Radar device and method for controlling a radar device

The radar device uses a statistical model and Bayesian filtering to accurately distinguish between stationary vehicles and overhead objects, improving collision avoidance systems by calculating probabilities based on multiple parameters.

DE102019104113B4Active Publication Date: 2024-11-07DENSO TEN LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
DE102019104113
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2018-03-16
Filing Date
2019-02-19
Publication Date
2024-11-07
Estimated Expiration
2039-02-19

AI Technical Summary

Technical Problem

Existing radar systems struggle with low accuracy in distinguishing between stationary objects and overhead objects, often mistakenly recognizing overhead objects as stationary objects.

Method used

A radar device and method that utilizes a statistical model and Bayesian filtering to calculate probabilities of targets being stationary vehicles or overhead objects based on multiple parameters, including side position difference, relative velocity, beam-independent slope, and angular power differences, and employs a Bayesian filtering method to improve accuracy.

Benefits of technology

Enables high-accuracy differentiation between stationary vehicles and overhead objects by using a combination of statistical models and Bayesian filtering, enhancing the reliability of vehicle collision avoidance systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

Radar device (1), comprising: a derivation unit (71, 72, 79) configured to derive detection distances of a target (T1, T2) existing in an area near a vehicle (A) equipped with the radar device (1), based on received signals acquired by transmitting a radar transmission wave to the area near the vehicle (A) and receiving a reflected wave from the target (T1, T2); a detection unit (76) configured to detect a distance at which a maximum value of received power was detected and a distance at which a minimum value of received power was detected, among the detection distances derived sequentially from the derivation unit (71, 72, 79); and a determination unit (76) configured to determine whether the target (T1, T2) is an upper object that will not collide with the vehicle (A), based on the distances detected by the detection unit (76), wherein, In a case where the difference between the distance at which the maximum value of the received powers was detected and the distance at which the minimum value of the received powers was detected is equal to or greater than a third threshold for the difference, the determination unit (76) determines that the target (T1, T2) is the upper object.
Need to check novelty before this filing date? Find Prior Art

Description

BACKGROUND TECHNICAL AREA

[0001] The present invention relates to a radar device and a method for controlling a radar device. RELATED TECHNOLOGY

[0002] According to the prior art, a radar device installed on the front side of the vehicle body of a vehicle or the like outputs a transmission wave to a transmission area outside the vehicle and receives the reflected waves from targets. It derives target data, including location information, etc., of the targets, and distinguishes stationary objects, such as stationary vehicles, etc., located in front of the vehicle from the target data. Then, a vehicle control device installed in the vehicle acquires information about the stationary vehicles, etc., from the radar device and controls the behavior of the vehicle based on the acquired information so that the vehicle avoids collision with, for example, the stationary vehicles, etc., to provide safe and comfortable travel for the vehicle user (see, for example, JP 2016-006383 A).

[0003] DE 698 29 777 T2 discloses a method for classifying objects detected by a radar device mounted on a vehicle. Changes in the amplitude of a received radar reflection from an object are monitored, and based on the amplitude changes, the object is classified as a ground-level object or an overhead object. EP 1 371 997 A1 relates to a method for determining whether a target detected by a radar is a stationary object on the road or not. A fluctuation in the reception level of a wave reflected from a target is determined with respect to the distance of the target. From the fluctuation in the reception level, a difference in the reception level between a maximum point and a minimum point is determined. If the determined difference is greater than a predetermined threshold, it is determined that the target is a stationary object on the road.In addition, the slope is determined based on the distance between the maximum and minimum points. If the determined slope is greater than a predefined threshold, it is determined that the target is a stationary object on the road. Furthermore, the distance between maximum points or between minimum points is determined. If the determined distance is less than a predefined threshold, it is determined that the target is a stationary object on the road. SUMMARY

[0004] However, the prior art technology described above has a problem in that the accuracy in distinguishing between stationary objects and upper objects is insufficient and upper objects are erroneously recognized as still objects.

[0005] An object of an example of embodiments of the present invention is to provide a radar device and a radar device control method for discriminating between stationary objects and overhead objects with high accuracy. This object is achieved by a radar device according to claim 1 and by a radar device control method according to claim 9.

[0006] According to one aspect of the present disclosure, there is provided a radar device including: a deriving unit configured to derive detection distances of a target existing in an area near a vehicle equipped with the radar device based on reception signals detected by transmitting a radar transmission wave to the area near the vehicle and receiving a reflected wave from the target; a detecting unit configured to detect, among the detection distances sequentially derived by the deriving unit, a distance at which a maximum value of reception powers has been detected; and a determining unit configured to determine, based on the distances detected by the detecting unit, whether the target is an upper object that will not collide with the vehicle.

[0007] According to exemplary embodiments of the present disclosure, it may be possible to distinguish, for example, between stationary vehicles and upper objects with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] One or more exemplary embodiments of the present invention will be described in detail based on the following figures, wherein: Fig. 1 is a schematic diagram illustrating an overview of target detection performed by a radar device according to an embodiment; Fig. 2 is a view illustrating the configuration of the radar apparatus according to the first embodiment; Fig. 3 is a view illustrating the relationship between transmitted waves and their reflected waves and the beat signals; Fig. 4A is a view for explaining tip extraction with respect to an up-section; Fig. 4B is a view for explaining peak extraction with respect to an Ab section; Fig. 5 is a conceptual view illustrating angles estimated by an azimuth calculation process as an azimuth spectrum; Fig. 6A is a view for explaining pairing based on azimuth angles and angular powers corresponding to the up-section and the down-section; Fig. 6B is a view for explaining a pairing result; Fig. 7A is a view for explaining a lateral position difference calculation method according to the embodiment; Fig. 7B is a view for explaining the background of the side position difference calculation according to the embodiment; Fig. 8 is a view illustrating the side position difference model according to the embodiment; Fig. 9A is a view for explaining a relative speed difference calculation method according to the embodiment; Fig. 9B is a view for explaining the background of the relative speed difference calculation according to the embodiment; Fig. 10 is a view illustrating the relative speed difference model according to the embodiment; Fig. 11 is a view for explaining the calculation of an up-beam inclination and a down-beam inclination as parameters for calculating a beam-independent inclination difference according to the embodiment; Fig. 12 is a view of a beam-independent tilt difference model according to the embodiment; Fig. 13 is a view for explaining an up / down beam power difference inclination calculation method; Fig. 14 is a view illustrating the uplink / downlink beam power difference tilt model according to the embodiment; Fig. 15A is a view illustrating the relationship between the angular powers and distances of a stationary vehicle according to the embodiment; Fig. 15B is a view illustrating the angular powers and distances of an upper object according to the embodiment; Fig. 16 is a view for explaining an uplink / downlink beam range differential calculation method according to the embodiment; Fig. 17 is a view illustrating the upstream / downstream beam range differential model according to the embodiment; Fig. 18 is a view illustrating an extrapolation cause-based probability table according to the embodiment; Fig. 19 is a view illustrating probability ratio distribution overlap between stationary vehicle distribution and upper objects; Fig. 20A is a view illustrating a weighting coefficient / intersection table A for recursively estimating individual likelihood ratios according to a first detection distance according to the embodiment; Fig. 20B is a view illustrating a weighting coefficient / intersection table B for recursively estimating individual likelihood ratios according to a first detection distance according to the embodiment; Fig. 21A is a view for explaining an attenuation amount range calculation method according to the embodiment; Fig. 21B is a view for explaining the attenuation amount range calculation method according to the embodiment; Fig. 22A is a view for explaining a rise amount range calculation method according to the embodiment; Fig. 22B is a view for explaining the increase amount range calculation method according to the embodiment; Fig. 23A is a view illustrating the relationship between the angular powers and distances of an upper object according to the embodiment; Fig. 23B is a view illustrating the relationship between the angular powers and distances of a stationary vehicle according to the embodiment; Fig. 24A is a view for explaining an upper object discrimination method according to the embodiment, which is performed based on a distance at which the maximum value of angular powers was detected and a distance at which the minimum value was detected; Fig. 24B is a view for explaining the upper object discrimination method according to the embodiment, which is performed based on a distance at which the maximum value of angular powers was detected and a distance at which the minimum value was detected; Fig. 24C is a view for explaining the upper object discrimination method according to the embodiment, which is performed based on a distance at which the maximum value of the angular powers was detected and a distance at which the minimum value was detected; Fig. 25A is a flowchart illustrating a target information output process according to the embodiment; Fig. 25B is a flowchart illustrating a subroutine for removing unnecessary targets according to the embodiment; and Fig. 25C is a flowchart illustrating a subroutine for calculating likelihood ratios 1 to 6 and a (total) likelihood ratio OverAll according to the embodiment. DETAILED DESCRIPTION

[0009] Hereinafter, a radar device and a method for controlling a radar device according to exemplary embodiments of the present invention will be described with reference to the accompanying drawings. However, the following embodiments are merely examples and are not intended to limit the present invention. In addition, the following embodiments mainly indicate configurations and processes according to the disclosed technology, and descriptions of other configurations and processes are omitted. Moreover, the embodiments and modifications can be appropriately combined within a corresponding scope. Furthermore, in the individual embodiments, identical components and identical processes are denoted by the same reference symbols, and descriptions of already explained components and processes are omitted. (Overview of target detection performed by a radar device according to the embodiment)

[0010] In the present embodiment, a radar device distinguishes stationary objects that are detection targets, such as stationary vehicles and upper objects, from a comparatively long distance. The following description will be made using stationary vehicles as examples of stationary objects; however, stationary objects are not limited to stationary vehicles. In other words, in the case of performing discrimination between stationary vehicles and upper objects based on target data items with respect to a single parameter, when the distribution of the target data items to stationary vehicles and the distribution of the target data items to upper objects overlap with each other, it is difficult to distinguish between the stationary vehicles and the upper objects based on a threshold value.

[0011] For this reason, in the present embodiment, a statistical model is constructed from an already known correlation between the value that a stationary vehicle and an upper object may have as a certain parameter and the probability based on the occurrence frequency of that value, and whenever the radar device performs a scanning operation with respect to each target, the probability that it is a stationary vehicle and the probability that it is an upper object are calculated based on the statistical model.Furthermore, whenever a sampling operation is performed, probability updating is performed using a Bayesian filtering method to improve reliability, and whether the target is a stationary vehicle or an upper object is determined based on the ratio (logit) between the probability of being a stationary vehicle and the probability of being an upper object. Since only one parameter is insufficient to perform discrimination, the present invention appropriately uses a plurality of parameters for which statistical models have been defined to complement its capabilities.

[0012] Fig. Figure 1 is a schematic diagram illustrating an overview of target detection performed by a radar device according to the embodiment. A radar device 1 according to the embodiment is mounted on a front part of a vehicle A, for example, on the radiator grille, and detects targets T (targets T1 and T2) that exist in the moving direction of the vehicle A. Targets T include moving targets and stationary targets. Fig. For example, the target T1 shown in Figure 1 is a preceding vehicle moving along the direction of movement of vehicle A, or a stationary object (including a stationary vehicle) that is stationary. In addition, the Fig. 1, for example, an upper object fixed above it in the direction of movement of the vehicle A, such as a traffic light, an overpass, a road sign, an information sign or the like, which is not a vehicle.

[0013] The radar device 1 is a scanning radar which transmits a down-transmission wave TW1 and an up-transmission wave TW2 alternately at intervals of, for example, 5 msec, as shown in Fig. 1, to ensure performance even in the case where the radar is tilted relative to the vertical axis of the vehicle A due to cargo or the vehicle's suspension. The down-transmission wave TW1 is transmitted downward from a down-transmission unit TX1 of the radar device 1 in the moving direction of the vehicle A. The up-transmission wave TW2 is transmitted upward from an up-transmission unit TX2 of the radar device 1 in the moving direction of the vehicle A. The down-transmission unit TX1 and the up-transmission unit TX2 are, for example, antennas.

[0014] As in Fig. As shown in Figure 1, since the scanning ranges of the downlink wave TW1 and the uplink wave TW2 partially overlap in the vertical direction of the vehicle A, the radar device 1 detects targets T in a wider range in the vertical direction compared to the case of using either the downlink wave TW1 or the uplink wave TW2. The radar device 1 receives the reflected waves of the downlink wave TW1 and the uplink wave TW2 from targets T through receiving units RX, thereby detecting the targets T. (CONFIGURATION OF THE RADAR DEVICE ACCORDING TO THE EMBODIMENT)

[0015] Fig. Figure 2 is a view illustrating the configuration of the radar device according to the first embodiment. The radar device 1 according to the embodiment uses, for example, a frequency-modulated continuous wave (FM-CW) among various types of millimeter waves for radars to detect targets T extending within the range of the vehicle A.

[0016] As in Fig. As shown in Figure 2, the radar device 1 is connected to a vehicle control device 2. The vehicle control device 2 is connected to a brake 3, etc. For example, when the traveling direction of the reflected wave of a transmission wave of the radar device 1 from the target T1 to the receiving antenna of the radar device 1 is a predetermined distance or less, so that the vehicle A is in danger of colliding with the target T1, the vehicle control device 2 controls the brake 3, the accelerator, the gear shift, etc., thereby controlling the behavior of the vehicle A so that the vehicle A avoids colliding with the target T1. An example of a system that performs such vehicle control is an adaptive cruise control (ACC) system.

[0017] The moving distance of the reflected wave of a transmission wave of the radar device 1 from the target T1 to the receiving antenna of the radar device 1 is referred to as the "longitudinal distance," and the distance of a target T in the left-right direction (vehicle width direction) of the vehicle A is referred to as the "lateral distance." The left-right direction of the vehicle A is also referred to as the lane width direction of the road on which the vehicle A is traveling. The "lateral distance" is the distance from the position of the center of the vehicle. When the target is on the right side of the vehicle A, the lateral distance is expressed as a positive value, and when the target is on the left side of the vehicle A, the lateral distance is expressed as a negative value. Hereinafter, the lateral distance is also referred to as the lateral position.

[0018] As also in Fig. As shown in Figure 2, the radar device 1 includes a transmitting unit 4, a receiving unit 5 and a signal processing unit 6.

[0019] The transmission unit 4 includes a signal generation unit 41, an oscillator 42, a switch 43, the downlink transmission unit TX1, and the uplink transmission unit TX2. The signal generation unit 41 generates a modulation signal whose voltage varies in a triangular waveform and supplies the modulation signal to the oscillator 42. The oscillator 42 generates a transmission signal whose frequency varies over time by performing frequency modulation on a continuous wave signal based on the modulation signal generated by the signal generation unit 41, and outputs the transmission signal to the downlink transmission unit TX1 and the uplink transmission unit TX2.

[0020] The switch 43 connects one of the downlink transmission unit TX1 and the uplink transmission unit TX2 to the oscillator 42. The switch 43 operates at predetermined times (for example, at intervals of 5 msec) under the control of a transmission control unit 61, which will be described below, to switch the connection between one of the downlink transmission unit TX1 and the uplink transmission unit TX2 and the oscillator 42. In other words, the switch 43 alternately connects the downlink transmission unit TX1 and the uplink transmission unit TX2 to the oscillator 42.

[0021] The downlink transmission unit TX1 and the uplink transmission unit TX2 output the downlink transmission wave TW1 and the uplink transmission wave TW2 from the vehicle A based on the transmission signal. Hereinafter, the downlink transmission unit TX1 and the uplink transmission unit TX2 are also referred to as the transmission units TX. Fig. 2 shows, as an example, the case where each of the number of downlink transmission units TX1 and the number of uplink transmission units TX2 is one; however, the number of downlink transmission units and the number of uplink transmission units can be changed according to the implementation. The transmission units TX are each configured with a plurality of antennas and output the downlink transmission wave TW1 and the uplink transmission wave TW2 via the plurality of antennas in different directions to cover the scanning areas. Hereinafter, the downlink transmission wave TW1 and the uplink transmission wave TW2 are collectively referred to as the transmission waves TW.

[0022] The downlink transmission unit TX1 and the uplink transmission unit TX2 may be connected to the oscillator 42 via the switch 43. Therefore, one of the downlink transmission wave TW1 and the uplink transmission wave TW2 is output from a transmission unit TX of the transmission units TX according to the switching operation of the switch 43. Furthermore, the transmission waves TW to be output are switched alternately according to the switching operation of the switch 43.

[0023] The receiving unit 5 includes individual receiving units 52, each connected to four receiving units RX. The receiving units RX are individual antennas that form a group antenna. Fig. 2 shows, as an example, the case where the number of receiving units RX is four; however, the number of receiving units RX can be changed according to the implementation. Each receiving unit RX receives the reflected waves RX from the targets T. Each individual receiving unit 52 processes the reflected waves RW received via a corresponding receiving unit RX.

[0024] Each of the individual receiving units 52 includes a mixer 53 and an analog-to-digital (A / D) converter 54. Receive signals obtained from reflected waves RW received by the receiving units RX are transmitted to the mixers 53. However, amplifiers may be arranged between the receiving units RX and the mixers 53.

[0025] The transmission signal distributed from the oscillator 42 of the transmission unit 4 is input to the mixers 53, where the transmission signal and the reception signals are mixed. As a result, beat signals representing beat frequencies, which are frequency differences between the frequency of the transmission signal and the frequencies of the reception signals, are generated. The beat signals generated by the mixers 53 are converted in the A / D converters 54 and then output to the signal processing unit 6.

[0026] The signal processing unit 6 is a microcomputer that includes a central processing unit (CPU), a storage unit 63, etc., and controls the entire radar device 1. The signal processing unit 6 stores a variety of data to be used in arithmetic operations, target information acquired by a data processing unit 7, etc. in the storage unit 63. In addition, the storage unit 63 stores lateral position difference models 63a, relative speed difference models 63b, beam-independent inclination difference models 63c, up / down beam power difference inclination models 63d, up / down beam range differential models 63e, an extrapolation-cause-based probability table 63f, a weighting coefficient / intercept table A 63g, and a weighting coefficient / intercept table B 63h, which will be described below.As the storage unit 63, for example, an EPROM (Erasable Programmable Read-Only Memory), a flash memory, and the like can be used; however, the storage unit is not limited to this.

[0027] The signal processing unit includes the transmission control unit 61, a Fourier transform unit 62, and the data processing unit 7 as functions implemented in software in the microcomputer. The transmission control unit 61 controls the signal generation unit 41 of the transmission unit 4 and controls the switching of the switch 43. The data processing unit 7 includes a peak extraction unit 70, an angle estimation unit 71, a pairing unit 72, a continuity determination unit 73, a filtering unit 74, a target classification unit 75, an unnecessary target removal unit 76, a grouping unit 77, and a target information output unit 78.

[0028] The Fourier transform unit 62 performs fast Fourier transformation (FFT) on the beat signal output from each of the plurality of individual receiving units 52. In this way, the Fourier transform unit 62 converts the beat signal with respect to the received signal of each of the plurality of receiving units RX into a frequency spectrum, which is frequency domain data. The frequency spectrum generated by the Fourier transform unit 62 is output to the data processing unit 7.

[0029] The peak extraction unit 70 extracts peaks exceeding a predetermined signal level from the frequency spectrum generated by the Fourier transform unit 62 with respect to an up-section in which the frequency of the transmission signal increases and a down-section in which the frequency decreases.

[0030] Now, with reference to Fig. 3, Fig. 4A and Fig. 4B the process of the tip extraction unit 70 is described. Fig. Figure 3 is a view illustrating the relationship between transmitted waves and reflected waves and the beat signals. Fig. 4A is a view for explaining peak extraction with respect to an up section. Fig. Figure 4B is a view for explaining peak extraction with respect to an Ab section. For better explanation, the Fig. 3 is assumed to be a wave ideally reflected from a target T. In addition, the wave in Fig. 3 is shown by a solid line and the reflected wave RW is shown by a dashed line.

[0031] In the upper part of the Fig. 3, the vertical axis indicates the frequency in GHz and the horizontal axis indicates the time in ms. In addition, the Fig. 3 is assumed to be output in a section between a time t1 and a time t2, and the upstream transmission wave TW2 is assumed to be output in a section between the time t2 and a time t3.

[0032] As in Fig. As shown in Figure 3, each of the downlink transmission wave TW1 and the uplink transmission wave TW2 is a continuous wave whose frequency increases and decreases with a predetermined period that has a predetermined frequency as its center and varies this frequency linearly with time. Herein, the center frequency of the fluctuation width of the downlink transmission wave TW1 and the uplink transmission wave TW2 is indicated by ΔF, the center frequency of the fluctuation width is indicated by f0, and the inverse number of a period in which the frequency increases and decreases is indicated by fm.

[0033] Since the reflected waves R are the reflected waves of the downstream transmission wave TW1 and the upstream transmission wave TW2 from the target T, each of the reflected waves, similar to the downstream transmission wave TW1 and the upstream transmission wave TW2, is a continuous wave whose frequency increases and decreases at a predetermined period with a predetermined frequency as the center. The reflected waves RW have a delay relative to the downstream transmission wave TW1 and the like. A delay time τ depends on the longitudinal distance from the vehicle A to the target T.

[0034] In addition, the frequency of each of the reflected waves RW from the transmitted wave TW is shifted by a frequency fd due to the Doppler effect according to the relative speed of the target T to the vehicle A.

[0035] Thus, the reflected waves RW each have a delay time depending on the length distance and the frequency shifted depending on the relative speed relative to the downstream transmission wave TW1 and the like. Therefore, as shown in the lower part of Fig. 3, in each of the beat signals generated by the mixers 53, the frequency of a part corresponding to the up-section in which the frequency of the transmission signal increases is different from the frequency of a part corresponding to the down-section in which the frequency decreases.

[0036] A beat frequency is the frequency difference between the frequency of the downlink transmission wave TW1 or TW2 and the frequency of a reflected wave RW. In the following, a beat frequency corresponding to an up-section is indicated by fup, and a beat frequency corresponding to a down-section is indicated by fdn. In the lower part of the Fig. 3 The vertical axis indicates the frequency in kHz and the horizontal axis indicates the time in msec.

[0037] In addition, the waveforms in the frequency domains of the beat frequency fup corresponding to the up section and the beat frequency fdn corresponding to the down section when Fourier transformation is performed by the Fourier transformation unit 62 as shown in Fig. 4A and Fig. 4B shown. In Fig. 4A and Fig. 4B, the vertical axis indicates signal power in dB and the horizontal axis indicates frequency in kHz.

[0038] The peak extraction unit 70 extracts peaks Pu and peaks Pd that exceed a predetermined signal power Pref from the Fig. 4A and Fig. 4B. Here, it is assumed that the peak extraction unit 70 extracts peaks Pu and Pd relative to each of the down-transmission wave TW1 and the up-transmission wave TW2 shown in Fig. 3. The predetermined signal power Pref may be fixed or it may be variable. Furthermore, the predetermined signal power Pref may be set to differ depending on the up-section and the down-section.

[0039] In the frequency spectrum of the Fig. 4A, peaks Pu are shown at the positions of the three frequencies fup1, fup2 and fup3. In addition, in the frequency spectrum of the Fig. 4B, peaks Pd are shown at the positions of the three frequencies fdn1, fdn2 and fdn3. In Fig. 4A and Fig. Figure 4B shows, as an example, the case where the number of peaks Pu and Pd are three; however, the number of peaks Pu that can appear is one or more, and the number of peaks Pd that can appear is also one or more. Below, the frequency is also expressed in units of bin. 1 bin corresponds to approximately 467 Hz.

[0040] When only relative speeds are considered, frequencies at which peaks appear in frequency spectra correspond to the length distances of the targets T. 1 bin corresponds to a length distance of about 0.36 m. Furthermore, for example, in the frequency spectrum of an up-section, peak Pu at a certain frequency fup means that there is a target at a length distance corresponding to the frequency fup. For this reason, the peak extraction unit 70 extracts the frequencies at which the peaks Pu and the peaks Pd have powers exceeding the predetermined signal power Pref relative to the up-section and the down-section. Hereinafter, frequencies extracted as described above are referred to as "peak frequencies."

[0041] Frequency spectra of an up-section and a down-section, as in Fig. 4A and Fig. 4B are obtained from a received signal received by one receiving unit RX. Therefore, the Fourier transform unit 62 derives frequency spectra of an up-section and a down-section from each of the received signals received by the four receiving units RX.

[0042] Since the four receiving units RX receive the reflected wave RW from the same target, the same peak frequencies are extracted from the frequency spectra of the four receiving units RX. However, since the positions of the four receiving units RX are different from each other, the phase of a reflected wave RW differs depending on the receiving units RX. For example, phase information items of received signals corresponding to the same bin differ depending on the receiving units RX. Furthermore, in the case where a plurality of targets are located at the same bin but at different angles, information about the plurality of targets is contained in a peak frequency signal in the frequency spectrum.

[0043] The angle estimation unit 71 separates information objects about the plurality of targets existing in the same bin from signals having a peak frequency relative to the up-section and the down-section through an azimuth calculation process, and estimates the angle of each of the plurality of targets. The targets existing in the same bin are targets that have substantially the same longitude separation. The angle estimation unit 71 estimates the angles of the targets based on the phase information objects of the received signals, focusing on the received signals corresponding to the same bin among all the frequency spectra of the four receiving units RX.

[0044] As a method for estimating the angles of targets as described above, widely known angle estimation methods such as ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques), MUSIC (Multiple Signal Classification), and PRISM (Panchromatic Remote-sensing Instrument for Stereo Mapping) can be used. In the above-mentioned manner, the angle estimation unit 71 calculates the plurality of peak angles and the power of signals corresponding to these angles from the signals having a frequency.

[0045] Fig. Figure 5 is a conceptual view illustrating angles estimated by the azimuth calculation process as an azimuth spectrum. In Fig. In Figure 5, the vertical axis indicates signal power in dB, and the horizontal axis indicates angle in degrees. In the angle spectrum, the angles estimated by the azimuth calculation process are shown as peaks Pa that exceed the predetermined signal power Pref. Hereinafter, the angles estimated by the azimuth calculation process are referred to as the "peak angles." Thus, the plurality of peak angles having a peak frequency represent the angles of the plurality of targets existing in the same bin.

[0046] The angle estimation unit 71 performs peak angle derivation as described above with respect to all peak frequencies in the frequency spectra of the up-section and the down-section.

[0047] Through the above-described process, the peak extraction unit 70 and the angle estimation unit 71 derive peak data items corresponding to the plurality of targets existing ahead of the vehicle A with respect to each up-section and down-section. Each peak data item includes parameters such as the above-described peak frequency, peak angle, and signal power corresponding to the peak angle (hereinafter referred to as the "angular power").

[0048] The pairing unit 72 performs pairing to associate the peaks Pu corresponding to the up-section with the peaks Pd corresponding to the down-section based on the degree of agreement between the peak angles and the angular powers calculated with respect to the up-section by the angle estimation unit 71 and the peak angles and the angular powers calculated with respect to the down-section, etc. Fig. Figure 6A is a view for explaining pairing based on azimuth angles and the angular powers corresponding to the up-section and the down-section. Fig. 6B is a view for explaining a pairing result. In Fig. 6A and Fig. 6B, each horizontal axis indicates “Angle (Azimuth)” and each vertical axis indicates “Distance (Longitudinal Distance)”.

[0049] As in Fig. As shown in Figure 6A, when a peak angle and angular power of a peak obtained by performing peak azimuth calculation with respect to the up-portion and a peak angle and angular power of a peak obtained by performing peak azimuth calculation with respect to the down-portion have small differences within predetermined margins, the pairing unit 72 pairs the corresponding peaks. In other words, the pairing unit 72 calculates, for example, Mahalanobis distances using the peak angles and peak powers corresponding to the frequency peaks corresponding to the up-portion and the down-portion. A well-known technology is used to calculate Mahalanobis distances. The pairing unit 72 associates a peak corresponding to the up-portion and a peak corresponding to the down-portion, which have a minimum Mahalanobis distance, with each other.

[0050] As described above, the pairing unit 72 associates peaks relating to the same target T with each other. In this way, the pairing unit 72 derives a target data item for each of the plurality of targets T existing ahead of the vehicle A. Since each target data item is obtained by associating two peaks, it is referred to as a data object pair.

[0051] Then the pairing unit 72 calculates, as in Fig. 6B, the relative speed and distance of each target T relative to the vehicle A from the peak pairs (denoted by “O” in Fig. 6B) of the up-section and the down-section. The pairing unit 72 can derive, for example, parameters (the longitudinal distance, the lateral distance, and the relative speed) of each target data item (each data item pair) using two peak data items of the up-section and the down-section that are the origin of the corresponding target data item. The radar device 1 detects the presence of targets T through pairing.

[0052] The downlink transmission unit TX1 and the uplink transmission unit TX2 alternately irradiate a beam (one scan), and whenever reflected waves RW are received in response to the beam irradiation, the above-described processes of the peak extraction unit 70, the angle estimation unit 71, and the pairing unit 72 are performed to derive the instantaneous values ​​of the parameters (the length distance, the lateral distance, and the relative speed) from each target data item.

[0053] The continuity determination unit 73 determines the temporal continuity between target data objects derived in the past process and target data objects derived in the most recent process. In other words, the continuity determination unit 73 determines whether a target data object derived in the past process and a target data object derived in the most recent process are target data objects of the same target. For example, the past process is the previous target data derivation process, and the most recent process is the current target data derivation process.Specifically, the continuity determination unit 73 estimates the position of the current target data object based on each target data object derived in the previous target data derivation process, and determines one of the target data objects closest to the estimated position among the target data objects derived in the current target data derivation and included in a predetermined range determined based on the estimated position as a target data object closest to the corresponding target data object derived in the past process. At this time, the continuity determination unit 73 determines the target data object closest to the corresponding target data object as a target data object having continuity with the corresponding target data object derived in the past process.

[0054] In addition, in the case where a target data object having continuity with a target data object derived in the past process was not derived in the last process; in other words, in the case where it is determined that a target data object derived in the past process does not have continuity, the continuity determination unit 73 performs the extrapolation process based on the parameters (the length distance, the lateral distance, and the relative speed) of the target data object derived in the past process, virtually deriving a target data object that was not derived in the last process.

[0055] The extrapolation data item derived through the extrapolation process is treated as a target data item derived in the last process. Furthermore, when the extrapolation process is performed a plurality of times consecutively or at a comparatively high frequency relative to a target data item, it is assumed that the corresponding target has been lost, and the corresponding target data item is deleted from a predetermined storage area of ​​the storage unit 63. Specifically, the information about the parameters corresponding to a target number indicating the corresponding target is deleted, and a value indicating that the parameters have been deleted (a value indicating that a deletion flag is in an OFF state) is set in association with this number.Target numbers are indices that identify individual target data objects, and target data objects are each given different numbers.

[0056] The filter unit 74 smooths the parameters (the longitudinal distances, the lateral distances, and the relative velocities) of each pair of target data objects derived in the past process and the most recent process in a time axis direction, thereby deriving a target data object. Target data objects that have undergone such a filtering process are also referred to as "internal filter data objects," in contrast to pair data objects that represent instantaneous values.

[0057] The target classification unit 75 classifies the individual targets based on their relative speeds into preceding vehicles, stationary objects (including stationary vehicles), and approaching vehicles. For example, the target classification unit 75 classifies targets that have relative speeds in the same direction as the speed of vehicle A and orders of magnitude greater than the speed of the vehicle into "preceding vehicles." The target classification unit 75 further classifies, for example, targets that have relative speeds substantially opposite to that of vehicle A into "stationary objects."Furthermore, the target classification unit 75 classifies, for example, targets that have relative speeds opposite to the speed of vehicle A and magnitudes greater than the speed of vehicle A into "approaching vehicles." However, "leading vehicles" may be targets that have relative speeds in the same direction as the speed of vehicle A and magnitudes smaller than the speed of vehicle A. In addition, "approaching vehicles" may be targets that have relative speeds opposite to that of vehicle A and magnitudes smaller than the speed of vehicle A.

[0058] The unnecessary target removal unit 76 determines upper objects, lower objects, rain, rectangular wave ghosts, etc., among the individual targets as unnecessary objects, and excludes the unnecessary objects from the output targets. A process for distinguishing upper objects among the unnecessary objects is described in detail below.

[0059] The grouping unit 77 groups a plurality of target data objects relating to the same target into one target data object. For example, the grouping unit 77 considers target data objects with detection positions and speeds that have small differences within predetermined ranges as target data objects relating to the same target and integrates them into one target data object to be output, thereby reducing the number of target data objects to be output.

[0060] The target information output unit 78 selects output items from a predetermined number (for example, ten) of target data items from the plurality of target data items derived or estimated by extrapolation, and outputs the selected target data items to the vehicle control device 2. Based on the longitudinal distances and the lateral distances of the target data items, the target information output unit 78 selects target data items at targets that exist on the lane of the vehicle A and closer to the vehicle A. Here, in the case where the vehicle A moves substantially along the center of a lane, a moving lane with a width of approximately 1.8 m from each of the two ends of the lane is defined as the lane of the vehicle A. However, the width defining the lane of the vehicle A may be appropriately changed in accordance with the design.

[0061] The target data objects derived in the above-described target data derivation process are stored as parameters, the target numbers indicating the individual target data objects, in the predetermined storage area of ​​the storage unit 63 and used in the subsequent target data derivation processes as target data objects derived in the past process.

[0062] In other words, the target data items derived in the past target data derivation processes are stored as "histories." Referring to a "peak frequency" stored, for example, as a "history" in the predetermined storage area of ​​the storage unit 63, the peak extraction unit 70 estimates a "peak frequency" with temporal continuity with this "history," and extracts a frequency from a range based on the estimated "peak frequency," for example, from a range between "[estimated peak frequency] - 3 (bins)" and "[estimated peak frequency] + 3 (bins)." In this way, the radar device 1 can quickly select "peak frequencies" corresponding to targets that need to be preferentially output to the vehicle control device 2. The peak frequency of a target data item estimated in the current process is referred to as the "estimation bin." (FIRST DISTINGUISHING PROCESS OF DISTINGUISHING BETWEEN STATIONARY VEHICLES AND UPPER OBJECTS ACCORDING TO THE EMBODIMENT)

[0063] Hereinafter, details of the process of discriminating between stationary vehicles and upper objects performed by the unnecessary target removing unit 76 according to the embodiment will be described with reference to Fig. 7A to Fig. 18 in the order from STEP 1 to STEP 3. < STEP 1: Calculating the probability ratios>

[0064] From the parameters based on each of the target data items acquired by each scanning operation and probability distribution models corresponding to the parameters, the unnecessary target removal unit 76 calculates six probability ratios (logarithmic probability ratios) with respect to the corresponding target based on the following formula 1. For each of the detection distances set at intervals of, for example, 10 m, probability distribution models to be used for calculating the six probability ratios are defined or constructed in advance based on the actual measurement data. Regarding the distances in the detection distance intervals, probability distribution models obtained by linear interpolation are used.

[0065] In addition, the probability distribution models used to calculate the probability ratios include the side position difference models 63a, the relative speed difference models 63b, the beam-independent inclination difference models 63c, the up / down beam power difference inclination models 63d, the up / down beam range differential models 63e, and the extrapolation cause-based probability table 63f, as described above with reference to Fig. 2. Details of the side position difference models 63a are described below with reference to Fig. 8. Details of the relative velocity difference models 63b are described below with reference to Fig. 10. Details of the beam-independent tilt difference models 63c are described below with reference to Fig. 12. Details of the upstream / downstream beam power difference tilt models 63d are described below with reference to Fig. 14. Details of the upstream / downstream beam range differential models 63e are described below with reference to Fig. 17. Details of the extrapolation-cause-based probability table 63f are described below with reference to Fig. 18 described. [Probability ratio n] = log([Probability n of being a stationary vehicle] / [Probability n of being an upper object]) = log(Probability n of being a stationary vehicle)−log(Probability n of being an upper object) (where n=1, 2, 3,…6)

[0066] A likelihood ratio 1 obtained when "n" in Formula 1 is "1" is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "lateral position difference" to be described below. A likelihood ratio 2 obtained when "n" in Formula 1 is "2" is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "relative speed difference" to be described below.A likelihood ratio of 3, obtained when "n" in Formula 1 is "3," is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "beam-independent inclination difference" to be described below. A likelihood ratio of 4, obtained when "n" in Formula 1 is "4," is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "up / down beam power difference," which is to be described below.A likelihood ratio of 5, obtained when "n" in Formula 1 is "5," is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "upward / downward beam range differential" to be described below. A likelihood ratio of 6, obtained when "n" in Formula 1 is "6," is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, which is derived based on a parameter "extrapolation cause," which is described below. • STEP 1-1: Logarithmic likelihood ratio calculation based on the “side position difference”

[0067] The unnecessary target removal unit 76 calculates the side position difference from the current scanning position detected by the current scanning operation and the previous side position detected by the previous scanning operation based on the following formula 2. [Page position difference] = [current page position] − [previous page position]

[0068] Fig. 7A is a view for explaining the lateral position difference calculation method according to the embodiment. In Fig. 7A, the horizontal axis indicates distance (detection distance) and the vertical axis indicates the page position. As in Fig. For example, as shown in Fig. 7A, the lateral position difference in the case where the lateral positions of a target were detected at intervals of 50 msec is calculated by subtracting the previous lateral position (which is, for example, (2) in Fig. 7A) from the current page position (which is, for example, (1) in Fig. 7A).

[0069] Fig. 7B is a view for explaining the background of the side position difference calculation according to the embodiment. As shown in Fig. As shown in Figure 7B, the lateral position difference calculation is based on the fact that (a) in the case of vehicles, the positions of the reflection points are stable; while (b) in the case of upper objects, since the positions of the reflection points are unstable because their side widths and side positions may shift, lateral position differences tend to be large.

[0070] Next, the unit distance of unnecessary targets 76 is calculated from an example of the side position difference models 63a shown in Fig. 8 and the side position difference calculated based on Formula 2, the “likelihood ratio 1” based on Formula 1. Fig. Figure 8 is a view illustrating the lateral position difference model according to the embodiment. The lateral position difference model 63a is a probability distribution model that has a horizontal axis indicating a lateral position difference in meters and a vertical axis indicating the probability, and represents the relationship between the lateral position difference and the probability of being a stationary vehicle and the relationship between the lateral position difference and the probability of being an upper object. The probability distribution model in Fig. 8, which represents the probability of being a stationary vehicle and the probability of being an upper object, is a model constructed in advance based on, for example, normal distributions using a maximum likelihood estimation method and an experimental design method. In the lateral position difference model 63a, the parameter characterizing the model can be adjusted for stationary vehicles and upper objects, respectively, to improve the accuracy of the determination.

[0071] In addition, Fig. 8, as an example of the lateral position difference model 63a, the lateral position difference model applied to the case where the distance from the vehicle A to a target is 80 m, which is one of the distances set at 10 m intervals from 10 m to about 150 m, is shown. Lateral position difference models applied to the case where the distance from the vehicle A to the target is other than the distances are not shown.

[0072] For example, it is assumed that the “side position difference” calculated based on formula 2 is “1”. In this case, Fig. 8, when the lateral distance is "1" on the horizontal axis, the probability of being a stationary vehicle and the probability of being an upper object are "0.13" and "0.27" on the vertical axis, respectively. Therefore, by substituting "0.13" and "0.27" for the terms "probability 1 of being a stationary vehicle" and "probability 1 of being an upper object" in Formula 1, "[Probability Ratios 1] = log(Probability 1 of being a stationary vehicle) - log(Probability 1 of being an upper object)" (Formula 1 when "n" is 1), the probability ratio 1 can be calculated. • STEP 1-2: Logarithmic probability ratio calculation based on the “relative speed difference”

[0073] From the differential of the distance difference between the current distance detected by the current scanning operation and the previous distance detected by the previous scanning operation and the current relative speed, the unnecessary target removal unit 76 calculates the relative speed difference based on the following formula 3. [Relative Speed ​​Difference] = −1 × [Current Distance] − [Previous Distance] [Update Interval] + [Current Relative Speed]

[0074] Fig. 9A is a view for explaining the relative speed difference calculation method according to the embodiment. As shown in Fig. 9A, for example, in the case where the distances of a target were detected at intervals of 50 msec, the differential of the distance difference between the current distance and the previous distance is calculated by subtracting the previous distance (which is, for example, (2) in Fig. 9A) from the current distance (which, for example, is (1) in Fig. 9A) and dividing the obtained difference (= [current distance] - [previous distance]) by the update interval (the ΔT (for example, 50 msec) in Fig. 9A). Then, the differential of the distance difference between the current distance and the previous distance is converted to the opposite sign and then converted to the current relative speed (which, for example, corresponds to (3) in Fig. 9A). This calculates the relative velocity difference.

[0075] Fig. 9B is a view for explaining the background of the relative speed difference calculation according to the embodiment. As shown in Fig. As shown in Figure 9B, in the case of upper objects, since they have longitudinal and lateral widths, the positions of the reflection points are unstable and can shift in the longitudinal and lateral directions. Therefore, the distances from the radar device 1 tend to change significantly, resulting in an increase in the value of the differential of the detection distance relative to time. The relative speed difference calculation described above is based on this fact.

[0076] Next, calculate the unit distance of unnecessary targets 76 from an example of the relative speed difference models 63b shown in Fig. 10 and the relative speed difference calculated based on Formula 3, the “likelihood ratio 2” based on Formula 1. Fig. Figure 10 is a view illustrating the relative speed difference model according to the embodiment. The relative speed difference model 63b is a probability distribution model that has a horizontal axis indicating a relative speed difference in m / s and a vertical axis indicating the probability, and represents the relationship between the relative speed difference and the probability of being a stationary vehicle and the relationship between the relative speed difference and the probability of being an upper object. The probability distribution model in Fig. 10, which represents the probability of being a stationary vehicle and the probability of being an upper object, is a model constructed in advance based on, for example, normal distributions using a maximum likelihood estimation method and an experimental design method. In the relative speed difference model 63b, the parameter characterizing the model can be adjusted for stationary vehicles and upper objects, respectively, to improve the accuracy of the determination.

[0077] In addition, Fig. 10, as an example of the relative speed difference models 63b, the relative speed difference model applied to the case where the distance from vehicle A to a target is 80 m, which is one of the distances set at 10 m intervals from 10 m to about 150 m, is shown. Relative speed difference models applied to the case where the distance from vehicle A to the target is other than the distances are not shown.

[0078] For example, it is assumed that the “relative speed difference” calculated based on formula 3 is “2”. In this case, Fig. 10, when the relative speed difference is "2" on the horizontal axis, the probability of being a stationary vehicle and the probability of being an upper object are "0.11" and "0.75" on the vertical axis, respectively. Therefore, by substituting "0.11" and "0.75" for the terms "probability 2 of being a stationary vehicle" and "probability 2 of being an upper object" in Formula 1, "[Probability ratio 2] = log(Probability 2 of being a stationary vehicle) - log(Probability 2 of being an upper object)" (Formula 1 when "n" is 2), the probability ratio 2 can be calculated. • STEP 1-3: Logarithmic likelihood ratio calculation based on the “beam-independent slope difference”

[0079] The unnecessary target removal unit 76 calculates the beam-independent tilt difference from the current distance and the current angular power acquired by the current scanning operation and the previous distance and the previous angular power acquired by the previous scanning operation based on the following formula 4. This calculation is based on the fact that since beam-independent tilt differences are related to power fluctuations caused by multipath effects, it is possible to distinguish stationary vehicles and upper objects based on beam-independent tilt differences. [Beam-independent tilt difference] = [Current upbeam−(downbeam) tilt] − [Previous upbeam−(downbeam) tilt]

[0080] Fig. 11 is a view for explaining the calculation of the up-beam inclination and the down-beam inclination as parameters for calculating a beam-independent inclination difference according to the embodiment. As shown in Fig. 11, for example, in the case of detecting the angular power of a target by transmitting the up-beam and the down-beam at intervals of 50 msec with respect to the down-beam, the down-beam inclination is calculated by subtracting the previous angular power difference (which is, for example, (2) in Fig. 11) from the current angular power difference (which, for example, is (1) in Fig. 11) and dividing the subtraction result by the distance difference (= [previous distance] - [current distance]). Furthermore, with respect to the upbeam, the upbeam inclination is calculated by subtracting the previous angular power difference (which, for example, corresponds to (4) in Fig. 11) from the current angular power difference (which, for example, is (3) in Fig. 11) and dividing the subtraction result by the distance difference (= [previous distance] - [current distance]). Then, the beam-independent slope difference of the upbeam is calculated by subtracting the previous upbeam slope from the current upbeam slope. Furthermore, the beam-independent slope difference of the downbeam is calculated by subtracting the previous downbeam slope from the current downbeam slope.

[0081] Next, calculate the unit distance of unnecessary targets 76 from an example of the beam-independent tilt difference models 63c shown in Fig. 12 and the beam-independent inclination difference calculated based on Formula 4, the “likelihood ratio 3” based on Formula 1. Fig. 12 is a view of a beam-independent tilt difference model according to the embodiment. The beam-independent tilt difference model 63c is a probability distribution model that has a horizontal axis indicating a beam-independent tilt difference in dB / m and a vertical axis indicating the probability, and represents the relationship between the beam-independent tilt difference and the probability of being a stationary vehicle, and the relationship between the beam-independent tilt difference and the probability of being an upper object. The probability distribution model in Fig. 12, which represents the probability of being a stationary vehicle and the probability of being an upper object, is a model constructed in advance based on, for example, normal distributions using a maximum likelihood estimation method and an experimental design method. In the beam-independent inclination difference model 63c, the parameter characterizing the model can be adjusted for stationary vehicles and upper objects, respectively, to improve the accuracy of the determination.

[0082] In addition, Fig. 12, as an example of the beam-independent tilt difference models 63c, the beam-independent tilt difference model is shown that applies to the case where the distance from the vehicle A to a target is 80 m, which is one of the distances set at 10 m intervals from 10 m to about 150 m. Beam-independent tilt difference models that apply to the case where the distance from the vehicle A to a target is other than the distances are not shown.

[0083] For example, it is assumed that the “beam-independent slope difference”, which is calculated based on formula 4, is “0”. In this case, Fig. 12, when the beam-independent inclination difference is "0" on the horizontal axis, the probability of being a stationary vehicle and the probability of being an upper object are "0.14" and "0.05" on the vertical axis, respectively. Therefore, by substituting "0.14" and "0.05" for the terms "probability 3 of being a stationary vehicle" and "probability 3 of being an upper object" in Formula 1, "[Probability ratio 3] = log(Probability 3 of being a stationary vehicle) - log(Probability 3 of being an upper object)" (Formula 1 when "n" is 3), the probability ratio 3 can be calculated. • STEP 1-4: Logarithmic likelihood ratio calculation based on the “upward / downward beam power difference slope”

[0084] The unnecessary target removal unit 76 calculates the up / down beam power difference inclination based on the following formula 5. This calculation is based on the fact that since up / down beam power difference inclinations are related to power fluctuations caused by multipath effects, it is possible to distinguish stationary vehicles and upper objects based on up / down beam power difference inclinations. [Upward / Downward Beam Power Difference Slope]=[Current Angular Power Difference]−[Previous Angular Power Difference][d1−d2]

[0085] Here, the term “[actual angular power difference]” can be interpreted as “[Interpolated upstream angular power (see (3') in Fig. 13 for example)] - [Current down beam angle power (see (3) in Fig. 13 for example)]” and the term “[Previous angular power difference]” as “(Previous up-beam angular power (see (2) in Fig. 13 for example)] - [Downward beam angle power interpolation (see (2') in Fig. 13 for example)]”.

[0086] Fig. 13 is a view for explaining an up / down beam power difference tilt calculation method according to the embodiment. As shown in Fig. 13, for example, in the case of detecting the angular power of a target by transmitting the up-beam and the down-beam at intervals of 50 msec, linear interpolation is performed on the previous down-beam angular power (which is, for example, (1) in Fig. 13) and the current downward beam angle power (which, for example, corresponds to (3) in Fig. 13) is performed, resulting in a downward beam angle power interpolation value (which is, for example, (2`) in Fig. 13) which corresponds to a distance d1 corresponding to the previous upward beam angle power (which, for example, is (2) in Fig. 13). In addition, linear interpolation is performed on the previous up-beam angle power (which, for example, is shown in (2) Fig. 13) and the current up-beam angle power (which, for example, corresponds to (4) in Fig. 13) is performed, whereby an up-beam angle power interpolation value (which corresponds, for example, to (3`) in Fig. 13), which corresponds to a distance d2, the current downward beam angle power (which, for example, is (3) in Fig. 13), is calculated.

[0087] Because the upstream and downstream beams are transmitted alternately and at 50 msec intervals, there are time delays between the times at which angular powers are detected by transmitting the upstream beam and the times at which angular powers are detected by transmitting the downstream beam. Therefore, these interpolation values ​​are used to calculate more accurate angular power differences at corresponding times.

[0088] Next, the unnecessary target removal unit 76 calculates the current angular power difference by subtracting the current down beam angular power (which is, for example, (3) in Fig. 13) from the upward angular power interpolation value (which is, for example, (3`) in Fig. 13). In addition, the unnecessary target removal unit 76 calculates the current angular power difference by subtracting the downward angular power interpolation value (which is, for example, (2`) in Fig. 13) from the current up-beam angle power (which, for example, is (2) in Fig. 13). Then, the unnecessary target removal unit 76 calculates the up / down beam power difference inclination based on Formula 5. In Formula 5, “d1” and “d2” are “d1” and “d2” in Fig. 13.

[0089] Next, the unit distance of unnecessary targets 76 is calculated from an example of the up / down beam power difference inclination models 63d shown in Fig. 14 and the up / down beam power difference slope calculated based on Formula 5, the “likelihood ratio 4” based on Formula 1. Fig. 14 is a view illustrating the up / down beam power difference slope model according to the embodiment. The up / down beam power difference slope model 63d is a probability distribution model having a horizontal axis indicating an up / down beam power difference slope dB / m and a vertical axis indicating the probability, representing the relationship between the up / down beam power difference slope and the probability of being a stationary vehicle, and the relationship between the up / down beam power difference slope and the probability of being an upper object. The probability distribution model in Fig. 14, which represents the probability of being a stationary vehicle and the probability of being an upper object, is a model constructed in advance based on, for example, normal distributions using a maximum likelihood estimation method and an experimental design method. In the up / down beam power difference inclination model 63d, the parameter characterizing the model can be adjusted for stationary vehicles and upper objects, respectively, to improve the determination accuracy.

[0090] In addition, Fig. 14, as an example of the up / down beam power difference inclination models 63d, the up / down beam power difference inclination model is shown, which is applicable to the case where the distance from the vehicle A to a target is 80 m, which is one of the distances set at 10 m intervals from 10 m to about 150 m. Up / down beam power difference inclination models applicable to the case where the distance from the vehicle A to a target is other than the distances are not shown.

[0091] For example, it is assumed that the “Up / Down beam power difference slope” calculated based on Formula 5 is “0”. In this case, Fig. 14, when the up / down beam power difference slope is "0" on the horizontal axis, the probability of being a stationary vehicle and the probability of being an upper object are "0.2" and "0.06" on the vertical axis, respectively. Therefore, by substituting "0.2" and "0.06" for the terms "probability 4 of being a stationary vehicle" and "probability 4 of being an upper object" in Formula 1, "[Probability Ratio 4] = log(Probability 4 of being a stationary vehicle) - log(Probability 4 of being an upper object)" (Formula 1 when "n" is 4), the probability ratio 4 can be calculated. • STEPS 1-5: Logarithmic likelihood ratio calculation based on the “upstream / downstream beam range differential”

[0092] Multipath effects of reflected waves affect angular performance, and the influence of multipath effects depends on the height of the targets. In the case of detecting a stationary vehicle and an upper object that exist at a large distance from the vehicle, the influence of multipath effects on the angular performance of the upper object is more obvious than the influence of multipath effects on the angular performance of the stationary vehicle.Furthermore, the distances at which nulls attributable to multipath appear (and the number of such nulls appearing at each distance) are unlikely to be affected by the radar installation angle relative to the vertical direction; in other words, the "up / down beam range differential" parameter, obtained by normalizing the variation of the up / down beam power difference with distance, is unlikely to be affected by the radar installation condition. A null refers to a region in the antenna radiation pattern between radiation lobes that are concave at the top near their maximum points and whose shapes near their minimum points resemble the shape of a cycloidal curve near their minimum point.

[0093] Fig. 15A is a view illustrating the relationship between the angular power and distance of a stationary vehicle. Fig. 15B is a view illustrating the relationship between the angular power and distance of an upper object. From a part in a Fig. 15A, it can be seen that at a distance of less than about 70 m or 80 m, since the influence of multipath effects on the stationary vehicle is strong, the angular power distribution is wide and the amount of change is large. At the same time, a part of a Fig. 15B, it can be seen that regardless of the distance, since the influence of multipath effects on the upper object is smaller, the angular power distribution is narrow and the amount of change is small.

[0094] For this reason, the unnecessary target removal unit 76 calculates the up / down beam range differential by means of a Fig. 16 shown procedure. Fig. 16 is a view for explaining the uplink / downlink beam range differential calculation method according to the embodiment. As shown in Fig. 16, for example, in the case of detecting the angular power of a target by transmitting the up-beam and the down-beam at intervals of 50 msec, relative to each angular power detected by transmitting one of the up-beam and the down-beam, linear interpolation is performed on angular powers detected by transmitting the other beam, thereby estimating an angular power corresponding to the distance corresponding to the detected angular power, and relative to the individual distances, angular power differences between the up-beam and the down-beam (corresponding to (1) to (6) in (a) of Fig. 16) is calculated.

[0095] Because the upstream and downstream beams are transmitted alternately and at 50 msec intervals, there are time delays between the times at which angular powers are detected by transmitting the upstream beam and the times at which angular powers are detected by transmitting the downstream beam. Therefore, interpolation values ​​obtained with respect to individual distances by linear interpolation are used to calculate more accurate angular power differences at appropriate times.

[0096] Next, the distance unit 76 calculates as in (b) in Fig. As shown in Figure 16, calculate the "up / down beam power difference" via |[angular power difference of (1)] - [angular power difference of (2)]| and the "distance difference" via |[distance from (1)] - [distance from (2)]|, and obtain the range of (1') by multiplying the "up / down beam power difference" and the "distance difference." The notation |*| denotes the absolute value of *.

[0097] Similarly, the distance unit of unnecessary targets calculates the range of (2'), the range of (3'), the range of (4'), and the range of (5'), respectively, via "|[Angular power difference of (2)] - [Angular power difference of (3)]lx{[Distance from (2)] - [Distance from (3)]}", "|[Angular power difference of (3)] - [Angular power difference of (4)]|×{[Distance from (3)] - [Distance from (4)]}", "|[Angular power difference of (4)] - [Angular power difference of (5)]|×{[Distance from (4)] - [Distance (5)]}" and "|[Angular power difference of (5)] - [Angular power difference of (6)]|×{[Distance from (5)] - [Distance (6)].

[0098] In Fig. Figure 16 (c) shows the ranges of the individual cycles calculated using the above method. Next, the unnecessary target removal unit 76 calculates the "up / down beam range differential" by dividing the sum of the ranges of the individual cycles (the up / down beam power differences) calculated with respect to the same target as described above by the sum of the individual range differences. [Up / down beam range differential] = ∑ [Up / down beam power differences] ∑ [Distance differences]

[0099] Next, the unit distance of unnecessary targets 76 is calculated from an example of the up / down beam range differential models 63e shown in Fig. 17 and the “up / down beam range differential” calculated as described above, the “likelihood ratio 5” based on Formula 1. Fig. Figure 17 is a view illustrating the up / down beam range differential model according to the embodiment. The up / down beam range differential model 63e is a probability distribution model having a horizontal axis indicating an up / down beam range differential in dB and a vertical axis indicating the probability and representing the relationship between the up / down beam range differential and the probability of being a stationary vehicle and the relationship between the up / down beam range differential and the probability of being an upper object. The probability distribution model in Fig. 17, which represents the probability of being a stationary vehicle and the probability of being an upper object, is a model constructed based on normal distributions, for example. In the upstream / downstream beam range differential model 63e, the parameter characterizing the model can be adjusted for stationary vehicles and upper objects, respectively, to improve the accuracy of the determination.

[0100] In addition, Fig. 17, as an example of the up / down beam range differential models 63e, the up / down beam range differential model is shown, which is applicable to the case where the distance from the vehicle A to a target is 80 m, which is one of the distances set at 10 m intervals from 10 m to about 150 m. Relative speed difference models, which are applicable to the case where the distance from the vehicle A to the target is other than the distances, are not shown.

[0101] For example, it is assumed that the “upstream / downstream beam range differential” calculated based on formula 6 is “1”. In this case, Fig. 17, when the up / down beam range differential is "1" on the horizontal axis, the probability of being a stationary vehicle and the probability of being an upper object are "0.52" and "0.1" on the vertical axis, respectively. Therefore, by substituting "0.52" and "0.1" for the terms "probability 5 of being a stationary vehicle" and "probability 5 of being an upper object" in Formula 1, "[Probability Ratio 5] = log(Probability 5 of being a stationary vehicle) - log(Probability 5 of being an upper object)" (Formula 1 when "n" is 5), the probability ratio 5 can be calculated. • STEPS 1-6: Logarithmic probability ratio calculation based on the “extrapolation causes”

[0102] In the case of upper objects, reflection is unstable due to the influence of multi-point reflection, shifting of reflection points, power attenuation attributable to multipath effects, etc., so extrapolation is often applied. For this reason, the probability of being a stationary vehicle and the probability of being an upper object are calculated based on the characteristics of the extrapolation. In other words, with respect to each target, it is determined in the current scanning process whether extrapolation was performed, and if extrapolation was performed, the cause of the extrapolation is identified. Then, based on whether extrapolation was performed and the cause of the extrapolation, the logarithmic probability ratios are determined from the extrapolation-cause-based probability table 63f. Fig. 18 is a view illustrating an extrapolation-cause-based probability table according to the embodiment. In Fig. 18, ellipses "..." are used to indicate omissions of specific numerical values. For example, when continuity determination unit 73 performs an extrapolation process, it stores information about whether extrapolation was performed and the extrapolation cause in the predetermined storage area of ​​storage unit 63.

[0103] For example, there are seven extraplotion causes: “History non-existent,” “Peak non-existent,” “Angle estimation failure,” “Discontinuity,” “Out of range (in bins) based on estimated frequency,” “Invalid Mahalanobis distance,” and “Pair non-existent.”

[0104] The "history nonexistent" cause refers to the case where it is impossible to acquire a history corresponding to an extracted current peak frequency, or there is no history corresponding to an extracted current peak frequency. The "peak nonexistent" cause refers to the case where the peak extraction unit 70 cannot extract a peak from a frequency spectrum generated by the Fourier transform unit 62. The "angle estimation failure" cause refers to the case where, although the peak extraction unit 70 has extracted a peak, the angle estimation unit 70 cannot estimate a target angle.

[0105] The cause “discontinuity” refers to the case where, although the pairing unit 72 succeeded in pairing, the continuity determination unit 73 determines that the corresponding target data object has no continuity with a target data object derived in the last process.

[0106] The cause "Out of range (in bins) based on estimated frequency" refers to the case where the actual position of a current target data object is not within the predetermined range based on the estimated position for the current target data object (for example, within a range between "(Estimated peak frequency] - 3 (bins)" and "[Estimated peak frequency] + 3 (bins)").

[0107] The "illegal Mahalanobis distance" cause refers to the case where the pairing unit 72 fails to pair because the minimum value of the Mahalanobis distances is greater than the predetermined value. The "pair nonexistent" cause refers to the case where the pairing unit 72 fails to pair due to a cause other than the "history nonexistent," "pair nonexistent," "angle estimation failure," "discontinuity," "out of range (in bins) based on the estimated frequency," and "illegal Mahalanobis distance" causes.

[0108] In the case where it is determined in the current scanning operation that no extrapolation has been performed, that is, in the case where normal detection has been performed, the unnecessary target removal unit 76 reads the logarithmic probability of being a stationary vehicle and the logarithmic probability of being an upper object, corresponding to the detection distance, from a column of the extrapolation-cause-based probability table 63f in Fig. 18, which represents the logarithmic probability with respect to "normal detection." Referring to the line "95 < R ≤ 105," for example, in the case where the detection distance R is 100 m, the logarithmic probability of being a stationary vehicle is -0.14, and the logarithmic probability of being an upper object is -0.16. Therefore, by substituting “-0.14” and “-0.16” for the terms “log(probability 6 of being a stationary vehicle)” and “log(probability 6 of being the upper object)” in Formula 1 “[likelihood ratio 6] = log(probability 6 of being a stationary vehicle) - log(probability 6 of being the upper object)” (Formula 1 when “n” is 6), the likelihood ratio 6 can be calculated.

[0109] In the case where it is determined in the current scanning operation that extrapolation has been performed and the cause of the extrapolation is “history non-existent”, the unnecessary target removal unit 76 reads the logarithmic probability of being a stationary vehicle and the logarithmic probability of being an upper object, corresponding to the detection distance, from a column of the extrapolation cause-based probability table 63f in Fig. 18, which represents the logarithmic probability of the cause "history nonexistent." Referring to the line "95 < R ≤ 105," for example, in the case where the detection distance R is 100 m, the logarithmic probability of being a stationary vehicle is -2.48, and the logarithmic probability of being an upper object is -1.13. Therefore, by substituting "-2.48" and "-1.13" for the terms "log(probability 6 of being a stationary vehicle)" and "log(probability 6 of being the upper object)" in Formula 1, "[likelihood ratio 6] = log(probability 6 of being a stationary vehicle) - log(probability 6 of being the upper object)" (Formula 1 when "n" is 6), the likelihood ratio 6 can be calculated. The same applies to the other extrapolation causes. < STEP 2: Calculating the likelihood ratio OverAll>

[0110] The unnecessary target removal unit 76 calculates the sum of the likelihood ratio 1 to the likelihood ratio 6 in the above-described STEPS 1-1 to 1-6 as the (total) likelihood ratio OverAll based on the following formula 7. [Probability ratio OverAll]=[Probability ratio 1]+ [Probability ratio 2]+[Probability ratio 3]+[Probability ratio 4]+[Probability ratio 5]+[Probability ratio 6] < STEP 3: Process of distinguishing between stationary vehicles and upper objects>

[0111] The unnecessary target removal unit 76 compares the likelihood ratio OverAll calculated in STEP 2 described above with a predetermined threshold, and thereby determines whether the target is a stationary vehicle or an upper object. In other words, if the likelihood ratio OverAll is equal to or greater than the predetermined threshold, the unnecessary target removal unit 76 determines that the target is a stationary vehicle; whereas, if the likelihood ratio OverAll is less than the predetermined threshold, the unnecessary target removal unit determines that the object is an upper object.

[0112] As described above, in the first discrimination process, with respect to each of the plurality of parameters, probability density functions are respectively defined, which define the relationship between values ​​that a stationary vehicle can take and the occurrence frequencies of the values, and the relationship between the values ​​that an upper object can take. From the current values, the probability of being a stationary vehicle (the probability of being a stationary vehicle) and the probability of being an upper object (the probability of being an upper object) are calculated based on the probability density functions with respect to the plurality of parameters. Furthermore, in the first discrimination process, whenever data acquisition is performed, Bayesian updating is performed to update prior probabilities with posterior probabilities.By repeating the above-mentioned process, whether each target is a stationary vehicle or a top object is determined in the Bayesian filtering manner using the logarithmic relationship between the probability of being a stationary vehicle and the probability of being a top object.

[0113] Therefore, according to the first discrimination process, from a comparatively long distance (about 80 m) from the target, it is possible to determine whether a target detected in the moving direction of the vehicle is likely to collide with the vehicle (for example, whether a detected target is a target for which control of the vehicle such as brake control is required), and it is possible to improve the detection rate so that control of the vehicle based on target detection can be performed at an appropriate timing by appropriate instructions.

[0114] Furthermore, in the first discrimination process, the likelihood ratios 1 to 6 are parameters unlikely to be affected by the installation height of the radar device 1 on the vehicle A and the installation angle of the radar device relative to the vertical direction, and regardless of the radar installation condition, it is possible to distinguish between stationary vehicles and upper objects with high accuracy. (Second discrimination process of distinguishing between stationary vehicles and upper objects according to the embodiment)

[0115] There is a case where the discrimination process described above cannot distinguish between stationary vehicles and upper objects with great accuracy. In other words, as in Fig. As shown in Figure 19, some probability ratios OverAll can be acquired based on the multitude of parameters of stationary vehicles and can also be acquired from upper objects. Fig. Figure 19 is a view illustrating a probability ratio distribution including a range common to stationary vehicles and overhead objects. Therefore, in the following second discrimination process, a method for separating probability ratios Overall will be described.

[0116] Hereinafter, causes Hi (where i is 1 or 2) are defined as the following Table 1. [Table 1] Ursache Definition H1 Ziel ist stationäres Fahrzeug. H2 Ziel ist oberes Objekt.

[0117] In addition, parameters Dj (where j is a value between 1 and 7) are defined as the following Table 2. Furthermore, all parameters D1 to D7 in the following Table 2 are assumed to be independent. [Table 2] Parameter Definition D1 Wahrscheinlichkeitsverhältnis, das auf Grundlage der Seitenabstandsdifferenz in dem aktuellen Abtastvorgang berechnet wurde (Wahrscheinlichkeitsverhältnis 1) D2 Likelihood ratio calculated based on the relative speed difference in the current scanning process (likelihood ratio 2) D3 Likelihood ratio calculated based on the beam-independent tilt difference in the current scan (likelihood ratio 3) D4 Likelihood ratio calculated based on the up / down beam power difference slope in the current scan (likelihood ratio 4) D5 Likelihood ratio calculated based on the upstream / downstream beam range differential in the current scan (likelihood ratio 5) D6 Likelihood ratio calculated based on the extrapolation cause in the current sampling process (likelihood ratio 6) D7 Likelihood ratio calculated based on the total up / down beam variation in the current scan (likelihood ratio 7)

[0118] In this case, the posterior probability P(H1 / D), which is the probability that the target is a stationary vehicle, is expressed with the parameters D (D1 to D7) as the following formula 8-1. Furthermore, the posterior probability P(H2 / D), which is the probability that a target is an upper object, is expressed with the parameters D (D1 to D7) as the following formula 8-2. P(H1|D)=P(H1|D1,D2,D3,D5,D6,D7) P(H2|D)=P(H2|D1,D2,D3,D5,D6,D7)

[0119] In the case of using a naive Bayesian filter, whether a target is a stationary target or a top object is determined by the logit; this is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being a top object, as the following formula 9. Logit=log{P(H1|D) / P(H2|D)} =log{P(D|H1)P(H1) / P(D|H2)P(H2)}

[0120] Herein, P(D|H1) and P(D|H2), because the parameters D (D1 to D7) were assumed to be independent, are expressed as the following formulas 10-1 and 10-2, respectively. P(D|H1)=P(D1|H1)P(D2|H1)P(D3|H1)P(D4|H1)P(D5|H1)P(D6|H1)P(D7|H1) P(D|H2)=P(D1|H2)P(D2|H2)P(D3|H2)P(D4|H2)P(D5|H2)P(D6|H2)P(D7|H2)

[0121] Using formulas 10-1 and 10-2, formula 9 is converted into the following formulas 11-1 and 11-2. Logit=log{(P(H1|D) / P(H2|D))}(11−1)=log{P(D1|H1)}−log{P(D1|H2)}(denoted as a part x1)+log{P(D2|H1)}−log{P(D2|H2)}(denoted as a part x2)+log{P(D3|H1)}−log{P(D3|H2)}(denoted as a part x3)+log{P(D4|H1)}−log{P(D4|H2)}(denoted as a part x4)+log{P(D5|H1)}−log{P(D5|H2)}(denoted as a part x5)+log{P(D6|H1)}−log{P(D6|H2)}(denoted as a part x6 denoted)+log{P(D7|H1)}−log{P(D7|H2)}(denoted as a part x7)+log{P(H1)}−log{P(H2)}(denoted as a part x0) (11−2)

[0122] When the first to eighth parts of Formula 11-2 are denoted as x1, x2, x3, x4, x5, x6, x7, and x0, respectively, as shown in Formula 11-2, Formula 11-2 is expressed as the following Formula 12. Logit=(x1+x2+x3+x4+x5+x6+x7)+x0

[0123] In Formula 12, the part in parentheses can be interpreted as a generalized linear model using x1 to x7 as predictor variables. Therefore, in the second discrimination process, the regression coefficients and intercepts are determined from the multivariate analysis of the part in parentheses of Formula 12, thereby improving the accuracy of distinguishing between stationary vehicles and upper objects.

[0124] Target data used to determine whether a target is a stationary vehicle or an overhead object is qualitative data, such as data on stationary vehicles and overhead objects. It is also possible to assume that all predictor variables (from the likelihood ratio 1 (x1) to the likelihood ratio 7 (x7)) are normal. Therefore, logistic regression analysis is used as a method for multivariate analysis. Considering multicollinearity and Akaike's information criterion (AIC), selection and integration of predictor variables are performed to determine a regression coefficient (weighting coefficient) and an intercept at each of the detection distances set at 10 m intervals. In this case, a coefficient (weighting coefficient) and an intercept are calculated by linear interpolation for each distance in the detection distance intervals.

[0125] In the second discrimination process, for example, in the case where the first detection distance of a target is a first distance between 80 m and 110 m, the weighting coefficient / intersection table A 63g (to be described later) is used to determine whether the target is a stationary vehicle or an upper object, and in the case where the first distance at which the target was first detected is not the first distance, the weighting coefficient / intersection table B 63h (to be described later) is used. As a result, the accuracy of discrimination between stationary vehicles and upper objects improves.

[0126] In the second discrimination process, the weighting coefficient / intersection table A 63g is used for recursively estimating individual probability ratios based on a first detection distance at which a target is detected, and the weighting coefficient / intersection table B 63h is used for recursively estimating individual probability ratios based on a first detection distance.

[0127] Fig. 20A is a view illustrating the weighting coefficient / intersection table A for recursively estimating individual likelihood ratios based on a first detection distance according to the embodiment.

[0128] Fig. 20B is a view illustrating the weighting coefficient / intersection table B for recursively estimating individual likelihood ratios based on a first detection distance according to the embodiment.

[0129] In the Fig. In the weighting coefficient / intersection table A 63g shown in Fig. 20A, weighting coefficients αi and intersection points β to be used to correct likelihood ratios i (where i is 1, 2, 3, 5, and 6) in the case where the first detection distance of a target is, for example, the first distance between 80 m and 100 m, are stored in association with each detection distance. In the Fig. In the weighting coefficient / intersection table B 63h shown in Fig. 20B, the weighting coefficients αi and intersection points β to be used to correct likelihood ratios i (where i is 1, 2, 3, 5, and 6) in the case where the first detection distance of a target is a different distance than the first range are stored in association with the detection distance.

[0130] In Fig. 20A and Fig.20B, from the predictor variable (parameter) selection based on multicollinearity and the aforementioned AIC, parameters D1 to D6, ie, likelihood ratios 1 to 6, were selected, and the likelihood ratio corresponding to parameter D7 was excluded. In addition, Fig. 20A and Fig. 20B, the parameters D3 and D4, ie probability ratios 3 and 4, from the predictor variable (parameter) selection are integrated into one. Therefore, in Fig. 20A and Fig. 20B Weighting coefficients αi (where i is 1, 2, 3, 5, and 6) and intersection points β likelihood ratios 1 to 3 and likelihood ratios 5 and 6 are stored in association with each detection distance.

[0131] The unnecessary target removal unit 76 performs regression correction on the likelihood ratios i (where i is 1 to 6) calculated in the same manner as those of the first discrimination process described above, using weighting coefficients and intersections read from the weighting coefficient / intersection table A 63g or the weighting coefficient / intersection table B 63h, and calculates a corrected likelihood ratio OverAll based on the following formula 13. [Corrected probability ratio OverAll]=(α1×[probability ratio 1]+β)+(α2×[probability ratio 2]+β)+α3×[probability ratio 3]+[probability ratio 4]2+β+α5×[probability ratio 5]+β+α6×[probability ratio 6]+β

[0132] In the case where the first detection distance is the above-mentioned first distance, the unnecessary target removing unit 76 reads the weighting coefficient αi and intersection points β for the individual probability ratios i (where i is 1 to 6) corresponding to each distance at which a scan was performed from the weighting coefficient / intersection table A 63g. Meanwhile, in the case where a first detection distance is not the above-mentioned first distance, the unnecessary target removing unit 76 reads weighting coefficients αi and intersection points β for the individual probability ratios i (where i is 1 to 6) corresponding to each distance at which a scan was performed from the weighting coefficient / intersection table B 63h.Variations in the weighting coefficients αi (where i is 1 to 6) and intersection points β read from the weighting coefficient / intersection table A 63g according to target detection distances in the case where the first detection distance is the first distance are relatively larger than variations in the weighting coefficients αi (where i is 1 to 6) and intersection points β read from the weighting coefficient / intersection table B 63h according to the target detection distances in the case where the first detection distance is not the first distance. The unnecessary target removal unit 76 calculates the corrected likelihood ratio OverAll from the read weighting coefficients αi and intersection points β for the individual likelihood ratios i (where i is 1 to 6) and the individual likelihood ratios i based on Formula 13.

[0133] In the second discrimination process, with respect to likelihood ratios updated by Bayesian updating whenever a sampling operation is performed, an overall likelihood ratio is calculated from the individual parameters (likelihood ratios 1 to 6), weighting is performed using weighting coefficients and intercepts depending on the initial detection distance of a target and obtained through logistic regression analysis. Therefore, in the second discrimination process, it is possible to improve the accuracy of distinguishing between stationary vehicles and overhead objects.

[0134] As also described above, likelihood ratio 1 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, obtained based on the "lateral position difference" parameter. Likelihood ratio 2 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, obtained based on the "relative velocity difference" parameter. Likelihood ratio 3 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, based on the "beam-independent inclination difference" parameter.The likelihood ratio 4 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, obtained based on the "upward / downward beam power difference slope" parameter. The likelihood ratio 5 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, based on the "upward / downward beam range differential" parameter. The likelihood ratio 6 is the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object, obtained based on the "extrapolation cause" parameter.

[0135] Now the characteristics of the weighting coefficients for the probability ratios used in the Fig.20A. Weighting coefficients α1 for likelihood ratio 1 to be used in the case where the distance is relatively short (for example, 80 m or less) are larger than weighting coefficients α1 to be used in the case where the distance is relatively long (for example, a distance of more than 80 m). Compared with the case where the distance to the target is relatively short, in the case where the distance is relatively long, since the lateral position differences of the upper objects become larger, the importance of the "lateral position difference" parameter becomes greater in determining whether a target is a stationary vehicle or an upper object.For this reason, a weighting coefficient α1 for the likelihood ratio 1 to be used in the case where the distance of the target is longer is set to be larger than a weighting coefficient α1 for the likelihood ratio 1 to be used in the case where the distance is relatively shorter.

[0136] In addition, the weighting coefficients α2 for the likelihood ratio 2 to be used in the case where the distance is relative are larger than the weighting coefficients α2 to be used in the case where the distance is relatively long. This is because as the distance decreases, since the intensities of the waves reflected from the target increase and it becomes possible to calculate the relative speed difference of the target more accurately, the significance of the "relative speed difference" parameter in determining whether the target is a stationary vehicle or an upper object becomes higher. However, as the distance to the target continues to decrease (for example, when the distance becomes 60 m or less), smaller weighting coefficients α2 are used.The reason for this is that as the distance decreases, the target disappears from the transmission range of the transmitted wave of the radar device 1; a reflected wave is therefore not received.

[0137] Furthermore, weighting coefficients α5 for the likelihood ratio 5 to be used in the case where the distance is relatively short are larger than weighting coefficients α5 to be used in the case where the distance is relatively long. This is because as the distance decreases, it is possible to detect more range differentials; the importance of the "up / down beam range differential" parameter becomes greater in determining whether the target is a stationary vehicle or an upper object. As described above, focusing on variations in the values ​​of the various parameters according to the distance of a stationary vehicle and variations in the values ​​of the various parameters according to the distance of an upper object, weighting coefficients for the various parameters are set so that they vary depending on target distances.It is therefore possible to distinguish precisely between stationary vehicles and upper objects.

[0138] Also, in the case where a distance (a first detection distance) at which the radar device 1 has detected a target of an object for the first time is not the first distance (for example, between 80 m and 110 m), weighting coefficients αi are calculated based on the values ​​shown in Fig. 20B. Meanwhile, when the first detection distance of a target is the first distance, weighting coefficients αi are calculated based on the weighting coefficient / intersection table B 63h shown in Fig. 20A shown weighting coefficient / intercept table A 63g. Thus, when calculating the weighting coefficients αi, weighting coefficients αi are calculated based on the Fig.20B, and if the distance of a target becomes a specific distance such as the first distance, weighting coefficients αi are calculated based on another table (for example, the one shown in Fig. 20A, the weighting coefficient / intersection table A (63g) is calculated using different weighting coefficients for the same parameters. Therefore, based on the distance to a target, optimal weighting can be performed on parameters used to determine the type of target, and the type of target can be reliably determined. [MODIFICATION OF THE FIRST AND SECOND DISCRIMINATION PROCESS](1) Likelihood Ratio OverAll

[0139] In the first discrimination process, if the likelihood ratio OverAll of a target is equal to or greater than the threshold, it is determined that the target is a stationary vehicle; whereas, if the likelihood ratio OverAll is less than the threshold, it is determined that the target is an upper object. The same applies to each corrected likelihood ratio OverAll of the second discrimination process. However, the determination of whether a target is a stationary target or an upper object is not limited to this. In the case of making the determination by comparing the certainty of being a stationary vehicle with a predetermined threshold, a value based on the likelihood ratio OverAll may be used as a magnification ratio C by which the certainty of being a stationary vehicle is multiplied.In other words, the product of the magnification ratio C and the certainty of being a stationary vehicle is calculated as the certainty of being a stationary vehicle compared to the threshold value predetermined for determination, and in the case where the certainty is equal to or greater than the predetermined threshold value, it is determined that the corresponding target is a stationary vehicle.

[0140] The certainty of being a stationary vehicle is an index indicating whether the target data object is a data object on a stationary vehicle. It can take a value, for example, between 0 and 100. As the certainty value increases, the probability of being a stationary vehicle also increases. The certainty of being a stationary vehicle is calculated using a variety of information objects contained in a target data object (such as the longitudinal distance, angular power, extrapolation frequency, etc.).

[0141] For example, assume that two thresholds, i.e., a first threshold and a second threshold (less than the first threshold), are provided. In the case where the likelihood ratio OverAll is equal to or greater than the first threshold, the magnification ratio C is set to 1. Since it is possible to determine that the certainty of being a stationary vehicle is high, the certainty of being a stationary vehicle is compared with the predetermined threshold without change to determine whether the target is a stationary vehicle. Meanwhile, in the case where a likelihood ratio OverAll is equal to or less than the second threshold, the magnification ratio C is set to 0.Since it is possible to determine that the certainty of being a stationary vehicle is low, 0 is set to make the certainty of being a stationary vehicle 0 to prevent the target from being determined to be a stationary vehicle.

[0142] Even in the case where the likelihood ratio OverAll is smaller than the first threshold and larger than the second threshold, the magnification ratio C is calculated as "([likelihood ratio OverAll] - [second threshold]) / ([first threshold] - [second threshold])." In other words, the magnification ratio C indicates the ratio of the likelihood ratio OverAll exceeding the second threshold to the first threshold exceeding the second threshold. For example, the case where the magnification ratio C is 0.5 represents that the product of 0.5 and the certainty of being a stationary vehicle is compared with the predetermined threshold to determine whether the target is a stationary vehicle.

[0143] As described above, the probability ratio OverAll is converted into a magnification ratio C, by which the certainty of being a stationary vehicle is multiplied, to expand the line for determining whether a target is a stationary vehicle. Therefore, it is possible to more comprehensively distinguish stationary vehicles by considering various factors. (2) Distinction between stationary vehicles and targets other than stationary vehicles

[0144] The first and second discrimination processes use the Bayesian filtering method to update the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object corresponding to each of the various parameters, based on the probability models calculated by calculating the various parameters with respect to each of the stationary vehicles and upper objects detected by each target detection process, and representing the correlation between the corresponding parameters and the probability of being a stationary vehicle and the probability of being an upper object with respect to each detection distance, and determining whether a target is a stationary vehicle or an upper object.However, the Bayesian filtering method is not limited to this and can also be applied to distinguish between stationary vehicles and targets other than stationary vehicles (for example, lower objects and the like).

[0145] Probabilistic models representing, for example, the correlation between each of the various parameters and the probability of being a stationary vehicle, and the correlation between the corresponding parameter and the probability of being a target other than a stationary vehicle with respect to each detection distance, are constructed in advance by calculating the various parameters related to each of the stationary vehicles and targets other than stationary vehicles whenever target detection is performed. These probabilistic models can be used to calculate the logarithm of the ratio of the probability of being a stationary vehicle to the probability of being an upper object corresponding to each of the various parameters and determine whether a target is a stationary vehicle or a target other than stationary vehicles. (THIRD DISTINGUISHING PROCESS OF DISTINGUISHING BETWEEN STATIONARY VEHICLES AND UPPER OBJECTS ACCORDING TO THE EMBODIMENT)

[0146] Next, a third discrimination process for discriminating between stationary vehicles and upper objects according to the embodiment will be described. The following mainly describes the differences from the first discrimination process, and some parts will be omitted to avoid duplication.

[0147] In the first discrimination process described above, in the process of removing unnecessary targets, a likelihood ratio OverAll is used as an index to determine whether a target is a stationary vehicle or an upper object (hereinafter also referred to as determination at a target). A likelihood ratio OverAll is also an example of a first index.

[0148] As mentioned above with reference to Fig. As described in Section 19, some overall probability ratios can be detected by stationary vehicles, and they can also be detected by upper objects. Therefore, in the third discrimination process, an index that differs from an overall probability ratio is additionally used to perform determination at a target.

[0149] Therefore, in the third discrimination process, for example, it is possible to perform determination on the target with high accuracy even in the case where it is difficult to unambiguously determine whether a target is a stationary vehicle or an upper object based on a likelihood ratio OverAll.

[0150] In particular, in the case where a target (a target T) is, for example, a stationary vehicle (the target T1), the wave reflected from the stationary vehicle penetrates when the vehicle A connected to the radar device 1 (see Fig. 1) approaches the stationary vehicle, it more easily penetrates the front surface of the receiving unit RX. Therefore, the angular power in the radar device, which is the received power of the reflected wave, tends to increase.

[0151] Meanwhile, in the case where a target is an upper object (target T2), as the vehicle approaches the upper object, the angular power in radar device 1 gradually disappears from the main transmission range of the transmission wave. Therefore, the angular power in radar device 1 tends to weaken or rarely changes.

[0152] For this reason, in the third discrimination process, considering the feature that in the case where a target is an upper object, the angular power is likely to weaken, a second index that calculates the amount of attenuation of the angular power according to the detection distance of each target (the length distance to the target) is calculated, and determination is made on the corresponding target based on the second index. As the second index, for example, an attenuation amount range A1 (see Fig. 21A, etc.) and this case is described below.

[0153] Furthermore, in the third discrimination process, considering the feature that in the case where a target is a stationary vehicle, the angular power is likely to increase, a third index representing the amount of increase in the angular power according to the detection distance of each target is calculated, and based on the third index, determination is made on the corresponding target. As the third index, for example, an increase amount range A2 (see Fig. 22A etc.), which is to be described below.

[0154] Next, an attenuation amount range A1, which is the second index, will be described first. The unnecessary target removal unit 76 calculates an attenuation amount range A1 based on the following formula 14. [Attenuation amount range A1]=∑{([Current angular power]−[Reference power])×[Difference between previous distance and current distance]}

[0155] Fig. 21A and Fig. 21B are views for explaining a method of calculating an attenuation amount range A1 according to the embodiment. Fig. 21A and Fig. 21 B are also diagrams with horizontal axes indicating distance (detection distance) and vertical axes indicating angular power. Fig. 21A is also an example of a diagram that can be obtained in the case where a target is an upper object, and Fig. 21B is an example of a diagram that can be obtained in the case where a target is a stationary vehicle.

[0156] In the case where an angular power obtained in the current process (1) in Fig.21A, the distance unit of unnecessary targets 76 sets the maximum value of the angular powers obtained in the past, and current processes are set as a reference power. In the Fig. 21A, an angular power Pa (corresponding to (2) in Fig. 21A) obtained in the previous process is the maximum value. Therefore, the unnecessary target removal unit 76 sets the angular power Pa as a reference power.

[0157] Then the distance unit of unnecessary targets 76 subtracts the reference power (corresponding to (2) in Fig. 21A) from the angular power recorded in the current process (corresponding to (1) in Fig.21A), and obtains the product of the calculated difference and the distance difference between the distance detected in the previous process and the distance detected in the current process (the difference between the previous distance and the current distance). As shown in Formula 14, the unnecessary target removal unit 76 performs such a calculation whenever the angular power of the same target is obtained, and calculates the sum of the results of such calculations as the attenuation amount range A1. In other words, the unnecessary target removal unit 76 calculates the attenuation amount range A1 using, for example, section measurement.

[0158] As in Fig.As shown in Figure 21A, in the case where a target is an upper object, the angular power attenuates as the vehicle approaches the upper object. Therefore, the attenuation amount range A1 increases whenever the process of calculating Formula 14 is repeated. In Fig. 21A, the hatched part represents the attenuation amount range A1 obtained when the target is at a distance Ya.

[0159] In contrast, in the case where a target is a stationary vehicle, as in Fig. 21B, the attenuation amount range A1 is less likely to increase other than in the case of an upper object. In particular, in the case where the angular power obtained in the current process, for example, (1) in Fig. 21B, sets the unit distance of unnecessary targets 76, since an angular power Pb obtained from the previous process (corresponding to (2) in Fig.21B), the maximum value is set, the angular power Pb as a reference power.

[0160] Then, the unnecessary target removal unit 76 calculates the attenuation amount range A1 by subtracting the reference power (corresponding to (2) in Fig. 21B) of the angular power used in the current process (corresponding to (1) in Fig. 21B), and obtaining the product of the calculated difference and the distance difference between the distance detected in the previous process and the distance detected in the current process (the difference between the previous distance and the current distance), and adding the product to the previous attenuation amount range.

[0161] In the case where a target is a stationary vehicle, the angular power may temporarily decrease for a specific reason, such as the influence of multipath effects; however, the angular power generally increases as the vehicle approaches the stationary vehicle. Therefore, for example, in the example in Fig. 21B, in the case where an angular power Pc (corresponding to (3) in Fig. 21B) in the current process, since the angular power Pc is greater than the angular power Pb set as the reference power, the unit distance of unnecessary targets 76 the reference power with the angular power Pc, which is the new maximum value.

[0162] In this case, since both of the terms “[Current angular power]” and “[Reference power]” in Formula 14 become the angular power Pc, the value of the part “(Current angular power] - [Reference power]” becomes zero. Therefore, the attenuation amount range A1 does not increase.

[0163] The reference power is given in the example in Fig. 21B also continuously updated until an angular power Pd, as determined by (4) in Fig. 21B, so that the attenuation amount range A1 does not increase. In the case where a target is a stationary vehicle, such as in the example in Fig. 21B, the attenuation range A1, after the angular power Pd shown by (4) is obtained, only slightly increases when the angular power temporarily attenuates for some reason such as the influence of multipath effects.

[0164] Next, the unnecessary target removal unit 76 compares the attenuation amount range A1 calculated as described above with a threshold value A1a. If the attenuation amount range A1 is equal to or greater than the threshold value A1a, the unnecessary target removal unit determines that the target is an upper object. Meanwhile, if the attenuation amount range A1 is smaller than the threshold value A1a, the unnecessary target removal unit 76 determines that the target is not an upper object, that is, the probability of being a stationary vehicle is high.

[0165] As described above, in the third discrimination process, since the attenuation amount range A1 uses the amount of attenuation of the angular power according to the target detection distance, it is possible to determine with high accuracy whether the target is a stationary vehicle or an upper object. In other words, since the determination is made based on the sum of the amounts of attenuation of the signal power obtained with respect to the target in a plurality of target detection processes, unlike the case of making the determination based on the instantaneous value of the attenuation amount of the signal power obtained with respect to the target in one target detection process, it is possible to determine more accurately whether the target is a stationary object or an upper object.

[0166] Furthermore, in the third discrimination process, in the case where the angular power obtained in the current process is larger than the reference power used in the previous process, the reference power is updated with the angular power obtained in the current process.

[0167] Therefore, for example, in the case where a target is a stationary vehicle and the angular power shows a tendency to increase continuously (see Fig. 21B), it is possible to prevent the attenuation amount range A1 from increasing slightly.

[0168] Therefore, when the attenuation amount range A1 is compared with the threshold A1a, if the attenuation amount range A1 is smaller than the threshold A1a, it is possible to determine that the probability that the target is a stationary vehicle is high.

[0169] Next, a rise amount range A2, which is a third index, will be described. The unnecessary target removal unit 76 calculates a rise amount range A2 based on the following formula 15. [Rise amount range A2]∑{([Current angular power]−[Reference value])×[Difference between previous distance and current distance]}

[0170] Fig. 22A and Fig. 22B are views for explaining a method of calculating a rise amount range A2 according to the embodiment. Fig. 22A and Fig. 22B are also diagrams with horizontal axes indicating distance (detection distance) and vertical axes indicating angular power. Fig. 22A is also an example of a diagram that can be obtained in the case where a target is an upper object, and Fig.22B is an example of a diagram that can be obtained in the case where a target is a stationary vehicle. Fig. 22A also shows an example of a state in which the angular power obtained in the case where a target is an upper object rarely changes.

[0171] First, with reference to Fig. 22B describes the method for calculating a gain range A2. The unnecessary target distance unit 76 sets a reference value based on the angular powers obtained at target detection distances equal to or greater than a predetermined distance Yb.

[0172] Specifically, the unnecessary target distance unit 76 calculates the average of the angular powers corresponding to the target distances equal to or greater than the predetermined distance Yb, that is, the angular powers from an angular power (1) in Fig.22B to an angular power which (2) in Fig. 22B. Then, the distance unit of unnecessary targets 76 sets the average value (referred to as the angular power Pe) as a reference value. Fig. 22B and Fig. 22A, which are to be described below, reference values ​​are indicated by white square markers.

[0173] Any distance can be set as the predetermined distance Yb. For example, a relatively long distance can be set, particularly a distance such that the reference value is set when the radar device 1 is far from the target.

[0174] Then, the unnecessary target removal unit 76 subtracts the reference value (herein the angular power Pe) from the angular power obtained in the current process (corresponding to, for example, (3) in Fig.22B), thereby obtaining the difference. Then, the unnecessary target distance unit 76 obtains the product of the obtained difference and the distance difference between the distance detected in the previous process and the distance detected in the current process (the difference between the previous distance and the current distance). The unnecessary target distance unit performs such a calculation whenever the angular power of the same target is obtained, and calculates the sum of the results of such calculations as the increase amount range A2.

[0175] As in Fig. As shown in Figure 22B, the angular power increases in the case where the target is a stationary target as the vehicle approaches the stationary vehicle. Therefore, the increase amount range A2 increases whenever the process of calculating Formula 15 is repeated. Fig.22B, the hatched part represents the increase amount range A2 obtained when the target is at a distance Yc.

[0176] Meanwhile, as in Fig. 22A, in the case where the target is an upper object, since the angular power rarely changes, unlike the case where the target is a stationary vehicle, the increase amount range A2 is less likely to increase. Specifically, the unnecessary target distance unit 76 sets the average (the angular power Pe) of the angular powers corresponding to the target distances equal to or greater than the predetermined distance Yb ​​as a reference value, that is, the angular powers from an angular power corresponding to (1) in Fig. 22B, to an angular power which (2) in Fig. 22B.

[0177] Then, the unnecessary target removal unit 76 subtracts the reference value (herein, the angular power Pe) from an angular power obtained in the current process (corresponding to, for example, (3) in Fig. 22A), whereby the difference is obtained. Here, the increase amount range A2 is an index indicating the amount of increase in the angular power as described above. Therefore, in the case where the obtained difference is a negative value, that is, in the case where the angular power has attenuated, it is not necessarily necessary to reflect the obtained difference in the increase amount range A2.

[0178] Specifically, only when the difference obtained by subtracting the reference value from the angular power obtained in the current process is a positive value, the distance unit 76 can obtain the product of the obtained difference and the difference between the previous distance and the current distance, and add the product to the previous increase amount range, thereby obtaining the increase amount range A2.

[0179] In Fig. 22A, the angular power shown in (3) is smaller than the reference value (herein, the angular power Pe). The difference between them is therefore a negative value. Therefore, this difference is not reflected in the increase amount range A2. In the example in Fig.22A, the difference when the reference value is subtracted from the angular power shown in (4) is also a positive value. Therefore, the unnecessary target distance unit 76 calculates the product of the obtained difference and the difference between the previous distance and the current distance, and reflects the product in the increase amount range A2. In this way, only the amounts of increase in the angular power are summed to obtain the increase amount range A2. It is therefore possible to improve the accuracy of the determination at the target, which will be described below.

[0180] In the example in Fig. 22A, the increase amount range A2 increases only slightly because the angular power rarely changes even if the angular power temporarily increases to a value larger than the reference value due to the influence of multipath effects.

[0181] Next, the unnecessary target removal unit 76 compares the increase amount range A2 calculated as described above with a threshold value A2a. If the increase amount range A2 is less than the threshold value A2a, the unnecessary target removal unit determines that the target is an upper object. Meanwhile, if the increase amount range A2 is equal to or greater than the threshold value A2a, the unnecessary target removal unit 76 determines that the target is not an upper object, that is, the possibility of being a stationary vehicle is high.

[0182] As described above, in the third discrimination process, since the increase amount range A2 representing the amount of increase of the angular power according to the target detection distance, it is possible to determine with high accuracy whether the target is a stationary vehicle or an upper object.

[0183] In addition, when calculating the attenuation amount range A1 or the gain amount range A2 as the angular power obtained in the current process, an instantaneous value is used. This makes it possible to improve the responsiveness when performing target determination. Although instantaneous values ​​of the angular power may be dispersed, since the attenuation amount range A1 or the gain amount range A2 is the sum of instantaneous values, such dispersion is unlikely to have a significant impact.

[0184] As described above, in the third discrimination process, the attenuation amount range A1 or the increase amount range A2 is used to determine the target. Therefore, in the case where instantaneous values ​​of the angular power clearly represent that the target is an upper object, regardless of the likelihood ratio Overall, it is possible to determine that the target is an upper object.

[0185] Therefore, in the third discrimination process, for example, it is possible to perform determination on the target with high accuracy even in the case where it is difficult to unambiguously determine whether a target is a stationary vehicle or an upper object based on the likelihood ratio OverAll.

[0186] Furthermore, in the third discrimination process, both an attenuation amount range A1 and an increase amount range A2 can be used to perform determination on a target. In this case, it is possible to perform determination on the target accurately.

[0187] For some reason, such as the influence of multipath effects, the angular power value may change drastically. Therefore, even if a target is, for example, an upper object, some instantaneous angular power values ​​may cause the attenuation magnitude range A1 or the gain magnitude range A2 to not represent a value that indicates the target is an upper object.

[0188] Even in this case, if one of the attenuation amount range A1 and the gain amount range A2 has a value that represents the target as an upper object, it is possible to determine that the target is an upper object. In other words, in the third discrimination process, it is possible to reduce the influence of multipath effects, and thus to accurately perform determination on a target.

[0189] Specifically, even though a target is an upper object, if the height of the upper object from the road is less than a certain value, the attenuation amount range A1 may become smaller than the threshold value A1a. In such a case, the value of the increase amount range A2 of the upper object with a height from the road less than the certain value becomes smaller than the threshold value A2a. The reason for this is that since the heights of upper objects from the road are higher than the heights of stationary vehicles, upper objects are less likely to be affected by multipath effects, so the increase amount ranges A2 of upper objects are smaller than those of stationary vehicles.As described above, in the case where a target is an upper object, even if the attenuation amount range A1 has a tendency to change like that of a stationary vehicle, if the increase amount range A2 is used to make the determination, it is possible to accurately determine the type of the target.

[0190] In the case where only accurate determination of upper targets is required, it may be considered to use only the increase amount range A2 to determine whether a target is an upper object. However, even in the case where a target is a stationary vehicle (for example, a truck), if the height of the stationary vehicle's body from the road is equal to or greater than a certain value, the angular performance is likely to be affected by multipath effects. For this reason, not only determination using the increase amount range A2 with a feature that appears remarkable in the case where a target is a stationary target, but also determination using the attenuation amount range A1 with a feature that appears remarkable in the case where a target is an upper object is performed.In this case, based on the multitude of criteria for determination, targets that meet all the conditions for stationary vehicles are determined as stationary vehicles. [MODIFICATION OF THE THIRD DISCRIMINATION PROCESS]

[0191] A modification of the third discrimination process will now be described. In the modification of the third discrimination process, similar to the modifications of the first and second discrimination processes, a value based on the attenuation amount range A1 or the increase amount range A2 can be used as a magnification ratio C by which the stationary vehicle certainty is multiplied.

[0192] Specifically, for example, in the case where an attenuation amount range A1 is equal to or greater than the threshold value A1a, the magnification ratio C may be set to 0 because it is possible to determine that the target is an upper object. In this case, the certainty of being a stationary vehicle becomes 0. Therefore, it may be possible to prevent the target from being determined as a stationary vehicle.

[0193] Similarly, for example, in the case where the increase amount range A2 is smaller than the threshold A2a, since it is possible to determine that the target is an upper object, the magnification ratio C is set to 0, whereby the certainty of being a stationary vehicle becomes 0. It is therefore possible to prevent the target from being determined as a stationary vehicle.

[0194] Although the case where the attenuation amount range A1 or the increase amount range A2 is converted into a magnification ratio C as described above, the present invention is not limited thereto. The attenuation amount range A1 or the increase amount range A2 may be converted into any other value, for example, a value added to or subtracted from the certainty of a stationary vehicle.

[0195] In the discrimination process described above, the processes of calculating the attenuation amount range A1 and the increase amount range A2, and the processes of comparing the attenuation amount range A1 and the increase amount range A2 with the threshold values ​​A1a and A2a are also performed after the process of calculating the likelihood ratio Overall. However, the present invention is not limited to this. These processes, such as the processes of calculating the attenuation amount range A1 and the increase amount range A2, may be performed before the process of calculating the likelihood ratio Overall, or at the same time as the process of calculating the likelihood ratio Overall.

[0196] Furthermore, in the third discrimination process, both the attenuation amount range A1 and the increase amount range A2, which are the second and third indexes, are used to perform determination on a target. However, the present invention is not limited to this. For example, either of the second and third indexes may be used. (FOURTH DISTINGUISHING PROCESS OF DISTINGUISHING BETWEEN STATIONARY VEHICLES AND UPPER OBJECTS ACCORDING TO THE EMBODIMENT)

[0197] Now, a fourth discrimination process for discriminating between stationary vehicles and upper objects according to the embodiment will be described with reference to Fig. 23A to Fig. 24C described. Fig. 23A is a view illustrating the relationship between the angular power and distance of an upper object according to the embodiment. Fig.23B is a view illustrating the relationship between the angular power and distance of a stationary vehicle according to the embodiment. Fig. 24A to Fig. 24C Views for explaining a method of discriminating an upper object based on a distance at which the maximum value of the angular power was detected and a distance at which the minimum value of the angular power was detected, according to the embodiment.

[0198] In the third discrimination process described above, it is determined whether a target is a stationary vehicle or an upper object by setting a reference power for the angular power, and the amount of attenuation or the amount of increase from the reference power is compared with the threshold according to the target detection distance.

[0199] However, since the amount of attenuation or increase in the angular power from the reference power depends on the reference power setting method, if a specific value is set as the reference power, the amount of attenuation or increase may inadvertently have abnormal values. In the case where the radar device 1 is configured to have enhanced general capabilities, it may also become difficult to set an appropriate reference power.

[0200] For this reason, in the fourth discrimination process, whether a target is a stationary vehicle or an upper object is determined by a method different from the third discrimination process, using the angular power which has a tendency to increase as the detection distance fluctuates in the case where a target is a stationary target but has a tendency to weaken as the detection distance fluctuates as described above.

[0201] As described above, especially in the case where a target is an upper object, the upper object gradually disappears from the main transmission range of the transmitted wave. Therefore, in the radar device 1, the angular power, which is the received power of the reflected wave, tends to gradually decrease.

[0202] Meanwhile, in the case where the target is a stationary vehicle, the reflected wave penetrates the front surface of the RX receiving units more easily. The angular power, which is the received power of the reflected wave, therefore tends to gradually increase.

[0203] For this reason, as the vehicle moves toward a target, the radar device 1 determines the target's distance at regular intervals. Furthermore, the radar device 1 detects the angular powers of the reflected waves from the target received at the individual distances.

[0204] At this time, the radar device 1 sequentially calculates the moving averages of the subsequences of a predetermined number of (for example, three) consecutively detected angular powers and uses the smoothed averages as the angular powers of the waves reflected from the target at the individual distances. Thus, the influence of noise and multipath effects is removed from the angular power, and the radar device 1 can detect the original fluctuations in the angular power.

[0205] Subsequently, the radar device 1 detects a distance Rmax at which the maximum angular power was detected and a distance Rmin at which the minimum angular power was detected from the distances subsequently detected until the distance of the target becomes a predetermined distance (for example, 60 m from the radar device 1).

[0206] In the case where the distances and angular powers, as in Fig. 23A, for example, with respect to a certain upper object, the radar device 1 detects a distance corresponding to a black diamond shown in Fig. 23A, at the distance Rmax, and detects a distance corresponding to a white diamond shown in Fig. 23A, at a distance Rmin.

[0207] In the case where the distances and angular powers, as in Fig. 23A, with respect to a stationary vehicle, the radar device 1 also detects a distance corresponding to a black diamond shown in Fig. 23B, as the distance Rmax, and detects a distance corresponding to a white diamond shown in Fig. 23B, as the distance Rmin.

[0208] As described above, in the radar device 1, in the case where a target is an upper object, the distance Rmax is greater than the distance Rmin; while in the case where a target is a stationary vehicle, the distance Rmin is greater than the distance Rmax.

[0209] In other words, in the case where a target is an upper object, the radar device 1 detects a minimum angular power when the target is at a relatively short distance, and detects a maximum angular power when the target is at a relatively long distance. In contrast, in the case where a target is a stationary vehicle, the radar device 1 detects a maximum angular power when the target is at a relatively short distance, and detects a minimum angular power when the target is at a relatively long distance.

[0210] In the Fig. 23A and Fig.In the examples shown in Fig. 23B, 100 m is set as a threshold value for the distance from the radar device 1. In this case, with respect to a target, if the distance Rmin is shorter than the threshold value and the distance Rmax is greater than the threshold value, it is possible to determine that the target is an upper object, and if the distance Rmin is greater than the threshold value and the distance Rmax is shorter than the threshold value, it is possible to determine that the target is a stationary vehicle.

[0211] However, in the case where the same threshold value is set for distances Rmax and Rmin, the radar device 1 may erroneously determine a stationary object as an upper object. For example, in the case where a target is a large-sized vehicle such as a truck or a bus, even if the vehicle is at a relatively large distance from the radar device 1, the angular power may increase to such an extent that the distance Rmax becomes longer than the threshold value. In this case, the radar device 1 may erroneously determine the truck or bus, which is a stationary vehicle, as an upper object.

[0212] For this reason, in the fourth discrimination process, a threshold RmaxTh for minimum angular powers and a threshold RminTh for maximum angular powers are provided separately, and each upper object is discriminated based on the result of the comparison between the distance Rmax and the threshold RmaxTh and the result of the comparison between the distance Rmin and the threshold RminTh.

[0213] Due to the tendency of the angular power described above, the distances Rmax of upper objects also tend to be longer than the distance Rmin ( Fig. 23A). For this reason, in the fourth discrimination process, each upper object is discriminated based on the result of the comparison between a difference (Rmax - Rmin) obtained by subtracting the distance Rmin from the distance Rmax and a difference threshold Rmax-RminTh.

[0214] In the fourth discrimination process, for example, a threshold RmaxTh, a threshold RminTh and a threshold Rmax-RminTh, as shown in Fig. 24A, Fig. 24B and Fig. 24C shown. Fig. 24A, Fig. 24B and Fig. 24C show verification experiment results obtained by detecting a variety of targets.

[0215] In Fig. 24A, white diamonds represent the distances Rmax and angular powers of upper objects, black squares represent distances Rmax and angular powers of small-sized stationary vehicles, and black triangles represent the distances Rmax and angular powers of large-sized stationary vehicles.

[0216] In Fig.24B, white diamonds represent the distances Rmin and angular powers of the upper objects, black squares represent the distances Rmin and angular powers of the small-sized stationary vehicles, and black triangles represent the distances Rmin and angular powers of the large-sized stationary vehicles. Fig. 24C also shows white diamonds representing the distance differences (Rmax-Rmin) of the upper objects.

[0217] In the fourth process of discrimination, as in Fig. 24A, in the case where the radar device 1 derives distances to each target in a detection range between 150 m and 60 m, a threshold value RmaxTh shorter than a distance that is the center of the detection range (for example, about 100 m) is set.

[0218] As a result, in the fourth discrimination process, targets with distances Rmax shorter than the threshold RmaxTh are excluded from targets that could be upper objects. Thus, in the fourth discrimination process, it is possible to exclude targets that have a high probability of being stationary vehicles with a distance Rmax shorter than the distance that is the center of the detection area from targets that could be upper objects.

[0219] In the fourth process of discrimination, as in Fig. 24B, in addition, in the case where the radar device 1 derives distances to each target in a detection range between 150 m and 60 m, a threshold value RminTh greater than a distance that is the center of the detection range (for example, about 100 m) is set.

[0220] As a result, in the fourth discrimination process, targets with distances Rmin greater than the threshold RminTh are excluded from targets that could be upper objects. Thus, in the fourth discrimination process, it is possible to exclude targets that have a high probability of being stationary vehicles with a distance Rmax shorter than the distance that is the center of the detection area from targets that could be upper objects.

[0221] In the fourth process of discrimination, as in Fig.24C, a positive value close to 0 (for example, a value less than or equal to 10) is also set as a threshold Rmax-RminTh. As a result, in the fourth discrimination process, targets with differences (Rmax-Rmin) smaller than the threshold are excluded from targets that might be upper objects. In this way, in the fourth discrimination process, it is possible to exclude targets that have a high probability of being stationary vehicles with a distance Rmax longer than the distance Rmin from targets that might be upper objects.

[0222] The reason for setting a positive value close to 0 as a threshold Rmax-RminTh is that, for example, when the reflected wave from a large-sized stationary vehicle is affected by noise or multipath effects, although the corresponding vehicle is a stationary vehicle, the distance Rmax slightly exceeds the distance Rmin.

[0223] Even in such cases, in the fourth discrimination process, since a positive value close to 0 was set as a threshold Rmax-RminTh, it is possible to exclude large-sized stationary vehicles from targets that might be upper objects based on the reflected waves affected by noise or multipath effects.

[0224] Therefore, in the fourth discrimination process, the targets that were not excluded from targets that could be upper objects by the three discrimination processes based on the threshold RmaxTh, the threshold RminTh, and the threshold Rmax-RminTh are determined as upper objects.

[0225] In other words, in the fourth discrimination process, targets with distances Rmax greater than or equal to the threshold RmaxTh, distance Rmin less than or equal to the threshold RminTh, and differences (Rmax-Rmin) greater than or equal to the threshold Rmax-RminTh are determined as upper objects. Therefore, the fourth discrimination process can discriminate upper objects more accurately.

[0226] For example, in the fourth discrimination process, when only the determination process based on the threshold value RmaxTh is performed, some large-sized stationary vehicles with distances Rmax close to 100 m (black triangles near 100 m, which are shown in Fig. 24A) may not be excluded from targets that may be upper objects.

[0227] Even in such a case where the determination process based on the threshold value RmaxTh could not exclude some large-sized stationary vehicles from targets that might be upper objects, in the fourth discrimination process, by performing the determination process based on the threshold value RminTh thereafter, it is possible to reduce the number of large-sized stationary vehicles.

[0228] Furthermore, in the fourth discrimination process, by performing the determination process based on the threshold value Rmax-RminTh as shown in Fig. As shown in Figure 24C, it is possible to definitively exclude large-scale stationary vehicles from targets that may be upper objects. It is therefore possible to accurately distinguish between stationary vehicles and upper objects. (Target information output process according to embodiment)

[0229] Fig. 25A is a flowchart illustrating a target information output process according to the embodiment. The signal processing unit 6 performs a target information extraction process at regular intervals (e.g., every five seconds). The target information extraction process starts when beat signals obtained from reflected waves RW are input to the signal processing unit 6 from the four receiving units RX.

[0230] First, the Fourier transform unit 62 of the signal processing unit 6 performs fast Fourier transform on the beat signals received by the plurality of individual receiving units 5 (STEP S101). Subsequently, the peak extraction unit 70 extracts from the peaks generated by the Fourier transform unit 62 that exceed a predetermined signal level, respectively, an up-section in which the frequency of the transmission signal increases and a down-section in which the frequency decreases (STEP S102).

[0231] Next, the angle estimation unit 71 separates, with respect to each up-section and down-section, information objects on a plurality of targets existing in the same bin from each peak frequency signal through an azimuth calculation process and estimates the angles of the plurality of targets (STEP S103).

[0232] Next, the pairing unit 72 associates each of a plurality of targets T existing in front of the vehicle A with each other, thereby deriving pair data items related to the targets T (STEP S104). Next, the continuity determination unit 73 determines whether a target data item derived in the past process and a target data object derived in the last process relate to the same target (STEP S105).

[0233] Next, the filtering unit 74 smooths the parameters (the longitudinal distances, the lateral distances, and the relative speeds) of each pair of target data items derived in the past process and the last process in the time axis direction, thereby deriving a target data item (internal filter data item) (step S106). Next, the target classification unit 75 classifies the individual targets into preceding vehicles, stationary objects (including stationary vehicles), and approaching vehicles based on their relative speeds (step S107).

[0234] Next, the unnecessary target removal unit 76 determines unnecessary targets, such as upper objects, lower objects, rain, etc., among the individual targets and excludes the unnecessary targets from output targets (STEP S108). The process of STEP S108 includes an upper object exclusion process, which is described below with reference to Fig. 25B is described.

[0235] Next, the grouping unit 77 copies a plurality of target data objects on the same target into one target data object (STEP S109). Next, the target information output unit 78 selects a predetermined number of target data objects as output objects from the plurality of target data objects derived or estimated by extrapolation and outputs the selected target data objects to the vehicle control device 2 (STEP S110). When STEP S110 ends, the signal processing unit 6 terminates the target information derivation process. (REMOVING UNNECESSARY TARGETS ACCORDING TO THE EMBODIMENT)

[0236] Fig. 25B is a flowchart illustrating a subroutine for removing unnecessary targets according to one embodiment. Fig.25B shows the process flow of removing upper objects according to the embodiment, which is a part of the process of removing unnecessary targets in STEP S108 shown in Fig. 25A is shown.

[0237] First, the unnecessary target removal unit 76 calculates the above-described likelihood ratios 1 to 6 and the likelihood ratio OverAll in the first discrimination process. Now, with reference to Fig. 25C describes an example of the process of calculating the likelihood ratios 1 to 6 and an overall likelihood ratio. Fig. 25C is a flowchart illustrating a subroutine for calculating likelihood ratios 1-6 and a (total) likelihood ratio OverAll according to the embodiment.

[0238] As in Fig.As shown in Figure 25C, the unnecessary target removal unit 76 first calculates the lateral position difference of a target based on Formula 2 (STEP S301). Next, the unnecessary target removal unit 76 calculates a likelihood ratio of 1 based on the lateral position difference calculated in STEP S301 and Formula 1 (STEP S302).

[0239] Next, the unnecessary target removal unit 76 calculates the relative speed difference of the target based on Formula 3 (STEP S303). Next, the unnecessary target removal unit 76 calculates a likelihood ratio 2 based on the relative speed difference calculated in STEP S303 and Formula 1 (STEP S304).

[0240] Next, the unnecessary target removal unit 76 calculates the beam-independent inclination difference based on Formula 4 (STEP S305). Next, the unnecessary target removal unit 76 calculates a likelihood ratio 3 based on the beam-independent inclination difference derived in STEP S305 and Formula 1 (STEP S306).

[0241] Next, the unnecessary target removal unit 76 calculates the up / down beam power difference inclination based on Formula 5 (STEP S307). Next, the unnecessary target removal unit 76 calculates a likelihood ratio of 4 based on the up / down beam power difference inclination based on the up / down beam power difference inclination calculated in STEP S307 and Formula 1 (STEP S308).

[0242] Next, the unnecessary target removal unit 76 calculates the up / down beam range differential based on Formula 6 (STEP S309). Next, the unnecessary target removal unit 76 calculates a likelihood ratio 5 based on the up / down beam range differential derived in STEP S309 and Formula 1 (STEP S310).

[0243] Next, the unnecessary target removal unit 76 determines whether extrapolation has been performed, and if extrapolation has been performed, the unnecessary target removal unit identifies the cause of the extrapolation (STEP S311). Next, the unnecessary target removal unit 76 calculates a likelihood ratio 6 based on the result obtained in STEP S311 by determining whether extrapolation has been performed, the cause of the extrapolation identified in STEP S311, and Formula 1 (STEP S312).

[0244] Next, the unnecessary target removal unit 76 calculates a probability ratio OverAll based on Formula 7 (STEP S313). After that, the unnecessary target removal unit 76 performs the process in Fig. 25B of STEP S202.

[0245] Again with reference to Fig.25, the removal of unnecessary targets is further described. After calculating the likelihood ratios 1 to 6 and the likelihood ratio Overall, the unnecessary target removal unit 76 calculates an attenuation amount range A1 and an increase amount range A2 (STEP S202).

[0246] Next, the unnecessary target removal unit 76 detects the distance Rmax at which the maximum angular power was detected and the distance Rmin at which the minimum angular power was detected (STEP S203), and calculates the difference (Rmax - Rmin) by subtracting the distance Rmin from the distance Rmax (STEP S204).

[0247] Next, the unnecessary target removal unit 76 determines whether the distance Rmax is greater than or equal to the threshold RmaxTh (STEP S205). If it is determined that the distance Rmax is shorter than the threshold RmaxTh ("No" in STEP S205), the unnecessary target removal unit 76 proceeds to STEP S208.

[0248] Meanwhile, if it is determined that the distance Rmax is greater than or equal to the threshold RmaxTh ("Yes" in STEP S205), the unnecessary target removal unit 76 determines whether the distance Rmin is less than or equal to the threshold RminTh (STEP S206). If it is determined that the distance Rmin is greater than the threshold RminTh ("No" in STEP S206), the unnecessary target removal unit 76 proceeds to STEP S208.

[0249] Meanwhile, in the case where the distance Rmin is less than or equal to the threshold RminTh (“Yes” in STEP S206), the unnecessary target distance unit 76 determines whether the difference (Rmax-Rmin) is greater than or equal to the threshold Rmax-RminTh (STEP S207).

[0250] In the case where the difference (Rmax-Rmin) is smaller than the threshold value Rmax-RminTh ("No" in STEP S207), the unnecessary target removal unit 76 proceeds to STEP S208. Meanwhile, in the case where the difference (Rmax-Rmin) is greater than or equal to the threshold value Rmax-RminTh ("Yes" in STEP S207), the unnecessary target removal unit 76 determines that the target is an upper object (STEP S211).

[0251] In STEP S208, the unnecessary target removal unit 76 also determines whether the attenuation amount range A1 calculated in STEP S202 is greater than or equal to the threshold A1a. If it is determined that the attenuation amount range A1 is greater than or equal to the threshold A1a ("Yes" in STEP S208), the unnecessary target removal unit 76 determines that the target is an upper object (STEP S211).

[0252] Meanwhile, if the attenuation amount range A1 is smaller than the threshold value A1a ("No" in STEP S208), the unnecessary target removal unit 76 determines whether the increase amount range A2 is smaller than the threshold value A2a (STEP S209). If it is determined that the increase amount range A2 is smaller than the threshold value A2a ("Yes" in STEP S209), the unnecessary target removal unit 76 determines that the target is an upper object (STEP S211).

[0253] Meanwhile, in the case where the increase amount range A2 is not smaller than the threshold A2a (“No” in STEP S209), the unnecessary target removal unit 76 determines whether the likelihood ratio OverAll is greater than or equal to the threshold (STEP S210).

[0254] If it is determined that the likelihood ratio OverAll is greater than or equal to the threshold value ("Yes" in STEP S210), the unnecessary target removal unit 76 determines that the target is a stationary vehicle (STEP S212). Meanwhile, if it is determined that the likelihood ratio OverAll is less than the threshold value ("No" in STEP S210), the unnecessary target removal unit 76 determines that the target is an upper object (STEP S211).

[0255] As described above, in the fourth discrimination process, a target is determined to be an upper object when a target satisfies all three conditions, namely, "[Distance Rmax] ≥ [Threshold RmaxTh]," "[Distance Rmin] ≤ [Threshold RminTh]," and "[Difference (Rmax - Rmin)] ≥ [Threshold Rmax-RminTh]." Therefore, in the fourth discrimination process, it is possible to accurately discriminate upper objects among the plurality of targets detected by the radar device 1.

[0256] In the present embodiment, when a target does not satisfy any of the conditions “[Distance Rmax] ≥ [Threshold RmaxTh]”, “[Distance Rmin] ≤ [Threshold RminTh]”, and “[Difference (Rmax - Rmin)] ≥ [Threshold Rmax-RminTh]”, the first, second, and third discrimination processes are performed on the corresponding target.

[0257] Therefore, even if some upper objects are excluded from targets that might be upper objects by the fourth discrimination process through the above-described determination process from STEP S208 to STEP S210, it is possible to accurately determine whether each target is an upper object or a stationary vehicle.

[0258] Herein, the case of determining each target that satisfies all three of the conditions “[Distance Rmax] ≥ [Threshold RmaxTh]”, “[Distance Rmin] ≤ [Threshold RminTh]”, and “[Difference (Rmax - Rmin)] ≥ [Threshold Rmax-RminTh]” has been described as an upper object; however, this is an example.

[0259] In the fourth discrimination process, any target that satisfies at least one of the conditions of being an upper object: "[Distance Rmax] ≥ [Threshold RmaxTh]," "[Distance Rmin] ≤ [Threshold RminTh]," and "[Difference (Rmax - Rmin)] ≥ [Threshold Rmax-RminTh]" can be determined as an upper object. Even in this case, it is possible to discriminate upper objects with a certain degree of accuracy.

[0260] The peak extraction unit 79, the angle estimation unit 71, and the pairing unit 72 are examples of an inference unit. The unnecessary target removal unit 76 is also an example of a detection unit, a calculation unit, and a determination unit. In addition, stationary vehicles are examples of a target that will collide with the vehicle (for example, a target that requires control on the vehicle, such as brake control), and upper objects are examples of a target that will not collide with the vehicle (for example, a target that does not require control on the vehicle, such as brake control).

[0261] Among the processes described in the embodiments, all or some of the processes referred to as being performed automatically may also be performed manually. Otherwise, among the processes described in the embodiments, all or some of the processes referred to as being performed manually may also be performed automatically using widely known methods.

[0262] The integration and distribution of the components described in the embodiments may also be appropriately changed depending on the processing load and processing efficiency. Furthermore, the processing methods, control methods, specific designations, data variety, and information variety, including parameters, may be appropriately changed, except where explicitly mentioned.

[0263] Various advantages and modifications will readily occur to those skilled in the art. The invention, in its broader aspects, is therefore not limited to the specific details and representative embodiments shown and described above. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents.

Claims

[1] Radar device (1) comprising: a deriving unit (71, 72, 79) configured to derive detection distances of a target (T1, T2) existing in an area near a vehicle (A) equipped with the radar device (1) based on reception signals acquired by transmitting a radar transmission wave to the area near the vehicle (A) and receiving a reflected wave from the target (T1, T2); a detection unit (76) configured to detect a distance at which a maximum value of reception power was detected and a distance at which a minimum value of reception power was detected among the detection distances sequentially derived by the derivation unit (71, 72, 79); and a determination unit (76) configured to determine whether the target (T1, T2) is an upper object that will not collide with the vehicle (A) based on the distances detected by the detection unit (76), wherein, in a case where the difference between the distance at which the maximum value of the received powers was detected and the distance at which the minimum value of the received powers was detected is equal to or greater than a third threshold value for the difference, the determination unit (76) determines that the target (T1, T2) is the upper object. [2] The radar device (1) according to claim 1, wherein, in a case where the distance at which a maximum value of the reception powers was detected is equal to or longer than a first threshold value for the distance, the determination unit (76) determines that the target (T1, T2) is the upper object. [3] The radar device (1) according to claim 1 or 2, wherein, in a case where the distance at which a minimum value of the reception powers was detected is equal to or shorter than a second threshold value for the distance, the determination unit (76) determines that the target (T1, T2) is the upper object. [4] The radar device (1) according to claim 1, wherein, in a case where the distance at which the maximum value of the reception powers was detected is equal to or longer than a first threshold value for the distance, the distance at which the minimum value of the reception powers was detected is equal to or shorter than a second threshold value for the distance, and the difference between the distance at which the maximum value of the reception powers was detected and the distance at which the minimum value of the reception powers was detected is equal to or greater than a third threshold value for the difference, the determination unit (76) determines that the target (T1, T2) is the upper object. [5] Radar device (1) according to claim 1 or 4, wherein the third threshold value for the difference is set to a positive value close to 0. [6] Radar device (1) according to one of claims 1 to 5, wherein the detection unit (76) detects a distance at which a maximum value of the moving averages of the reception powers was detected and a distance at which a minimum value of the moving averages was detected. [7] Radar device (1) according to one of claims 1 to 6, further comprising a calculation unit (76) configured to set a reference value based on a reception power of a reflected wave received at a target detection distance equal to or longer than a predetermined distance, and configured to calculate an increase amount of the reception power of the reflected wave by accumulating a product, the product being obtained by calculating a difference by subtracting the reference value from the reception power of the reflected wave detected in a current process and multiplying the difference by a detection distance difference between a detection distance in a previous process and a detection distance in the current process, wherein, in a case where the increase amount of the reception power of the reflected wave calculated by the calculation unit is smaller than a threshold value for an increase amount, the determination unit (76) determines that the target (T1, T2) is the upper object. [8] Radar device (1) according to one of claims 1 to 7, wherein the derivation unit (71, 72, 79) derives a plurality of parameters relating to the target (T1, T2) based on received signals at regular intervals, and the determination unit (76) determines whether the target (T1, T2) is the upper target (T1, T2) based on the plurality of parameters and the detection distances derived by the derivation unit (71, 72, 79) and based on the probability ratios obtained by calculations based on probability models representing known correlations between a probability of being a target (T1, T2) that exists in a moving direction of the vehicle (A) and will collide with the vehicle (A) when the vehicle (A) is running in the moving direction, and a probability of being a target (T1, T2) that exists in the moving direction of the vehicle (A) but will not collide with the vehicle (A) even though the vehicle (A) is running in the moving direction for each of the detection distances. [9] A radar device control method which a control device of a radar device (1) performs, the method comprising: Deriving detection distances of a target (T1, T2) existing in an area near a vehicle (A) equipped with the radar device based on a reception signal acquired by transmitting a radar transmission wave to an area close to the vehicle (A) and receiving a reflected wave from the target (T1, T2); detecting a distance at which a maximum value of the received power was detected and a distance at which a minimum value of the received power was detected among the detection distances derived sequentially; and Determining whether the target (T1, T2) is an upper object that will not collide with the vehicle (A) based on the detected distances, wherein, in a case where the difference between the distance at which the maximum value of the received power was detected and the distance at which the minimum value of the received power was detected is equal to or greater than a third threshold value for the difference, it is determined that the target (T1, T2) is the upper object.

Citation Information

Patent Citations

  • Radar system, vehicle control system, and signal processing method

    JP2016006383A

  • METHOD AND DEVICE FOR CLASSIFICATION OF HIGH LEVEL OBJECTS

    DE69829777T2

  • Method for detecting stationary object on road by radar

    EP1371997A1