JP2018063130A - Radar device and continuity determination method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a radar device and a continuity determination method capable of accurately determining the temporal continuity of a target.SOLUTION: The radar device includes a deriving unit, an estimating unit, a setting unit, and a determining unit. The deriving unit derives observation data regarding a target including one or more of distance, relative speed, and angle, to the target based on reflected wave from the target. The estimating unit generates estimated data from the observation data. The setting unit sets a prediction range based on the estimated data. When the current observation data derived by the deriving unit is within the prediction range, the determining unit determines that the preceding observation data and the current observation data derived by the deriving unit are temporally continuous observation of the same target data. The setting unit includes a determining unit that determines the size of the prediction range based on an error of the observation data previously derived by the deriving unit.SELECTED DRAWING: Figure 1B

Description

  Embodiments disclosed herein relate to a radar apparatus and a continuity determination method.

  Conventionally, a radar device that transmits a transmission wave and detects a target based on a reflected wave from the target is known. Such a radar apparatus repeatedly performs, every predetermined time, processing for deriving observation data including distance and relative velocity data from the target based on the reflected wave from the target, and the target corresponding to the observation data derived this time. It is determined whether or not the same target as the target corresponding to the previously derived observation data exists.

  Specifically, the radar apparatus calculates prediction data obtained by predicting the observation data derived this time, and if the observation data derived this time is included within the prediction range based on the prediction data, the radar data was derived last time. The observation data and the observation data derived this time are determined to be the same target having temporal continuity (see, for example, Patent Document 1).

JP 2006-128751 A

  In the radar apparatus described above, it is desired to improve the accuracy of determining the temporal continuity of a target.

  One aspect of the embodiments has been made in view of the above, and an object thereof is to provide a radar apparatus and a continuity determination method that can accurately determine temporal continuity of a target.

  A radar apparatus according to an aspect of an embodiment includes a derivation unit, a prediction unit, a setting unit, and a determination unit. The deriving unit derives observation data of the target including at least one of distance, relative speed, and angle data with respect to the target based on a reflected wave from the target. The prediction unit generates prediction data obtained by predicting the observation data derived by the deriving unit. The setting unit sets a prediction range based on the prediction data. When the observation data derived this time by the derivation unit is within the prediction range, the determination unit is continuous in time with the observation data previously derived by the derivation unit and the observation data derived this time. It is determined that the observation data is the same target. The setting unit includes a determining unit that determines the size of the prediction range based on an error in observation data derived in the past by the deriving unit.

  According to the radar device and the continuity determination method according to one aspect of the embodiment, the temporal continuity of a target can be accurately determined.

FIG. 1A is a diagram illustrating a positional relationship between a radar apparatus mounted on a vehicle and a target. FIG. 1B is a diagram illustrating an example of changes in observation data, prediction data, and a prediction range. FIG. 2 is a diagram illustrating a configuration of the radar apparatus according to the embodiment. FIG. 3 is a diagram illustrating an example of a confirmed data table including past confirmed data stored in the storage unit illustrated in FIG. 2. FIG. 4 is a diagram illustrating an example of the configuration of the setting unit illustrated in FIG. FIG. 5 is a diagram illustrating an example of error variation widths of distance observation data, speed observation data, and angle observation data. FIG. 6 is a diagram illustrating an example of a relationship between a distance prediction range, distance prediction data, distance observation data, and distance determination data. FIG. 7 is a flowchart illustrating an example of a processing procedure executed by the data processing unit.

  Hereinafter, embodiments of a radar apparatus and a continuity determination method disclosed in the present application will be described in detail with reference to the accompanying drawings. In addition, this invention is not limited by embodiment shown below.

[1. Target detection method using radar device]
FIG. 1A is a diagram illustrating an example of a positional relationship between a radar device mounted on a vehicle and a target. As shown in FIG. 1A, a radar apparatus 1 according to the embodiment is mounted on a vehicle 5 such as an automobile and detects a target in front.

  The radar apparatus 1 transmits a transmission wave and derives observation data including data on the distance, relative speed, and angle between the radar apparatus 1 (vehicle 5) and the target based on the reflected wave of the transmission wave from the target. Is repeated every predetermined time. In the following description, the observation data will be described as including distance, relative speed, and angle data. However, the observation data may be at least one of distance, relative speed, and angle data. Data other than the above data may be included. In the following, distance data among observation data is described as distance observation data, relative speed data among observation data is described as speed observation data, and angle data among observation data is described as angle observation data. There is a case.

  The distance to the target (hereinafter sometimes simply referred to as a distance) is the distance until the reflected wave reflected from the target reaches the radar apparatus 1, and the distance observation data is the distance from the target. It is data indicating an instantaneous value. The relative speed with the target (hereinafter sometimes simply referred to as speed) is the relative speed between the target and the radar apparatus 1, and the speed observation data is an instantaneous value of the relative speed with the target. It is the data shown. The angle with the target (hereinafter sometimes simply referred to as an angle) is an angle at which the reflected wave reflected from the target reaches the radar apparatus 1, and the angle observation data is an angle of the target and the target. It is data indicating an instantaneous value.

  In the example illustrated in FIG. 1A, the other vehicle 6 exists within the detection range L of the radar device 1, and the rear end portion of the other vehicle 6 is detected by the radar device 1 as a target.

  The radar apparatus 1 determines whether observation data of the same target as the previously derived observation data exists in the observation data derived this time. That is, the radar apparatus 1 determines whether or not there is temporal continuity between the observation data derived last time and the observation data derived this time. Specifically, the radar apparatus 1 generates prediction data obtained by predicting the observation data derived this time. The predicted data includes predicted distance, angle, and speed data, and the radar apparatus 1 determines the predicted ranges of distance, angle, and speed based on the predicted data of distance, angle, and speed. Set.

  The distance prediction data (hereinafter may be described as distance prediction data) is data indicating the instantaneous value of the distance to the predicted target, and is the speed prediction data (hereinafter referred to as speed prediction data). Is data indicating an instantaneous value of the relative speed with respect to the predicted target. Further, angle prediction data (hereinafter sometimes referred to as angle prediction data) is data indicating an instantaneous value of an angle with a predicted target.

  When the radar apparatus 1 is the same target having temporal continuity with the observation data derived in the previous time, the observation data included in the prediction range for each of the distance, the angle, and the velocity among the observation data derived this time. Perform continuity determination.

  When there are a plurality of observation data included in the prediction range for each of the distance, angle, and speed among the observation data derived this time, the radar apparatus 1 has, for example, the distance, angle, and speed specified by the observation data as the prediction data. The observation data having the closest distance, angle, and velocity specified in (1) is determined to be the same target having temporal continuity with the observation data derived last time.

  For example, the radar apparatus 1 weights the difference between the distance observation data and the distance prediction data, the difference between the speed observation data and the speed prediction data, and the difference between the angle observation data and the angle prediction data. It is possible to determine that the observation data with the smallest result of adding the differences is the same target having temporal continuity with the observation data derived last time.

  Further, the radar apparatus 1 calculates the Mahalanobis distance for the prediction data in a multi-dimensional space of distance, velocity, and angle, and the observation data having the closest Mahalanobis distance is the same as the observation data derived in the previous time and having the temporal continuity. It can be determined that it is a target.

  Here, when the error Δ of the observation data is large, the continuity determination cannot be performed accurately, and the followability to the target may be deteriorated. Therefore, the radar apparatus 1 according to the embodiment changes the prediction range based on the difference between the past prediction data and the past observation data. That is, the radar apparatus 1 determines the difference between past distance observation data and past distance prediction data, the difference between past speed observation data and past speed prediction data, and past angle observation data and past angle prediction data. Based on the difference between the prediction range and the prediction range.

  FIG. 1B is a diagram illustrating an example of changes in observation data, prediction data, and a prediction range. In the example shown in FIG. 1B, for easy understanding, the three-dimensional observation data, prediction data, and prediction range represented by three parameters including distance, angle, and velocity data are converted into one-dimensional for convenience. It is expressed by observation data, prediction data, and prediction range.

As illustrated in FIG. 1B, at time t _n−2 , the radar apparatus 1 determines that the observation data is within the prediction range, and thus determines that the target is a temporally continuous target from time t _n−3. based on _n-2 of the observation data to determine the deterministic data at time _{t n-2.} For example, the radar device 1, the distance, angle and for each speed parameter, a time t _n-2 of the observed data and the time t _n-2 of the prediction data and the sum are weighted respectively, the time t _n of such addition result _-2 as finalized data.

Next, the radar device 1, based on the determined data at time _{t n-2,} to calculate the predicted data at time _{t n-1,} the time _{t n-1} of the predicted data from the predetermined range of times _{t n-1} Set as prediction range. In the example shown in FIG. 1B, since the time _{t n-1} of the observational data falls within the expected range of time _{t n-1,} the radar device 1, the time _{t n-1} of the observed data is time from time _{t n-2} to determine that the observation data of the successive target object, as in the case of the time t _n-2, on the basis of the time t _n-1 of the observed data and the time t _n-1 of the predicted data at time t Determine _n-1 deterministic data.

Similarly, the radar apparatus 1, based on the time t _n-1 of the determined data, to calculate the predicted data at time t _n, setting a predetermined range of the predicted data at time t _n as a prediction range of times t _n. Since the observed data at time t _n is within the expected range of time t _n, the radar device 1, the observation data at time t _n is determined that the target object temporally continuous from the time t _n-1, the time based on the prediction data of the observed data and the time t _n of t _n, as in the case of the time t _n-2, to determine the determined data at time t _n.

  Here, a method for setting the prediction range will be described. The radar apparatus 1 determines the size of the prediction range based on the error Δ of a plurality of past observation data. For example, the radar apparatus 1 can determine the size of the prediction range based on the average value or variation of the error Δ of a plurality of past observation data. Specifically, the radar apparatus 1 sets the difference between the past observation data and the corresponding prediction data (or confirmed data) as the error Δ of the observation data, and describes the average value of the error Δ (hereinafter referred to as an error average value Δav). Is set to a predetermined value or more, the size of the prediction range is set small.

  Further, when the past error average value Δav is equal to or greater than a predetermined value, the radar apparatus 1 multiplies the error average value Δav by a larger coefficient as the variation width Δw of the error Δ of the observation data increases, and the multiplication result is large. As a result, the size of the prediction range can be set smaller. The variation width Δw is a temporal variation width of the observation data error Δ, for example, the maximum value (positive maximum value) and minimum value (negative maximum value) of the observation data error Δ within a predetermined past time. ). The fluctuation of the error Δ of the observation data is caused by, for example, a detection error of the radar apparatus 1 or a fluctuation of an object reflection point with respect to a transmission wave of the radar apparatus 1.

  The radar apparatus 1 can set the size of the prediction range to be smaller as the error average value Δav is larger, or can set the size of the prediction range to be smaller as the variation width Δw of the error Δ is larger. . In addition, the radar apparatus 1 can make the reduction rate (change rate) of the prediction range the same for the distance, angle, and speed from the target, and the reduction rate of the prediction range by the distance, angle, and speed, respectively. Can be set individually. Hereinafter, a configuration example of the radar apparatus 1 according to the embodiment will be specifically described.

[2. Configuration example of radar apparatus 1]
FIG. 2 is a diagram illustrating a configuration of the radar apparatus 1 according to the embodiment. The radar device 1 according to the embodiment uses, for example, FM-CW (Frequency Modulated-Continuous Wave), which is a frequency-modulated continuous wave, among various types of millimeter wave radars, to detect targets existing around the vehicle 5. To detect.

  The radar apparatus 1 includes a transmission unit 10, a reception unit 20, and a signal processing unit 30, and is connected to a vehicle control device 2 that controls the behavior of the vehicle 5. The vehicle control device 2 performs vehicle control such as PCS (Pre-crash Safety System) and AEB (Advanced Emergency Braking System) based on the detection result of the target by the radar device 1. The radar device 1 may be used for various purposes other than the on-vehicle radar device (for example, monitoring of an aircraft or a ship).

[2.1. Transmitter 10]
The transmission unit 10 includes a signal generation unit 11, an oscillator 12, and a transmission antenna 13. The signal generation unit 11 generates a modulation signal whose voltage changes in a triangular wave shape and supplies the modulation signal to the oscillator 12. The oscillator 12 generates a transmission signal ST whose frequency changes over time based on the modulation signal generated by the signal generation unit 11 and outputs the transmission signal ST to the transmission antenna 13.

  The transmission antenna 13 converts the transmission signal ST from the oscillator 12 into a transmission wave TW and outputs the transmission wave TW to the outside of the vehicle 5. The transmission wave TW output from the transmission antenna 13 is FM-CW whose frequency increases and decreases at a predetermined period. The transmission wave TW transmitted to the front of the vehicle 5 from the transmission antenna 13 is reflected by a target such as another vehicle to become a reflected wave RW.

[2.2. Receiver 20]
The receiving unit 20 includes a plurality of receiving antennas 21 that form an array antenna and a plurality of individual receiving units 22 connected to the plurality of receiving antennas 21. Each receiving antenna 21 receives the reflected wave RW from the target and converts the reflected wave RW into a received signal SR. Each individual receiving unit 22 processes the received signal SR obtained by the corresponding receiving antenna 21. The number of receiving antennas 21 and individual receiving units 22 shown in FIG. 2 is four, but may be three or less or five or more.

  Each individual receiving unit 22 includes an amplifier 23, a mixer 24, and an A / D (Analog / Digital) converter 25. A reception signal SR obtained from the reflected wave RW received by the reception antenna 21 is amplified by an amplifier 23 (for example, a low noise amplifier) and then sent to the mixer 24. The mixer 24 mixes a part of the transmission signal ST and the reception signal SR to remove unnecessary signal components and generates a beat signal SB.

  Thereby, a beat signal SB indicating a beat frequency that is a difference between the frequency of the transmission signal ST and the frequency of the reception signal SR is generated. The beat signal SB generated by the mixer 24 is converted into a digital signal by the A / D converter 25 and then output to the signal processing unit 30.

[2.3. Signal processor 30]
The signal processing unit 30 includes a transmission control unit 31, a Fourier transform unit 32, and a data processing unit 33. The signal processing unit 30 is a microcomputer including, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an input / output port, and the like, and controls the entire radar apparatus 1.

  The microcomputer CPU functions as the transmission control unit 31, the Fourier transform unit 32, and the data processing unit 33 by reading and executing the program stored in the ROM. Note that all of the transmission control unit 31, the Fourier transform unit 32, and the data processing unit 33 can be configured by hardware such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

[2.3.1. Transmission control unit 31]
The transmission control unit 31 controls the signal generation unit 11 of the transmission unit 10 and causes the signal generation unit 11 to output a modulation signal whose voltage changes in a triangular wave shape to the oscillator 12. As a result, the transmission signal ST whose frequency changes with the passage of time is output from the oscillator 12 to the transmission antenna 13.

[2.3.2. Fourier transform unit 32]
The Fourier transform unit 32 performs Fast Fourier Transform (FFT) processing on the beat signal SB output from each of the plurality of individual reception units 22. Accordingly, the Fourier transform unit 32 converts the beat signal SB output from each of the plurality of individual reception units 22 into frequency spectrum data.

  The frequency spectrum converted by the Fourier transform unit 32 is output to the data processing unit 33. The frequency spectrum data includes signal level information for each frequency bin (BIN) set at a frequency interval corresponding to the frequency resolution of the Fourier transform unit 32.

[2.3.3. Data processing unit 33]
The data processing unit 33 includes an observation data deriving unit 34, a continuity determining unit 35, a filter unit 36, and an output unit 37. The data processing unit 33 derives observation data of the distance, speed, and angle with the target based on the frequency spectrum data converted by the Fourier transform unit 32, and performs vehicle control on target information corresponding to the observation data. Output to device 2. The vehicle control device 2 controls the behavior of the vehicle 5 based on the target information acquired from the radar device 1.

[2.3.4. Observation data deriving unit 34]
The observation data deriving unit 34 performs an observation data deriving process for deriving observation data including data on the distance, velocity, and angle with the target based on the frequency spectrum data converted by the Fourier transform unit 32. Such observation data derivation processing is repeatedly executed by the observation data deriving unit 34 every predetermined period Ts (for example, 1/20 second), and the derived observation data is determined from the observation data deriving unit 34 for each predetermined period Ts. Is output to the unit 35.

  The observation data deriving unit 34 includes a peak extracting unit 41, an angle estimating unit 42, and a pairing unit 43. The peak extraction unit 41 includes peaks that exceed a predetermined signal level in the frequency spectrum converted by the Fourier transform unit 32, and an up period in which the frequency of the transmission signal ST increases and a down period in which the frequency of the transmission signal ST decreases. Extract in each section. Hereinafter, the frequency extracted in this way is referred to as “peak frequency”.

  Since the reflected waves RW from the same target are received by the four receiving antennas 21, the extracted peak frequencies are the same between the frequency spectra of the four beat signals SB. In addition, when there are a plurality of targets at different angles in the same frequency bin, information about the plurality of targets is included in the signal of one peak frequency in the frequency spectrum.

  Since the positions of the four receiving antennas 21 are different from each other, the phases of the reflected waves RW received by the four receiving antennas 21 are different from each other. For this reason, the phase information of the reception signal SR having the same frequency bin is different for each reception antenna 21.

  The angle estimation unit 42 separates information about a plurality of targets existing in the same frequency bin from a signal of one peak frequency by a predetermined angle calculation process for each of the up section and the down section. Estimate the angle of each mark. The angle estimation unit 42 pays attention to signals of the same frequency bin corresponding to each peak frequency (hereinafter referred to as a peak signal) in all frequency spectra of the four beat signals SB, and based on the phase information of these peak signals. Estimate the angle of the target. Hereinafter, the angle of the target estimated by the observation data deriving unit 34 is referred to as “peak angle”, and the signal power of the peak angle is referred to as “angle power”.

  The angle estimation unit 42 derives such a peak angle for all peak frequencies in the frequency spectrum of the up section and the down section. The angle estimation unit 42 uses, for example, a predetermined angle estimation method such as ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques), DBF (Digital Beam Forming), or MUSIC (Multiple Signal Classification). Done. Thereby, the angle estimation part 42 can estimate the power of several peak angles and the signal of each of these peak angles from the signal of one frequency.

  Through the above processing, the peak extraction unit 41 and the angle estimation unit 42 derive peak data corresponding to each of a plurality of targets existing in front of the vehicle 5 in each of the up section and the down section. The peak data includes data such as the peak frequency, the peak angle, and the power of the peak angle signal (hereinafter referred to as angle power).

  The pairing unit 43 determines the peak of the up section and the peak of the down section based on the degree of coincidence between the peak angle and angle power of the up section calculated by the angle estimation section 42 and the peak angle and angle power of the down section. Perform pairing to associate.

  For example, the pairing unit 43 pairs the peaks whose peak angle and angular power are close within a predetermined range among the angle estimation results of the peaks in the up period and the down period. For example, the pairing unit 43 calculates the Mahalanobis distance using the peak angle and angular power of the frequency peak of each of the up section and the down period, and calculates the peak of the up section and the peak of the down period at which the Mahalanobis distance becomes the minimum value. Associate. A known technique can be used to calculate the Mahalanobis distance.

  In this way, the pairing unit 43 associates peaks related to the same target. Accordingly, the pairing unit 43 derives observation data relating to each of the plurality of targets existing in front of the vehicle 5. Since such observation data is obtained by associating two peaks, it is also called “pair data”.

  The pairing unit 43 derives, as observation data, data indicating the instantaneous values of the speed and distance of each target with respect to the vehicle 5 (radar apparatus 1) from the peak of the paired up section and down period. For example, the pairing unit 43 derives the velocity from the difference between the peak frequencies of the two peak data in the up section and the down section that are the basis of the observation data (pair data), and from the sum of the peak frequencies of the two peak data. The distance can be derived. Further, the pairing unit 43 instantaneously calculates the angle of each target with respect to the vehicle 5 (radar apparatus 1) from the average value of the peak angles of the two peak data of the up section and the down section, which are the sources of the observation data (pair data). Data indicating the value is derived as observation data.

The pairing unit 43 outputs observation data Z (t) for each target including distance observation data zr (t), velocity observation data zv (t), and angle observation data za (t) to the continuity determination unit 35. . The distance observation data zr (t), velocity observation data zv (t), and angle observation data za (t) are data indicating instantaneous values of the vertical distance, velocity, and angle at the time t with respect to the target. t represents a calculation time, and in the following, the latest calculation time is represented as t _n and the previous calculation time is represented as t _n−1 .

[2.3.5. Continuity determination unit 35]
The continuity determination unit 35 determines temporal continuity between observation data derived in the past and newly derived observation data. The continuity determination unit 35 includes a prediction unit 51, a setting unit 52, and a determination unit 53.

[2.3.5.1. Prediction unit 51]
The prediction unit 51 uses the observation data Z (t) newly derived for each target based on the observation data after filtering filtered by the filter unit 36 in the past (hereinafter referred to as fixed data F (t)). ) Is generated as predicted data P (t).

  Note that the prediction data P (t) includes distance prediction data pr (t), speed prediction data pv (t), and angle prediction data pa (t), which are prediction data of distance, speed, and angle with respect to the target. . Further, the fixed data F (t) includes distance fixed data fr (t), speed fixed data fv (t) and angle fixed data fa (t) which are fixed data of distance, speed and angle with respect to the target. . The determined data F (t) is output from the radar apparatus 1 to the vehicle control apparatus 2 as information on the distance, speed, and angle with the target.

  The storage unit 60 stores past observation data Z (t) and past confirmed data F (t), and the prediction unit 51 stores the past confirmed data F (t) stored in the storage unit 60. Based on this, prediction data P (t) is generated. FIG. 3 is a diagram illustrating an example of a data table including past observation data Z (t) and past confirmed data F (t) stored in the storage unit 60. Such a data table is stored in the storage unit 60 by the filter unit 36, for example.

In the data table shown in FIG. 3, observation data Z (t) and fixed data F (t) for each predetermined period Ts are associated with each target ID. For example, the observation data Z (t _n−M ) to Z (t _n−1 ) and the fixed data F (t _n− ) of the target with the target ID “0001” at each time t _n−M to time t _n−1 . _M ) to F (t _n-1 ) are set in the data table.

The observation data Z (t _n−M ) to Z (t _n−1 ) include distance observation data zr (t _{n− M} ) to zr (t _n−1 ), velocity observation data zv (t _n−M ) to zv (t _n-1 ) and angle observation data za (t _n-M ) to za (t _n-1 ) are included. Further, the final data F (t _n−M ) to F (t _n−1 ) include distance final data fr (t _n−M ) to fr (t _n−1 ) and speed final data fv (t _n−M). ) To fv (t _n−1 ) and angle determination data fa (t _n−M ) to fa (t _n−1 ).

  The prediction unit 51 performs state estimation calculation from the transition of each target, and generates prediction data P (t) for the target with each target ID. For example, the motion model of each target is defined from the state transition of each target, and the prediction data P (t) for the target with each target ID can be generated by a tracking filter using the motion model. The prediction unit 51 can determine the state transition of each target based on, for example, past confirmed data F (t) stored in the storage unit 60.

  Examples of motion models that can be used in the prediction unit 51 include a uniform acceleration linear motion model, a constant velocity linear motion model, and a meandering motion model. The tracking filters that can be used in the prediction unit 51 include a Kalman filter, a Bayes filter, There are particle filters. In the following, a particle filter using a constant acceleration linear motion model will be described as an example, but the present invention is not limited to this example. For example, a particle filter using a constant velocity linear motion model or a Kalman filter using a constant acceleration linear motion model may be used.

  The prediction unit 51 generates prediction data for a target with each target ID by a particle filter using a uniform acceleration linear motion model. Such a particle filter is one of generalized tracking models having high tracking characteristics with respect to linear motion or nonlinear motion.

The prediction unit 51, for example, from the past confirmed data F (t _n−M ) to F (t _n−1 ), the past target's longitudinal distance, lateral distance, longitudinal speed, lateral speed, longitudinal acceleration, and lateral acceleration. Is calculated. Here, the longitudinal distance, the longitudinal speed, and the longitudinal acceleration are parameters relating to the vehicle traveling direction, and the lateral distance, the lateral speed, and the lateral acceleration are parameters relating to the vehicle width direction. Then, the prediction unit 51 uses, for example, the following formula (1) as a state variable with the past (for example, previous) longitudinal distance, lateral distance, longitudinal speed, lateral speed, longitudinal acceleration, and lateral acceleration of each target as state variables. The current vertical distance, horizontal distance, vertical speed, horizontal speed, vertical acceleration, and lateral acceleration of each target are estimated.

In the above formula (1), x (t _n ) indicates the current target state, x (t _n-1 ) indicates the previous target state, and F (•) indicates the target state. It is a function that represents movement. U (t _n ) represents modeling noise.

Here, an observation model considering nonlinear motion is defined as shown in the following equation (2). In the following equation (2), H (•) is a function for converting the predicted data of the current vertical distance, lateral distance, vertical speed, lateral speed, vertical acceleration, and lateral acceleration into the current data z (t _n ). . n (t _n ) is observation noise and follows a Gaussian distribution.

The prediction unit 51 calculates prediction data P (t _n ) that is prediction data of the current observation data z (t _n ) based on the above equation (2). That is, the prediction unit 51 obtains the prediction data P (t _n ) by performing an operation such that H (x (t _n )) + n (t _n ) = p (t _n ). The prediction data P (t _n ) includes distance prediction data pr (t _n ), speed prediction data pv (t _n ), and angle prediction data pa (t _n ).

Note that the calculation method of the prediction data P (t _n ) by the prediction unit 51 is not limited to the above-described example. For example, when the speed of the target is constant and the lateral distance does not change, the prediction unit 51 calculates the distance prediction data pr (t _n ) and the speed prediction by calculating the following formulas (3) to (5). Predicted data P (t _n ) including data pv (t _n ) and angle predicted data pa (t _n ) can also be obtained. The prediction unit 51 can also obtain the prediction data P (t _n ) by other known estimation methods.

[2.3.5.2. Setting unit 52]
The setting unit 52 sets the prediction range A (t _n ) based on the prediction data P (t _n ) and the error Δ (t) between the observation data Z (t). For example, the setting unit 52 determines the size of the prediction range A (t _n ) based on the error Δ (t) of the past observation data Z (t), and predicts based on the prediction data P (t _n ). The arrangement of the range A (t _n ) is determined.

  FIG. 4 is a diagram illustrating an example of the configuration of the setting unit 52. As illustrated in FIG. 4, the setting unit 52 includes an error average value calculation unit 55, a variation width calculation unit 56, and a range determination unit 57 (an example of a determination unit).

The error average value calculation unit 55 can obtain the distance variance value sqrBr (t _n-1 ) by the calculation of the following equation (6). In the following formula (6), m is an integer from 1 to M−1, zr (t _n−m−1 ) is past distance observation data from the previous time, and fr (t _n−m−1). ) Is the distance determination data in the past from the previous time. Zr (t _n-1 ) is the previous distance observation data, and fr (t _n-1 ) is the previous distance determination data.

The error average value calculation unit 55 can obtain the distance error average value Br (t _n-1 ) by the calculation of the following equation (7). The distance error average value Br (t _n-1 ) is an average value of the error Δr (t) of the distance data zr (t) with respect to the distance determination data fr (t). The distance determination data fr (t) is a value estimated as a true value, and therefore the distance error average value Br (t _n-1 ) can be said to be an error average value with respect to the true value of the distance data.

The above formula (7) can be expressed as the following formula (8). In the following formula (8), sqrBr (t _n−2 ) is a distance variance value calculated last time. Therefore, the error average value calculation unit 55 can quickly calculate the distance error average value Br (t _n-1 ) by storing the previous distance variance value sqrBr (t _n-2 ). . The same applies to a speed error average value Bv (t _n-1 ) and an angle error average value Ba (t _n-1 ) described later.

Similarly, the average error value calculation unit 55 calculates the average speed error value Bv (t _n-1 ). The speed error average value Bv (t _n-1 ) is an average value of the error Δv (t) of the speed observation data zv (t) with respect to the speed determination data fv (t). The speed determination data fv (t) is a value estimated as a true value, and therefore the speed error average value Bv (t _n-1 ) is an error average value with respect to the true value of the speed observation data zv (t). Can do.

Specifically, the error average value calculation unit 55 can obtain the speed error average value Bv (t _n-1 ) by the calculation of the following equation (9). In the following formula (9), zv (t _n−m−1 ) is past speed determination data than the previous time, fv (t _n−m−1 ) is past speed prediction data than the previous time, and zv (T _n-1 ) is the previous speed determination data, and fv (t _n-1 ) is the previous speed prediction data.

Similarly, the average error value calculation unit 55 calculates the average angle error value Ba (t _n−1 ). The angle error average value Ba (t _n-1 ) is an average value of the error Δa (t) of the angle observation data za (t) with respect to the angle determination data fa (t). The angle determination data fa (t) is a value estimated as a true value, and therefore the angle error average value Ba (t _n-1 ) is an error average value with respect to the true value of the angle observation data za (t). Can do.

Specifically, the error average value calculation unit 55 can obtain the angle error average value Ba (t _n-1 ) by calculation of the following equation (10). In the following formula (10), za (t _n−m−1 ) is past angle observation data from the previous time, fa (t _n−m−1 ) is past angle determination data from the previous time, and za (T _n-1 ) is the previous angle observation data, and fa (t _n-1 ) is the previous angle determination data.

  The variation width calculation unit 56 also includes the variation width ΔR of the error Δr (t) of the distance observation data zr (t), the variation width ΔV of the error Δv (t) of the speed observation data zv (t), and the angle observation data. A variation width ΔA of the error Δa (t) of za (t) is detected. Here, the variation widths ΔR, ΔV, and ΔA will be described. FIG. 5 is a diagram illustrating an example of error variation widths ΔR, ΔV, and ΔA of distance observation data zr (t), velocity observation data zv (t), and angle data za (t).

As shown in FIG. 5, the variation width ΔR is a difference between the maximum positive value rmaxp and the negative maximum value rmaxn of the error Δr (t) of the distance observation data zr (t). The error Δr (t) of the distance observation data zr (t) is the difference between the distance observation data zr (t) and the distance determination data fr (t). For example, Δr (t _n−1 ) = zr ( tn _-1 ) -fr (tn _-1 ).

Similarly, the variation width ΔV is a difference between the maximum positive error value vmaxp and the negative maximum error value vmaxn of the error Δv (t) of the speed observation data zv (t), and the speed observation data zv (t ) Error Δv (t) is a difference between the speed observation data zv (t) and the speed determination data fv (t). For example, Δv (t _n−1 ) = zv (t _n−1 ) −fv (t _n−1 ).

Similarly, the variation width ΔA is the difference between the maximum positive error value amaxp and the maximum negative error value amaxn of the error Δa (t) of the angle observation data za (t), and the angle observation data za (t ) Error Δa (t) is a difference between the angle observation data za (t) and the angle determination data fa (t). For example, Δa (t _n−1 ) = za (t _n−1 ) −fa (t _n−1 ).

  The error Δr (t) of the distance observation data zr (t), the error Δv (t) of the velocity observation data zv (t), and the error Δa (t) of the angle observation data za (t) are obtained by the observation data deriving unit 34. This occurs due to the derivation accuracy of the observation data Z (t), the change in the reflection point of the object, etc. For example, when another vehicle traveling ahead cuts into the own lane, that is, the other vehicle ahead traveling in the adjacent lane is When the own vehicle (vehicle 5) suddenly interrupts the own lane on which the vehicle is traveling, the other vehicle moves in the lateral direction (direction from the adjacent lane to the own lane) at a relatively high speed.

  Since the shape of the other vehicle generally includes a curved surface or unevenness, the reflection point of the other vehicle relative to the transmission wave TW from the radar device 1 is relatively increased as the other vehicle moves in the lateral direction. It fluctuates greatly. Therefore, when another vehicle ahead in the adjacent lane suddenly cuts into the own lane in which the host vehicle is traveling, even if temporal continuity is obtained, as shown in FIG. The variation which is a temporal variation of the error (particularly, the variation of the error Δa (t) of the angle observation data za (t)) is large.

  Note that the example shown in FIG. 5 shows the variation in the observation data error when another vehicle traveling ahead cuts into the own lane. For example, when the other vehicle is far from the own vehicle, the speed observation is performed. The variation of the error Δv (t) of the data zv (1) and the variation of the error Δr (t) of the distance observation data zr (t) are larger than the variation of the error Δa (t) of the angle observation data za (t). In some cases.

Returning to FIG. 4, the description of the setting unit 52 will be continued. The range determination unit 57 of the setting unit 52 includes a distance prediction range Ar (t _n ) and a speed prediction range Av (t _n ) based on the calculation result by the error average value calculation unit 55 and the calculation result by the variation width calculation unit 56. ) And the angle prediction range Aa (t _n ).

The range determination unit 57 compares the distance error average value Br (t _n-1 ) with the distance error average value threshold Thr, and compares the speed error average value Bv (t _n-1 ) with the speed error average value threshold Thv. Then, the angle error average value Ba (t _n-1 ) is compared with the angle error average value threshold value Tha.

Then, the range determination unit 57 calculates a coefficient corresponding to the variation width ΔR when Br (t _n−1 )> Thr, Bv (t _n−1 )> Thv, and Ba (t _n−1 )> Tha. A coefficient Kv corresponding to Kr, a variation width ΔV, and a coefficient Ka corresponding to the variation width ΔA are determined. The range determination unit 57 uses the determined coefficients Kr, Kv, and Ka to determine the score C (t _n−1 ) by the calculation of the following equation (11).

The range determination unit 57 determines the sizes of the distance prediction range Ar (t _n ), the speed prediction range Av (t _n ), and the angle prediction range Aa (t _n ) based on the score C (t _n−1 ). . Thereby, the distance prediction range Ar (t _n ), the speed prediction range Av (t _n ), and the angle prediction range Aa (t _n ) having a size corresponding to the average error, distance, and variation width of the distance, speed, and angle are set. be able to.

  For example, the setting unit 52 compares the variation width ΔR with the variation threshold value ΔThr, compares the variation width ΔV with the variation threshold value ΔThv, and compares the variation width ΔA with the variation threshold value ΔTha. The setting unit 52 sets the coefficient Kr to a small value (for example, 0.1) if ΔR ≦ ΔThr, and sets the coefficient Kr to a large value (for example, 1) if ΔR> ΔThr.

  Similarly, the setting unit 52 sets the coefficient Kv to a small value (for example, 0.1) when ΔV ≦ ΔThv, and sets the coefficient Kv to a large value (for example, 1) when ΔV> ΔThv. The setting unit 52 sets the coefficient Ka to a small value (for example, 0.1) when ΔA ≦ ΔTha, and sets the coefficient Ka to a large value (for example, 1) when ΔA> ΔTha.

The setting unit 52, the smaller the score C _{(t n-1),} the distance predicted range Ar _(t n), the rate prediction range Av _{(t n)} and angle estimated range magnitude of Aa _{(t n),} respectively The prediction ranges Ar (t _n ), Av (t _n ), and Aa (t _n ) can be determined so as to decrease.

The setting unit 52 increases the coefficient Kr as the variation width ΔR is larger, increases the coefficient Kv as the variation width ΔV is larger, and increases the coefficient Ka as the variation width ΔA is larger. The setting unit 52 increases the distance prediction range Ar (t _n ), the speed prediction range Av (t _n ), and the angle prediction range Aa (t _n ) as the score C (t _n-1 ) increases. The prediction ranges Ar (t _n ), Av (t _n ), and Aa (t _n ) can be determined so as to decrease.

The range determination unit 57 determines the size of the prediction ranges Ar (t _n ), Av (t _n ), and Aa (t _n ), and then, based on the prediction data P (t _n ), the distance prediction range Ar (t _n ), the arrangement of the speed prediction range Av (t _n ) and the angle prediction range Aa (t _n ) is determined.

For example, the range decision portion 57, so that the center value of the distance expected range Ar _{(t n)} is coincident with the distance predicted data pr _{(t n),} determines the arrangement of the distance expected range Ar _{(t n).} Similarly, the range determining unit 57, the center value of the speed estimated range Av _{(t n)} determines the arrangement of the speed estimated range Av _{(t n)} to match the speed prediction data pv _{(t n).} Moreover, the range determining unit 57, the center value of the angle expected range Aa _{(t n)} determines the arrangement of the angular expected range Aa _{(t n)} to match the angle prediction data pa _{(t n).}

The range determination unit 57, as the distance value shifted from the center value of the expected range Ar _{(t n)} is coincident with the distance predicted data pr _{(t n),} determines the arrangement of the distance expected range Ar _{(t n)} You can also The same applies to the speed prediction range Av (t _n ) and the angle prediction range Aa (t _n ).

The method of setting the prediction range A (t) by the range determination unit 57 is not limited to the above-described example, and the range determination unit 57 includes the distance prediction range Ar (t _n ), the speed prediction range Av (t _n ), and each size of the angle expected range Aa (t _n) can be changed individually.

For example, the range determination unit 57 can reduce the distance prediction range Ar (t _n ) as the variation width ΔR increases when Br (t _n−1 )> Thr. Similarly, when Bv (t _n−1 )> Thv, the range determination unit 57 decreases the speed prediction range Av (t _n ) as the variation width ΔV increases, and Ba (t _n−1 )> Tha. In this case, the larger the variation width ΔA, the smaller the angle prediction range Aa (t _n ).

Moreover, the range determining unit 57, a distance error mean value Br a _{(t n-1)} The larger the distance expected range Ar _{(t n)} can be continuously or stepwise reduced. Similarly, the range determination unit 57 decreases the speed prediction range Av (t _n ) continuously or stepwise as the speed error average value Bv (t _n−1 ) increases, and the angle error average value Ba (t _{n− The} angle prediction range Aa (t _n ) can be reduced continuously or stepwise as ₁ ) increases.

In addition, the range determining unit 57 increases the value CR (= k1 × Br (t _n−1 ) + k2 × ΔR) obtained by weighting the distance error average value Br (t _n−1 ) and the variation width ΔR. The distance prediction range Ar (t _n ) can be reduced continuously or stepwise. As for the weighting, the larger the variation width ΔR, the larger the weighting coefficient, thereby increasing the weighted value CR.

Similarly, the range determination unit 57 can change the speed prediction range Av (t _n ) based on the value CV obtained by weighting the speed error average value Bv (t _n−1 ) and the variation width ΔV. . In addition, the range determination unit 57 can change the angle prediction range Aa (t _n ) based on a value CA obtained by weighting each of the angle error average value Ba (t _n−1 ) and the variation width ΔA.

Further, when the variation of the error Δ of the observation data Z is biased, the setting unit 52 excludes the range where the error Δ is relatively small from the prediction range A (t _n ), thereby predicting the range A (t _n ). Can be reduced.

  In addition, although the range determination part 57 demonstrated the example which sets the prediction range A (t) by making the difference | error of observation data Z (t) into the difference of observation data Z (t) with respect to definite data F (t), such an example It is not limited to. For example, the range determination unit 57 can set the prediction range A (t) with the error of the observation data Z (t) as the difference between the observation data Z (t) and the prediction data P (t).

  In this case, the range determination unit 57 converts the distance determination data fr (t), the speed determination data fv (t), and the angle determination data fa (t) into the distance prediction data pr (t), the speed prediction data pv (t), and The above-described processing can be performed in place of the angle prediction data pa (t).

For example, the range determination unit 57 converts zr (t) -fr (t), zv (t) -fv (t), za (t) -fa (t) to zr (t) -pr (t), zv ( t) -pv (t) and za (t) -pa (t). Then, the range determination unit 57 performs the calculations of the above formulas (6) to (11), the distance error average value Br (t _n-1 ), the speed error average value Bv (t _n-1 ), and the angle error average. The value Ba (t _n-1 ) can be determined.

The range determination unit 57 sets Δr (t _n−1 ) = zr (t _n−1 ) −pr (t _n−1 ) and Δv (t _n−1 ) = zv (t _n−1 ) −pv. (T _n-1 ) and Δa (t _n-1 ) = za (t _n-1 ) -pa (t _n-1 ). Thereby, the variation widths ΔR, ΔV, ΔA can be obtained.

The range determination unit 57 calculates the formula (11) when Br (t _n−1 )> Thr, Bv (t _n−1 )> Thv, and Ba (t _n−1 )> Tha. To obtain a score C (t _n-1 ). The range determination unit 57 determines the prediction ranges Ar (t _n ), Av (t _n ), and Aa (t _n ) according to the score C (t _n−1 ).

Similarly, the range determination unit 57 similarly uses the distance prediction range Ar (t _n ) and the speed prediction when the error of the observation data Z (t) is the difference between the observation data Z (t) and the prediction data P (t). The sizes of the range Av (t _n ) and the angle prediction range Aa (t _n ) can be individually changed.

[2.3.5.3. Determination unit 53]
Determination unit 53 shown in FIG. 2 determines the observed data Z the previously derived (t _n-1), this derived observation data Z and (t _n) is whether the same target object. Based on the observation data Z (t _n ) and the prediction range A (t _n ), the determination unit 53 determines that the current observation data Z (t _n ) is the same target as the past observation data Z (t _n−1 ). It is determined whether or not there is.

For example, the determination unit 53 determines whether or not the observation data Z (t _n ) exists within the prediction range A (t _n ) of each target. Then, the prediction range A _{(t n)} of the observed data Z _{(t n)} present in the predicted data P _{(t n)} to the closest observation data Z _{(t n)} is the last derived the observed data Z (t _n-1 ) is determined to be observation data Z (t _n ) continuous in time.

For example, the determination unit 53 includes distance data zr (t _n ), speed data zv (t _n ), angle data za (t _n ), distance prediction data pr (t _n ), speed prediction data pv (t _n ) and The Mahalanobis distance between the prediction data P (t _n ) and each observation data Z (t _n ) is calculated using the angle prediction data pa (t _n ). Then, the determination unit 53, prediction data P _{(t n)} the shortest observation data Mahalanobis distance between Z _{(t n)} predicted data P _{(t n)} to the current observation data Z of target object corresponding _{(t n} ).

Note that the determination unit 53 determines the difference between the distance data zr (t _n ) and the distance prediction data pr (t _n ), the difference between the speed data zv (t _n ) and the speed prediction data pv (t _n ), and the angle. The observation data Z (t _n ) having the smallest value obtained by weighting and adding the difference between the data za (t _n ) and the angle prediction data pa (t _n ) is the target corresponding to the prediction data P (t _n ). It can also be determined as the current observation data Z (t _n ).

[2.3.6. Filter unit 36]
The filter unit 36 shown in FIG. 2 derives the definite data F (t _n ) by smoothing the current observation data Z (t _n ) of the target in the time axis direction. For example, the filter unit 36 derives a value obtained by adding the weighted result to the current observation data Z (t _n ) of the target and the prediction data P (t _n ) as the final data F (t _n ). Can do.

For example, the filter unit 36 calculates the distance determination data fr (t _n ), the speed determination data fv (t _n ), and the angle determination included in the determination data F (t _n ) by calculating the following formulas (12) to (14). Data fa (t _n ) can be obtained. In the following formulas (12) to (14), Cr, Cv, and Ca are coefficients for weighting, and are changed depending on, for example, the behavior of the target.

  FIG. 6 is a diagram illustrating an example of a relationship between the distance prediction range Ar (t), the distance prediction data pr (t), the distance data zr (t), and the distance determination data fr (t). As shown in FIG. 6, distance determination data fr (t) is determined based on the relationship between the distance data zr (t) and the distance prediction data pr (t). In FIG. 6, the distance data zr (t) within the distance prediction range Ar (t) is determined as the distance data zr having temporal continuity, and the average value or variation width of the error Δ (t) is determined. This shows that the size of the distance prediction range Ar (t) is changed.

  Note that the filter unit 36 is not limited to the above-described processing and can be variously modified as long as variations in the target distance, speed, and angle with respect to the true value can be suppressed.

[2.3.7. Output unit 37]
The output unit 37 outputs target information including the confirmed data F (t _n ) of each target to the vehicle control device 2.

[3. Processing by Data Processing Unit 33]
Next, an example of the flow of processing executed by the data processing unit 33 will be described using a flowchart. FIG. 7 is a flowchart illustrating an example of a processing procedure executed by the data processing unit 33, and is a process that is repeatedly executed.

As shown in FIG. 7, the observation data deriving unit 34 of the data processing unit 33 is based on the frequency spectrum converted by the Fourier transform unit 32, and at least one data among distance, speed, and angle data with respect to each target. The current observation data Z (t _n ) including is derived (step S11). In addition, the continuity determination unit 35 of the data processing unit 33 calculates the current prediction data P (t _n ) for each target (step S12).

Further, the continuity determination unit 35 determines an error average value Δav (for example, distance error average value) of past observation data (for example, Z (t _n−1 ) to Z (t _n−M−1 )) of each target. Br (t _n-1 ), speed error average value Bv (t _n-1 ), angle error average value Ba (t _n-1 )) are calculated (step S13). The continuity determination unit 35 also includes a variation width Δw (for example, ΔR, ΔV, ΔA) of past observation data (for example, Z (t _n−1 ) to Z (t _n−M−1 )) of each target. ) Is calculated (step S14).

The continuity determination unit 35 determines the size of the prediction range A (t _n ) for each target based on the error average value Δav and the variation width Δw of the past observation data of each target (step S15). Then, the continuity determination unit 35 determines that the prediction range A (t _n ) whose size is adjusted in step S15 has a predetermined relationship with the prediction data P (t _n ) (for example, the prediction range A (t _n ) has the prediction range A ( t _n) to set the expected range a _{(t n)} to be the central value) of the (step S16). Note that the order of step S11 and steps S12 to S16 may be reversed. Further, the order of steps S12 to S14 and the order of steps S15 and S16 may be interchanged.

Next, the continuity determination unit 35 determines whether or not the current observation data Z (t _n ) is in the prediction range A (t _n ) (step S17). When the continuity determination unit 35 determines that the current observation data Z (t _n ) is present in the prediction range A (t _n ) (step S17, Yes), the observation data Z closest to the prediction data P (t _n ) (T _n ) is determined as observation data Z (t _n ) of the target corresponding to the prediction data P (t _n ) (step S18).

Next, the filter unit 36 smoothes the current observation data Z (t) of the target in the time axis direction and derives definite data F (t) (step S19). Then, the continuity determination unit 35 determines whether or not the processing of steps S17 and S18 for all the prediction ranges A (t _n ) of the target detected last time has been completed (step S20).

If the processing for all of the expected range A (t _n) determined not to be finished (step S20; No), the continuity determining section 35, for processing for the remaining expected range A (t _n), The process returns to step S17. Meanwhile, the continuity determining section 35, when determining that there is no prediction range A _{(t n)} to the current observation data Z _{(t n)} (step S17; No), or, all of the predicted range A _(t n) When it is determined that the process for the process is finished (step S20; Yes), the data processing unit 33 finishes the process shown in FIG. 7, and repeatedly executes the process from step S11 after a predetermined time.

  In the above-described example, the radar apparatus 1 detects a target using FM-CW (Frequency Modulated-Continuous Wave), but the radar apparatus 1 is an FCM (Fast Chirp Modulation) type radar apparatus. May be. In this case, the radar apparatus 1 performs two-dimensional fast Fourier transform processing on the beat signal SB for each chirp wave whose frequency continuously increases or decreases, and includes observation data including distance and speed acceleration values from the target. Is derived.

As described above, the radar apparatus 1 according to the embodiment includes the observation data deriving unit 34 (an example of a deriving unit), the predicting unit 51, the setting unit 52, and the determining unit 53. The observation data deriving unit 34 derives observation data Z (t) including at least one of the target data zr (t), zv (t), and za (t) based on the reflected wave from the target. The prediction unit 51 generates prediction data P (t) obtained by predicting the observation data Z (t) derived by the observation data deriving unit 34. The setting unit 52 sets the prediction range A (t) based on the prediction data P (t) generated by the prediction unit 51. When the observation data Z (t _n ) derived this time by the observation data deriving unit 34 is within the prediction range (t _n ) predicted by the prediction unit 51, the setting unit 52 derives the previous time by the observation data deriving unit 34. It is determined that the observed data Z (t _n-1 ) and the observation data Z (t _n ) derived this time are observation data of the same target that is temporally continuous. The setting unit 52 determines a size of the prediction range A (t _n ) based on the error of the observation data Z (t) derived in the past by the observation data deriving unit 34 (an example of a determination unit). Is provided. As described above, since the radar apparatus 1 determines the size of the prediction range A (t) according to the error of the observation data Z (t), even when the error of the observation data Z (t) is large. The temporal continuity of the target can be accurately determined.

The setting unit 52 also determines an error average value Δav (for example, Br (t _n−1 ), Bv (t _n−1 ), Ba () of the observation data Z (t) derived in the past by the observation data deriving unit 34. t _n-1 )) is calculated. The range determination unit 57 decreases the prediction range A (t _n ) as the error average value Δav calculated by the error average value calculation unit 55 increases. Thereby, for example, when the error average value Δav of the observation data Z is large, the prediction range A (t) is small, so that the temporal continuity of the target can be accurately determined.

The setting unit 52 includes a variation width calculation unit 56 that calculates a variation width Δw (for example, ΔR, ΔV, ΔA) of the error of the observation data Z (t) derived in the past by the observation data deriving unit 34. The range determination unit 57 decreases the prediction range A (t _n ) as the variation width Δw calculated by the variation width calculation unit 56 increases. Thereby, for example, when the variation width Δw, which is the temporal variation width of the error Δ of the observation data Z, is large, the prediction range A (t) is small, so that the temporal continuity of the target is accurate. Can be judged well.

Further, the range determining unit 57 uses the error average value Δav (for example, Br (t _n−1 ), Bv (t _n−1 ), Ba (t _n−1 )) derived by the error average value calculating unit 55. When a threshold value (for example, Thr, Thv, Tha) is exceeded, the prediction range A (t _n ) is decreased. As a result, when the error average value Δav of the observation data Z is equal to or greater than the threshold value, the prediction range A (t) is reduced, so that the temporal continuity of the target can be accurately determined.

In addition, the range determination unit 57 decreases the prediction range A (t _n ) as the variation width Δw calculated by the variation width calculation unit 56 increases. As a result, when the error average value Δav of the observation data Z is equal to or greater than the threshold value, the prediction range A (t) becomes smaller as the variation width Δw is larger, so the temporal continuity of the target is accurately determined. be able to.

Further, the range determination unit 57 weights the error average value Δav derived by the error average value calculation unit 55 according to the magnitude of the variation width Δw calculated by the variation width calculation unit 56 to obtain a score C ( The prediction range A (t _n ) is reduced based on t _n−1 ). It is possible to in this way the expected range of the magnitude corresponding to the score A (t _n), by appropriately adjusting the weighting, it is possible to accurately determine the temporal continuity of the target.

  In addition, the observation data Z (t) includes distance, speed and angle data zr (t), zv (t) and za (t) with respect to the target. The error average value calculation unit 55 calculates Br (t), Bv (t), and Ba (t) that are error average values of the data zr (t), zv (t), and za (t). When the Br (t), Bv (t), and Ba (t) exceed the respective threshold values Thr, Thv, and Tha, the range determination unit 57 predicts the prediction ranges Ar (t), Av (t), and Aa (t). Make it smaller. Accordingly, the sizes of the prediction ranges Ar (t), Av (t), and Aa (t) can be appropriately set by comprehensively considering the three relationships of distance, speed, and angle.

Further, the variation width calculation unit 56 calculates the variation widths ΔR, ΔV, and ΔA of the errors Δr, Δv, and Δt of the data zr (t), zv (t), and za (t). The range determination unit 57 performs weighting on Br (t _n−1 ), Bv (t _n−1 ), and Ba (t _n−1 ) according to the sizes of the variation widths ΔR, ΔV, and ΔA and adds them. To obtain a score C (t _n-1 ). Thereby, it is possible to accurately set the prediction ranges Ar (t), Av (t), and Aa (t) having a size corresponding to the score, comprehensively considering the three relationships of distance, speed, and angle. it can.

  The radar apparatus 1 includes a filter unit 36 (an example of a correction unit). The filter unit 36 corrects the observation data Z (t) derived by the observation data deriving unit 34 based on the prediction data P (t) predicted by the prediction unit 51, and the correction result is confirmed data F (t). Output as (an example of definite data). The setting unit 52 calculates the difference between the observation data Z (t) and the confirmed data F (t) obtained by correcting the observation data Z (t) as an error Δ of the observation data Z (t). Thereby, for example, when the confirmed data F (t) is close to the true value, the error Δ of the observation data Z (t) can be detected with high accuracy.

  The setting unit 52 calculates a difference between the observation data Z (t) and the prediction data P (t) as an error Δ of the observation data Z (t). Thereby, for example, when the predicted data P (t) is closer to the true value than the confirmed data F (t), the error Δ of the observed data Z (t) can be detected with high accuracy.

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

DESCRIPTION OF SYMBOLS 1 Radar apparatus 10 Transmitting part 20 Receiving part 30 Signal processing part 31 Transmission control part 32 Fourier transform part 33 Data processing part 34 Observation data deriving part (an example of deriving part)
35 Continuity determination unit 36 Filter unit (an example of a correction unit)
37 output unit 41 peak extraction unit 42 angle estimation unit 43 pairing unit 51 prediction unit 52 setting unit 53 determination unit 55 error average value calculation unit 56 variation width calculation unit 57 range determination unit (an example of a determination unit)
60 storage unit

Claims (10)

A derivation unit for deriving observation data of the target including at least one of data of distance, relative velocity and angle with the target based on a reflected wave by the target;
A prediction unit that generates prediction data obtained by predicting the observation data;
A setting unit for setting a prediction range based on the prediction data;
When the observation data derived this time by the deriving unit is within the prediction range, observation of the same target in which the observation data previously derived by the deriving unit and the observation data derived this time are continuous in time A determination unit that determines that the data is data,
The setting unit
A radar apparatus comprising: a determining unit that determines a size of the prediction range based on an error of observation data derived in the past by the deriving unit. The setting unit
An error average value calculation unit for calculating an average value of errors of the observation data derived in the past,
The determination unit
The radar apparatus according to claim 1, wherein the prediction range is reduced as the average value increases. The setting unit
A variation width calculation unit that calculates a variation width that is a temporal variation range of the error of the observation data derived in the past,
The determination unit
The radar apparatus according to claim 1, wherein the prediction range is reduced as the variation width increases. The setting unit
An error average value calculation unit for calculating an average value of errors of the observation data derived in the past,
The determination unit
The radar apparatus according to claim 1, wherein the prediction range is reduced when the average value exceeds a threshold value. The setting unit
A variation width calculation unit that calculates a variation width that is a temporal variation range of the error of the observation data derived in the past,
The determination unit
The radar apparatus according to claim 4, wherein the prediction range is reduced based on a score obtained by weighting the average value according to the size of the variation width. The observation data includes the distance, the relative velocity, and the angle data,
The error mean value calculation unit is
An average value of errors of the distance, the relative speed, and the angle data is calculated,
The determination unit
The radar apparatus according to claim 5, wherein the prediction range is reduced when an average value of errors of the distance, the relative speed, and the angle data exceeds a threshold value. The variation width calculator is
A variation width that is a temporal variation range of each error of the distance, the relative speed, and the angle data is calculated,
The determination unit
The average value of each error of the distance, the relative speed and the angle data is weighted according to the variation width of the error of the distance, the relative speed and the angle data and added. The radar apparatus according to claim 6, wherein the score is obtained. The setting unit
The radar apparatus according to claim 1, wherein a difference between the observation data and the prediction data is calculated as the error of the observation data. A correction unit that corrects the observation data based on the prediction data and outputs the correction result as confirmed data,
The setting unit
The radar apparatus according to claim 1, wherein a difference between the observation data and the confirmed data obtained by correcting the observation data is calculated as the error of the observation data. A continuity determination method executed by a computer,
A derivation step of deriving observation data of the target including at least one of distance, relative velocity and angle data with the target based on a reflected wave by the target;
A prediction step of generating prediction data obtained by predicting the observation data;
A setting step for setting a prediction range based on the prediction data;
Observation of the same target in which the observation data derived last time by the derivation step and the observation data derived this time are continuous in time when the observation data derived this time by the derivation step is within the prediction range. A determination step of determining that the data is data,
The setting step includes
A continuity determination method comprising: a change step of changing a size of the prediction range based on an error of observation data derived in the past by the derivation step.

Images