times. It is performed by any of the following methods.

[0102] Steps ST14, ST15, ST17, and ST18 are steps in which the object determination unit 27 performs the identification determination of the category of the stationary object that reflected the transmission wave based on the cumulative score for each category of the object calculated by the membership degree calculation unit 26 and obtains the determination result.

[0103] The radar signal processing device 2 is realized by a hardware configuration by a computer. As shown in FIG. 9, it includes a CPU (Central Processing Unit) 100, a large-capacity semiconductor memory (RAM: Random Access Memory) 200, a storage device (ROM: Read only memory) 300 such as a hard disk device or an SSD device, an input interface unit 400, an output interface unit 500, and a signal path (bus) 600.

[0104] The CPU 100 controls and manages the RAM 200, the ROM 300, the input interface unit 400, and the output interface unit 500. The CPU 100 loads the program stored in the ROM 300 into the RAM 200, and the CPU 100 executes various processes based on the program loaded into the RAM 200.

[0105] A part of the storage area of the RAM 200 constitutes the feature amount storage unit 23 and the membership degree database 25. The program stored in the ROM 300 and the CPU 100 function as the primary feature amount extraction unit 21, the data storage processing unit 22, the secondary feature amount extraction unit 24, the membership degree calculation unit 26, and the object determination unit 27. The input unit shown in FIG. 9 corresponds to the reception information generation unit in the transmission and reception device 1, and the output unit corresponds to a display or a voice output device that displays or outputs the determination result of the braking mechanism of the moving body or the radar signal processing device 2 by voice.

[0106] The radar signal processing method from step ST1 to step ST6 and from step ST11 to step ST18 is performed by the CPU 100 executing processing according to the program stored in the ROM 300. That is, the program to be executed by the CPU 100 stored in the ROM 300 is the observation time T k For each, a procedure for extracting, as primary feature amounts, information indicating the distance to the object, information indicating the azimuth of the object, and information indicating the reception intensity value of the received wave (reflected wave) reflected from the object by the transmitted wave from the reception information; a procedure for assigning an identification number for the object that reflected the transmitted wave to the extracted primary feature amounts; a procedure for calculating, as a secondary feature amount which is the amount of change in the reception intensity value with respect to time, the amount of change in the reception intensity value from the primary feature amounts to which the same identification number is assigned; for the calculated secondary feature amounts, obtaining a membership degree score using the distribution of membership degrees representing a score area in which a membership degree score for the secondary feature amounts is given for each category of the object, and adding the membership degree scores for the secondary feature amounts to which the same identification number is assigned at a plurality of observation times for each category of the stationary object to calculate a cumulative score; a procedure for performing an identification determination of the category of the stationary object that reflected the transmitted wave based on the calculated cumulative score for each category of the object and obtaining a determination result; is a radar signal processing program for execution.

[0107] The procedure for calculating the secondary feature amount is performed at the observation time T k The reception intensity values indicated by the reception intensity information, which is a primary feature extracted for each observation, are corrected using the distance indicated by the distance information and the direction indicated by the direction information in the primary feature extracted at the same observation time, and the secondary feature is calculated using the amount of change in the normalized reception intensity values that are independent of distance and direction.

[0108] As described above, the radar signal processing device according to the first embodiment includes the secondary feature extraction unit 24 that calculates, as a secondary feature, the amount of change in the reception intensity value over time from primary feature values to which the same identification number is assigned; the attribution calculation unit 26 that calculates, for the secondary feature values calculated by the secondary feature extraction unit 24, the attribution score using the distribution of membership that represents a score region in which the attribution score for the secondary feature values is assigned to each of a plurality of different categories of stationary objects, and calculates a cumulative score by adding up the attribution scores for the secondary feature values to which the same identification number is assigned at a plurality of observation times for each object category; and the object determination unit 27 that performs identification and determination of the category of a stationary object that has reflected the transmission wave using the cumulative score for each object category calculated by the attribution calculation unit 26 and obtains a determination result. Therefore, when the radar signal processing device itself is in a stationary state, it is possible to determine the stationary category using the secondary feature value that is an effective feature for discrimination, which is composed of the amount of change in the reception intensity value indicated by the reception intensity information, which is a primary feature for stationary objects and stationary people.

[0109] Furthermore, in the radar signal processing device according to the first embodiment, the secondary feature calculated by the secondary feature extractor 24 is the amount of change in the reception intensity value normalized by the distance and direction indicated by the distance information and the direction information extracted as the primary feature, relative to the reception intensity value indicated by the reception intensity information extracted as the primary feature. This makes it possible to reduce the memory space and memory capacity of the attribution database 25, which stores information representing the distribution of the attribution degrees, i.e., the distribution of the attribution degrees, given as scores that quantify the attribution degrees for secondary feature values in a predetermined category of a plurality of different stationary objects. Furthermore, the observation time T k When information indicating secondary features linked to the image data and assigned with identification numbers is stored in the feature storage unit 23, the memory space and memory capacity of the feature storage unit 23 can be reduced.

[0110] Another example of the radar signal processing device 2 according to the first embodiment In the radar signal processing device 2 according to the first embodiment described above, the secondary feature calculated by the secondary feature extractor 24 is the amount of change with respect to the reception intensity value normalized by the distance and the direction. In another first embodiment, however, the distance to the object is divided into a plurality of sections, and the secondary feature for each distance in the divided sections is used to identify and determine the category of a stationary object.

[0111] That is, in another embodiment 1, the attribution database 25 stores information representing the distribution of attribution scores, which are scores that quantify the attribution of secondary features for each category of stationary objects, corresponding to each of a plurality of distance sections into which the distance to a stationary object is divided, i.e., the distribution of attribution scores. The attribution calculation unit 26 determines to which of a plurality of distance categories the secondary feature to which an identification number has been assigned in the secondary feature extraction unit 24 belongs, depending on the distance indicated by the distance information, which is a primary feature extracted by the primary feature extraction unit 21, and calculates an attribution score for each category of stationary objects using the distribution of attribution for each category of stationary objects stored in the attribution database 25 corresponding to the determined distance category.

[0112] In an environment where the characteristics of a stationary object reflecting a transmitted wave change, for example, when the distance from a moving body on which an automotive radar device equipped with the radar signal processing device 2 is mounted to a stationary object is short, the tendency of the distribution of the degree of attribution for secondary features may change. By calculating the membership scores according to a plurality of distance sections as in the other embodiments described above, it is possible to reduce the deterioration of the discrimination performance of object categories.

[0113] In addition, in the embodiments, deformation of any component or omission of any component is possible.

Industrial Applicability

[0114] The radar signal processing device according to the present disclosure is suitable for a radar signal processing device of an in-vehicle radar device mounted on a moving body such as an automobile or indoor mobility.

Explanation of Signs

[0115] 1 Transceiver, 2 Radar signal processing device, 21 Primary feature amount extraction unit, 22 Data storage processing unit, 23 Feature amount storage unit, 24 Secondary feature amount extraction unit, 25 Affinity database, 26 Affinity calculation unit, 27 Object determination unit.

Claims

1. A radar signal processing device that determines the category of a stationary object in a state where it itself has stopped, comprising: a primary feature quantity extraction unit that extracts, as primary feature quantities, information indicating the distance to the object, information indicating the azimuth of the object, and information indicating the reception intensity value of the received wave from reception information by the received wave reflected from the object at each observation time; a data storage processing unit that associates each of the primary feature quantities at a plurality of observation times extracted by the primary feature quantity extraction unit with the object that reflected the transmitted wave and assigns an identification symbol of the object to the primary feature quantity; a secondary feature quantity extraction unit that calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned in the data storage processing unit; a membership degree calculation unit that obtains a membership degree score using a distribution of membership degrees representing a score region in which a membership degree score with respect to the secondary feature quantity is given to each category of a plurality of different stationary objects for the secondary feature quantity calculated by the secondary feature quantity extraction unit, and calculates a cumulative score by adding the membership degree scores for the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object; an object determination unit that determines the identification of the category of the stationary object that reflected the transmitted wave based on the cumulative score for each category of the object calculated by the membership degree calculation unit and obtains a determination result; and wherein the secondary feature quantity calculated by the secondary feature quantity extraction unit is the amount of change with respect to the reception intensity value normalized by the distance and azimuth from the distance indicated by the distance information extracted as the primary feature quantity and the azimuth indicated by the azimuth information with respect to the reception intensity value indicated by the reception intensity information extracted as the primary feature quantity, a radar signal processing device.

2. A radar signal processing device that determines the category of a stationary object in a state where it itself has stopped, comprising: a primary feature quantity extraction unit that extracts, as primary feature quantities, information indicating the distance to the object, information indicating the azimuth of the object, and information indicating the reception intensity value of the received wave from reception information by the received wave reflected from the object at each observation time; a data storage processing unit that associates each of the primary feature quantities at a plurality of observation times extracted by the primary feature quantity extraction unit with the object that reflected the transmitted wave and assigns an identification symbol of the object to the primary feature quantity; A secondary feature quantity extraction unit that calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned in the data storage processing unit; For the secondary feature quantity calculated by the secondary feature quantity extraction unit, a membership degree calculation unit that obtains a membership degree score using the distribution of membership degrees representing a score region in which a membership degree score for each category of a plurality of different stationary objects with respect to the secondary feature quantity is given, and adds the membership degree scores for the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object to calculate a cumulative score; An object determination unit that performs identification determination of the category of a stationary object that reflected the transmission wave based on the cumulative score for each category of the object calculated by the membership degree calculation unit and obtains a determination result; comprising; The secondary feature quantity calculated by the secondary feature quantity extraction unit is obtained by subtracting the propagation attenuation amount based on the propagation distance indicated by the distance information extracted as the primary feature quantity from the reception intensity value indicated by the reception intensity information extracted as the primary feature quantity, and adding the gain difference from the maximum value of the directivity gain of the antenna that radiates the transmission wave based on the azimuth indicated by the azimuth information extracted as the primary feature quantity. A radar signal processing device that is the amount of change with respect to the reception intensity value calculated by doing.

3. A radar signal processing device that determines the category of a stationary object in a state where it has stopped itself, A primary feature quantity extraction unit that extracts, as primary feature quantities, information indicating the distance to the object, information indicating the azimuth of the object, and information indicating the reception intensity value of the reception wave from the reception information of the reception wave reflected by the object for each observation time; A data storage processing unit that associates each of the primary feature quantities at a plurality of observation times extracted by the primary feature quantity extraction unit with the object that reflected the transmission wave and assigns an identification symbol of the object to the primary feature quantity; A secondary feature quantity extraction unit that calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned in the data storage processing unit; For the secondary feature quantity calculated by the secondary feature quantity extraction unit, a membership degree calculation unit that obtains a membership degree score using the distribution of membership degrees representing a score region in which a membership degree score for each category of a plurality of different stationary objects with respect to the secondary feature quantity is given, and adds the membership degree scores for the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object to calculate a cumulative score; An object determination unit that performs identification determination of the category of a stationary object that has reflected a transmission wave based on the cumulative score for each category of the object calculated by the attribution degree calculation unit, and obtains a determination result; comprising; The attribution degree score obtained by the attribution degree calculation unit is a radar signal processing device obtained for the secondary feature amount calculated by the secondary feature amount extraction unit corresponding to a plurality of distance sections, using a distribution of a plurality of attribution degrees corresponding to each of the plurality of distance sections obtained by dividing the distance to the stationary object into a plurality of sections.

4. The radar signal processing device according to any one of claims 1 to 3, wherein the attribution degree score for calculating the cumulative score in the secondary feature amount extraction unit is an attribution degree score multiplied by a weight coefficient smaller than 1 when the distance indicated by the distance information extracted as the primary feature amount is farther than the set distance.

5. The radar signal processing device according to any one of claims 1 to 3, wherein the attribution degree score for calculating the cumulative score in the secondary feature amount extraction unit is an attribution degree score multiplied by a weight coefficient that becomes smaller as the distance indicated by the distance information, which is the primary feature amount, is divided into a plurality and the section indicating a farther distance is indicated.

6. The radar signal processing device according to any one of claims 1 to 3, wherein the object determination unit obtains a determination result that the object indicated by the category of the object having the largest cumulative score among the cumulative scores for each category of the object is the stationary object that has reflected the transmission wave.

7. The radar signal processing device according to any one of claims 1 to 3, wherein the object determination unit obtains a determination result that the object indicated by the category of the object having the largest score growth rate among the cumulative scores for each category of the object is the stationary object that has reflected the transmission wave.

8. The radar signal processing device according to any one of claims 1 to 3, wherein the object determination unit obtains a determination result represented by a ratio with respect to the object indicated by the category of the object according to the cumulative score for each category of the object.

9. The radar signal processing device according to any one of claims 1 to 3, wherein when the cumulative scores of all object categories are less than the threshold value, the object determination unit obtains a determination result that the category of the object that has reflected the transmission wave is unknown.

10. The object determination unit divides the set time into a plurality of segments at regular intervals in chronological order, and performs identification determination of the category of the object that reflected the transmission wave based on the cumulative tendency of a plurality of cumulative scores obtained by adding the membership scores of all the observation times within each of the plurality of fixed times. The radar signal processing apparatus according to any one of claims 1 to 3.

11. A step in which the primary feature quantity extraction unit extracts, as primary feature quantities, information indicating the distance to the object, information indicating the azimuth of the object, and information indicating the reception intensity value of the reflected wave from the reception information by the reflected wave that reflects the transmission wave from the object for each observation time; A step in which the data storage processing unit assigns an identification number to the object that reflected the transmission wave to the extracted primary feature quantity; A step in which the secondary feature quantity extraction unit calculates, as a secondary feature quantity, the amount of change in the reception intensity value with respect to time from the primary feature quantities to which the same identification number is assigned; A step in which the membership degree calculation unit obtains a membership degree score using the distribution of membership degrees representing a score region in which a membership degree score for the secondary feature quantity is given for each category of the object for the calculated secondary feature quantity, and adds the membership degree scores for the secondary feature quantities to which the same identification number is assigned at a plurality of observation times for each category of the object to calculate a cumulative score; A step in which the object determination unit performs identification determination of the category of the stationary object that reflected the transmission wave based on the calculated cumulative score for each category of the object, and obtains a determination result; comprising The step of calculating the secondary feature quantity is a radar signal processing method in which the secondary feature quantity extraction unit corrects each reception intensity value indicated by the reception intensity information, which is the primary feature quantity extracted for each observation time, by the distance and azimuth indicated by the distance information and azimuth information in the primary feature quantity extracted at the same observation time, and calculates the secondary feature quantity based on the amount of change in the normalized reception intensity value that does not depend on the distance and azimuth.

12. A procedure for extracting, as primary feature quantities, information indicating the distance to the object, information indicating the orientation of the object, and information indicating the received intensity value of the reflected wave from the received information of the reflected wave that has reflected the transmitted wave from the object at each observation time; a procedure for assigning an identification number to the object that has reflected the transmitted wave to the extracted primary feature quantity; a procedure for calculating, as a secondary feature quantity, the amount of change in the received intensity value with respect to time from the primary feature quantity to which the same identification number has been assigned; for the calculated secondary feature quantity, obtaining a membership score using the distribution of membership degrees representing a score region in which a membership score for the secondary feature quantity is given for each category of the object, and calculating a cumulative score by adding the membership scores for the secondary feature quantities to which the same identification number has been assigned at a plurality of observation times for each category of the object; and a procedure for performing an identification determination of the category of the stationary object that has reflected the transmitted wave based on the calculated cumulative score for each category of the object and obtaining a determination result. A radar signal processing program for causing a computer to execute, The procedure for calculating the secondary feature amount is the observation time T ｋ For each, the received signal strength values indicated by the received signal strength information, which is the primary feature amount extracted for each, are corrected by the distance and azimuth indicated by the distance information and azimuth information in the primary feature amount extracted at the same observation time, and the secondary feature amount is calculated based on the change amount of the normalized received signal strength value that is independent of distance and azimuth. A radar signal processing program

13. A procedure for extracting, as primary feature quantities, information indicating the distance to the object, information indicating the orientation of the object, and information indicating the received intensity value of the reflected wave from the received information of the reflected wave that has reflected the transmitted wave from the object at each observation time; a procedure for assigning an identification number to the object that has reflected the transmitted wave to the extracted primary feature quantity; a procedure for calculating, as a secondary feature quantity, the amount of change in the received intensity value with respect to time from the primary feature quantity to which the same identification number has been assigned; for the calculated secondary feature quantity, obtaining a membership score using the distribution of membership degrees representing a score region in which a membership score for the secondary feature quantity is given for each category of the object, and calculating a cumulative score by adding the membership scores for the secondary feature quantities to which the same identification number has been assigned at a plurality of observation times for each category of the object; and a procedure for performing an identification determination of the category of the stationary object that has reflected the transmitted wave based on the calculated cumulative score for each category of the object and obtaining a determination result. A recording medium storing a radar signal processing program for causing a computer to execute, The procedure for calculating the secondary feature amount is the observation time T ｋ A recording medium that corrects each received signal strength value indicated by the received signal strength information, which is a primary feature amount extracted for each, by the distance and azimuth indicated by the distance information and azimuth information in the primary feature amount extracted at the same observation time, and calculates the secondary feature amount based on the amount of change in the normalized received signal strength value that is independent of the distance and azimuth.

Images