JP2015054603A - Object detection device - Google Patents

Abstract

An object detection apparatus capable of discriminating a moving object and a stationary object in a predetermined vehicle situation.
An object detection apparatus 100 for detecting an object based on radar reflected from the object, the position recording means 22 for recording the position information of the object, and the own vehicle acquired by the position information acquisition means. When the current position is a predetermined place, based on past position information of the same object recorded by the position recording means, a determination means 232 for determining whether the object is a stationary object; and the determination means And alarm control means 233 for controlling the alarm means 14 based on the determination result.
[Selection] Figure 11

Description

  The present invention relates to an object detection device that detects an object based on radar reflected from the object.

  Driving assistance that detects moving objects around the host vehicle and alerts the driver is known. In addition to detecting obstacles in the traveling direction of a vehicle with a radar or a camera, a technique for detecting a moving object in a direction that becomes a driver's blind spot and outputting an alarm or the like is being put into practical use (for example, Patent Literature 1). Patent Document 1 discloses a vehicle obstacle detection device that detects the position of the host vehicle and the moving direction of the host vehicle, and changes the direction when an obstacle is detected. By detecting the position of the host vehicle, for example, an intersection road that intersects the traveling road of the host vehicle, such as an intersection or a junction of an expressway, is detected, and the position and movement direction of the host vehicle with respect to the detected road configuration are detected. Based on this, the detection direction of an object detection means such as a millimeter wave or a laser is set.

JP 2005-231450 A

  However, in the conventional obstacle detection method, there is a case where it is difficult to distinguish between a moving object and a stationary object, and there is a problem that it is difficult to appropriately call attention.

  For example, when a merging vehicle enters a lane adjacent to the lane of the host vehicle at the merging point, but the merging vehicle joins the rear side of the host vehicle, the merging vehicle on the rear side is detected as a stationary object. There is.

  An object of this invention is to provide the target object detection apparatus with which the discrimination | determination precision of a moving object and a stationary object improved in view of the said subject.

  The present invention is an object detection device that detects an object based on radar reflected from the object, and includes a position recording unit that records position information of the object, and a current vehicle acquired by the position information acquisition unit. When the position is a predetermined place, based on past position information of the same object recorded by the position recording unit, a determination unit that determines whether or not the object is a stationary object, and a determination result of the determination unit And alarm control means for controlling the alarm means based on the above.

  It is possible to provide a target object detection apparatus with improved accuracy of discrimination between moving objects and stationary objects.

It is an example of the figure explaining the subject etc. which an object detection apparatus solves. It is an example of the system block diagram of a target object detection apparatus. It is an example of the figure explaining the target information detected by a radar apparatus. It is an example of the functional block diagram of driving assistance ECU. It is an example of the figure which illustrates the composition of a Kalman filter typically. It is an example of the figure which shows target object data. It is an example of the flowchart figure which shows the procedure in which a target object detection apparatus detects a target object and alarms (the 1). It is an example of the flowchart figure which shows the procedure in which a target object detection apparatus detects a target object and alarms (the 2). It is an example of the figure which illustrates typically the prediction point of an object, an observation point, etc. It is an example of the figure which illustrates typically the prediction point of an object, an observation point, etc. It is an example of the figure explaining the switching of the application in a junction. It is an example of the figure explaining the switching of the application at the time of curve driving | running | working.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. However, the technical scope of the present invention is not limited to this embodiment.

[About assignment]
First, the object detection apparatus has two detection functions of BSM (Blind Spot Monitoring-System) and LCA (Lane Change Assist).
BSM: Detects a vehicle in a blind spot. The detection area is a short distance range on the rear side of the vehicle, and this detection area may be referred to as a BSM area.
LCA: A moving object with a high relative speed is detected. The detection area is the rear side of the vehicle farther than the BSM, and this detection area may be referred to as an LCA area.

  The problem etc. which the target object detection apparatus of this embodiment solves are demonstrated using FIG. FIG. 1A shows a situation where a moving object is determined as a stationary object. The host vehicle is equipped with an object detection device that periodically detects an object on the left rear side with a radar. White circles are past prediction points for stationary objects, and black circles are the latest prediction points for stationary objects. The prediction points will be described later.

  As a procedure for recognizing a stationary object or a moving object of the object detection apparatus, BSM recognizes an object that continues to be reflected directly beside it as a stationary object.

  However, when a stationary object such as a wall is laid in front of the junction point in this way, it may be difficult to distinguish between the stationary object and the moving object. The host vehicle recognizes a stationary object (wall) that continues to be reflected directly beside it as a stationary object. However, if a merging vehicle at approximately the same speed as the host vehicle merges at this merging point, Remains to recognize the right side point as a stationary object. This is because the feature amount of the moving object is difficult to appear in a short time, and the past determination state recognized as a stationary object tends to be maintained in order to suppress frequent alarms. Therefore, there is a case where it is not delayed for recognizing the merged vehicle as a moving object.

  FIG. 1B shows an example of a situation where a stationary object is determined as a moving object. The white triangle is the past prediction point of the moving object, and the black triangle is the latest prediction point of the moving object. There are guardrails along the road, but other vehicles behind are detected by LCA before entering the curve. When the vehicle enters a right turn curve, there is a reflection point of the guard rail at a distance, so the guard rail may be detected as another vehicle that has been detected so far. In addition, as the vehicle heads toward the curve exit, the reflection point of the guardrail may be detected as a moving object with high relative speed (moving object approaching the host vehicle). That is, a reflected wave from a guardrail or the like is detected as a moving object that gradually approaches from the rear side when viewed from the host vehicle, as shown in a dotted frame in the figure. For this reason, the object detection device may determine that a stationary object such as a guardrail or a wall is a moving object and alerts it even when there is no moving object.

  The object detection device of the present embodiment makes it possible to discriminate between a stationary object and a moving object with higher accuracy in the situation as shown in FIGS.

[Outline of object detection device]
The object detection apparatus of this embodiment suppresses recognizing a moving object as a stationary object or recognizing a stationary object as a moving object as follows. First, a case where a moving object is determined to be a stationary object as shown in FIG.
(1) The target object detection apparatus acquires the position information of the host vehicle using a GNSS (Global Navigation Satellite System) or the like.
(2) The observation point is not detected because the wall is interrupted at the junction as shown in FIG. For this reason, the prediction points are detected discontinuously or the positions of the prediction points vary. On the other hand, when the merging vehicle merges with the host vehicle at a substantially constant speed, the object detection device starts detecting a stationary object directly beside the host vehicle. The target object detection device detects a target point (another vehicle) that has been determined to be a stationary object by detecting that it is a merging point from position information and using the feature amount (discontinuous time, variation, etc.) of the predicted point. Switch to moving objects.
(3) Since the moving object is detected in the BSM area by determining that the object is a moving object, the object detection device can output an alarm to the driver.

  In this way, when another vehicle joins to the rear side at the joining point, the switching from the stationary object to the moving object is judged based on the feature value of the predicted point, so the joining of traveling in the BSM area while suppressing alarm delay etc. The presence of a vehicle can be warned.

Next, a case where a stationary object is determined as a moving object as shown in FIG.
(1) The object detection device acquires the position information of the host vehicle using GNSS or the like.
(2) When the current position of the vehicle is near a curve, the object detection device prohibits an alarm by LCA.
(3) When a reflected wave from a guardrail or the like is determined as a moving object that gradually approaches from the rear side, an alarm can be suppressed because an alarm is prohibited.

On the other hand, the mobile body detected in the LCA area may actually approach gradually from the rear side. In this case, since an alarm should be given, the object detection device operates as follows.
(1) The object detection device acquires the position information of the host vehicle using GNSS or the like.
(2 ′) When the current position of the vehicle is near the curve, the object detection device prohibits an alarm by LCA. The object detection device records that the object is detected in the LCA area.
(3 ′) When an object is detected in the BSM area, if the object is an object detected in the LCA area, it is determined that the object is a moving body and an alarm is issued.

  In this way, when a moving object is replaced with a stationary object by a curve, it is possible to suppress alarming the stationary object, and when the moving object approaches as it is, the object is continuously detected from the LCA area to the BSM area. An alarm can be issued depending on whether or not.

[Configuration example]
FIG. 2 shows an example of a system configuration diagram of the object detection apparatus. In the object detection device 100, a radar device 11, a driving support ECU (Electronic Control Unit) 13, a navigation ECU 12, an operation device 14, and sensors (yaw rate sensor 15, wheel speed sensor 16, steering angle sensor 17) are CAN ( It is connected via an in-vehicle network such as Controller Area Network. As will be described later, the radar device 11 periodically (for example, 100 milliseconds) detects the distance to the object, the relative speed, the azimuth, and the reflected power (hereinafter, these may be collectively referred to as target information). It is transmitted to the driving support ECU 13. The relative position of the target with respect to the host vehicle is obtained from the distance and the direction.

  Further, the radar device 11 of the present embodiment detects an object from the side to the rear side. The radio wave transmission / reception units of the radar device 11 are respectively arranged inside the left end corner and the right end corner of the rear bumper of the vehicle made of a material that transmits radio waves such as resin. The center of the radar transmission direction is arranged to be about 45 to 70 degrees, for example, with respect to the direction parallel to the axle. The irradiation angle (radar detection range) can be designed to be 90 to 120 degrees, for example. The elevation angle is almost zero (parallel to the road surface). A parallel running vehicle can be detected based on the irradiation range of the radar. In addition, the radar apparatus 11 which detects the target object ahead of the own vehicle may be further attached.

  The radar device 11 is, for example, an FMCW (Frequency Modulated Continuous Wave) radar or a pulse radar. The radar apparatus 11 generates a beat signal for each reception antenna by mixing the transmission signal and the reception signal with a mixer. The time from when the transmission signal is transmitted until the reception signal is received is proportional to the distance to the object, and the frequency of the beat signal is shifted by the relative speed. Therefore, distance and relative speed can be obtained by FFT analysis of the beat signal, for example. In addition, the radar apparatus 11 is transmitting, switching automatically the radar of the frequency and frequency distribution suitable for BSM and LCA. Accordingly, the moving object is detected from each of the BSM area and the LCA area without being conscious of the driver.

  The (direction θ) of the target obstacle can be obtained by a monopulse method, DBF (Digital Beam Forming) processing, MUSIC (Multiple Signal Classification) analysis, Capon analysis, or the like.

  Examples of the sensors include a yaw rate sensor 15, a wheel speed sensor 16, a steering angle sensor 17, and the like. The yaw rate sensor 15 detects the rotational speed at which the direction perpendicular to the vehicle axle rotates horizontally with respect to the road surface. Further, the traveling direction can be detected by accumulating the rotational speed. The wheel speed sensor 16 is disposed on each wheel and detects the rotation speed of each wheel. The steering angle sensor 17 detects the steering angle of the steering wheel by the driver. In addition, it has a general sensor mounted on the vehicle.

  Navi ECU12 detects the position of the own vehicle using GNSS. The position information (latitude, longitude, altitude) of the host vehicle is detected by GNSS, and the current position is specified by accumulating the travel distance detected by the vehicle speed sensor from that position. Further, the traveling direction of the host vehicle is detected by a yaw rate sensor or a gyro sensor. The navigation ECU 12 displays a road map and the vehicle position on a display (not shown) and guides the route to the destination. The road map may be stored in advance in the vehicle, or may be downloaded from the outside by communicating with a server using a communication device.

  The driving assistance ECU 13 provides various driving assistances such as BSM and LCA based on the target information. BSM and LCA can be referred to as operation modes of the object detection apparatus 100, and programs for controlling the respective modes are referred to as BSM application and LCA application.

  The actuating device 14 is various on-vehicle devices for driving assistance that output an alarm in order to avoid an abnormal approach to an object. Examples of the operation device 14 include an alarm buzzer, a warning lamp, an audio output device, and a steering motor. It is common for BSM and LCA to have different activation devices 14 to operate, or to use the same activation device 14 but with different warning methods. In BSM and LCA, alarm may be performed by one or more of the following activation devices 14, and the alarm method is not limited in this embodiment. In the LCA, it is possible to set a warning condition that the winker switch is turned on or that the driver steers the steering wheel in the direction of the moving object.

  The alarm buzzer warns that there is a possibility of abnormally approaching the object by blowing the buzzer on the meter panel, and the warning lamp abnormally approaches the object by turning on / flashing the warning light on the meter panel. Call attention to the dangers. Moreover, you may arrange | position a warning lamp in the place which turns a line of sight when a driver | operator confirms a rear side, such as a side mirror.

  The audio output device outputs a message (for example, “There is another vehicle behind the vehicle”) from the speaker. The steering motor alerts the driver by transmitting vibrations. In the LCA, as one form of alarm, when the driver steers the steering wheel in the direction in which the object is present, an alarm may be given by applying rotational torque to the steering shaft in a direction opposite to the steering direction.

  FIG. 3 is an example of a diagram illustrating target information detected by the radar device 11. In the figure, only the radar device 11 at the left end corner is shown. The radar device 11 receives and analyzes a reflected wave obtained by the reflection of the radar on the object, and periodically measures the position of the observation point (triangle in the figure). However, the position of the target (observation point) detected by the radar is known to vary slightly, and the observation point is input to the tracking filter to generate a prediction point or a smooth point. It is treated as a position, and an alarm is output. Prediction points and smooth points generated by the tracking filter are created from past observation points and prediction points, and therefore are less likely to vary than observation points.

  In the tracking filter, the prediction point generated last is updated using the observation point. A plurality of observation points may be detected due to radar scattering or the like. In this case, for example, the prediction point is updated using the observation point closest to the prediction point within a predetermined distance. Extracting an appropriate observation point for such a finally generated prediction point is called pairing. In the figure, the prediction point ○ and the observation point Δ at time t4 are paired.

  FIG. 4 shows an example of a functional block diagram of the driving support ECU 13. The driving support ECU 13 includes a detection unit 21, a tracking unit 22, and an alarm unit 23. The detection unit 21 acquires target information detected by the radar device 11 and outputs an observation point. The detection unit 21 has an observation point calculation unit 212. The observation point calculation unit 212 converts the target information into coordinates (x, y) on a two-dimensional plane with the radar position of the host vehicle as the origin.

  The tracking unit 22 tracks the observation point based on the tracking filter and the motion equation, and outputs the object data. The object data will be described with reference to FIG. The tracking unit 22 includes a pairing unit 221, a duplication determination unit 222, and a tracking filter 223. The pairing unit 221 pairs the prediction point and the observation point. The pairing is performed because a pair of observation points is necessary for the tracking filter 223 to calculate a prediction point. For example, the observation point closest to the prediction point is determined as a pairing target. In addition, when there are a plurality of objects in the irradiation range of the radar, a plurality of observation points are detected, so it is necessary to select an observation point and pair it with a prediction point.

  When there are a plurality of prediction points paired with one observation point, the overlap determination unit 222 determines a prediction point for canceling the pairing with the observation point. For example, when the prediction point a and the observation point 1 are paired in the pairing set A and the prediction point b and the observation point 1 are paired in the pairing set B, the observation point is based on the evaluation value described later. A prediction point to be paired with 1 is determined (a prediction point for canceling pairing is determined).

  The tracking filter 223 may be a filter suitable for object tracking, such as a Kalman filter, an αβ filter, an αβγ filter, or a particle filter. The tracking filter 223 may track only the observation point on the assumption that the object moves linearly at a constant speed.

  The alarm unit 23 discriminates between a moving object and a stationary object, or warns an object that has been identified as a moving object. The alarm unit 23 includes an application switching unit 231, a moving / stationary object discrimination unit 232, and an alarm determination unit 233.

  The application switching unit 231 switches between the BSM application and the LCA application. As described above, each of BSM and LCA is provided as one application (BSM application, LCA application), but one program may provide the functions of BSM and LCA. In addition, an operation mode switches by switching an application.

  When the application switching unit 231 switches to the BSM application, the alarm determination unit 233 determines whether or not to alarm based on the BSM alarm determination condition. When the application switching unit 231 switches to the LCA application, the alarm determination unit 233 determines whether to perform the LCA alarm determination. It is determined whether or not to warn based on the condition. Application switching is performed, for example, from BSM to LCA every time a prediction point is calculated in one cycle (or from LCA to BSM first with LCA).

  The moving object / stationary object determination unit 232 determines whether the object is a moving object or a stationary object based on the position information of the host vehicle and the feature amount of the predicted point or smooth point. The feature amount is, for example, predicted point / smooth point variation, continuity, reflection intensity, or the like, or a value calculated from these. For example, the variation of the prediction points / smooth points is evaluated by taking the positions of the same prediction points in time series and analyzing the variance or standard deviation in the X-axis direction or Y-axis direction. It is estimated that the variation in the position in the X-axis direction or the Y-axis direction becomes large when the wall is gone and the observation point cannot be found from the state where the wall is detected as the object at the confluence.

  Continuity is the number of times that observation points that are paired with prediction points of the same object are successively generated. When there is no observation point within a predetermined distance from the prediction point, the observation point is created based on a motion model such as a constant velocity linear motion from a past prediction point or the like. That is, when the observation point is not detected, the tracking filter 223 does not immediately discard the prediction point in consideration of the fact that the radar device 11 cannot receive the observation point due to an influence such as flooding. For example, if an observation point that can be paired is not found more than a predetermined number of times, the predicted point is discarded.

  Therefore, predicted points continue to exist until an observation point is not found up to a predetermined number of measurements. For example, there may be a case where a wall disappears and an observation point cannot be found from a state where a wall is detected as an object at a junction. When the merged vehicle becomes a parallel running vehicle, an observation point that can be paired with the predicted point is detected again. However, it can be estimated that the observation point is interrupted in this way when the reliability of the prediction point is low (ie, the stationary object is replaced with the moving object).

  The reflection intensity is the intensity of electric power when an FFT analysis is performed on a beat signal including a reflected wave from the object. When the object exists, the power intensity shows a peak in the result of FFT analysis (time vs. power intensity). When this peak is equal to or greater than the threshold, the radar device 11 determines that there is an observation point, but it can be said that the higher the power intensity, the clearer the object is present. For example, in the process of switching from a stationary object to a moving object, the power intensity should decrease, so that a stationary object and a moving object can be distinguished based on the power intensity.

The moving object / stationary object discrimination unit 232
(i) Prediction point / smooth point variation is greater than threshold,
(ii) Of the past n times, the number of times that observation was not possible exceeds the threshold
(iii) It is determined that the object has been switched to a moving object when one or more of the states where the clarity of reflection is equal to or less than the threshold value continues n times or more are satisfied.

  In addition, the moving object / stationary object discriminating unit 232 detects the object detected in the LCA area in the BSM area in addition to continuously measuring the object when the vehicle position is traveling on a curve. It is determined that the object is a moving object. This utilizes the fact that when the object is a wall (stationary object), it is difficult to detect in the BSM area even if the moving object is replaced by a stationary object. Details will be described later.

  Note that the moving object / stationary object determination unit 232 can estimate that the vehicle is traveling on a curve using not only the position information of the navigation ECU 12 but also the yaw rate and the steering angle. In addition, it can be detected that the vehicle is in the vicinity of a curve or a merging point from road-to-vehicle communication.

  The alarm determination unit 233 determines whether to issue an alarm according to the alarm determination conditions of LCA and BSM. In both BSM and LCA, the object of alarm is a moving object.

  In the case of BSM, the warning determination unit 233 determines that a warning is given to a moving object existing in a blind spot within a predetermined processing distance (for example, within 2 to 3 meters) from the host vehicle. The warning unit 23 alerts the driver, for example, by turning on a warning lamp. Since there is no audible alarm, the presence of a vehicle other than the blind spot can be alarmed without making the driver feel bothersome.

  In the case of LCA, the alarm determination unit 233 determines that an alarm is issued when a moving object with a large relative speed within 5 meters is detected, for example. Then, the alarm unit 23 informs the driver by operating a blinker switch in the direction of the moving object, or by sounding an alarm sound or turning on a warning lamp when the steering angle of the steering wheel exceeds a predetermined value. Call attention.

[Example of tracking filter]
FIG. 5 is an example of a diagram schematically illustrating the configuration of the Kalman filter. The Kalman filter is a method for obtaining the most probable estimated value of the state variable x for optimal control. In the Kalman filter, the state of the system is described by the equation of state (1) and the observation equation (2).

ｘは状態変数のベクトルであり、本実施形態はｘ方向の位置ｘとｘ方向の速度ｖ_ｘ（ｘの一次微分）、及び、ｙ方向の位置ｙとｙ方向の速度ｖ_ｙ（ｙの一次微分）の４つの要素を有している。

x is a vector of state variables, and in this embodiment, the position x in the x direction and the velocity v _{x in the} x direction (first derivative of x), and the position y in the y direction and the velocity v _{y in the} y direction (first order of y). It has four elements (differentiation).

z is a vector indicating the observed value, and in this embodiment, the position (x, y) on the two-dimensional plane is used as an element. v _t is the noise of the process, and the Kalman filter assumes that the noise follows a Gaussian distribution. w _t is the noise of the observed value and is also assumed to follow a Gaussian distribution. Note that the variance matrix of v is Q, and the variance matrix of w is R.

  F is a matrix relating the state of the system at time t and the state of the system at time t + 1. H is a matrix that links the state variable x and the observation value z. F and H can be obtained by assuming that the motion model of the target within a minute time is, for example, constant velocity linear motion.

  In the Kalman filter, two prediction points are obtained. One is a prediction point at time t estimated before the observation value at time t is obtained. This is called a prediction point of the next cycle, and is represented by a symbol having “−” on x. The other is a prediction point at time t + 1 estimated from the observed value at time t. This is called the current prediction point and is represented by a symbol with “^” on x.

  As shown in FIG. 5, the predicted point of the next cycle is calculated by equations (B-1) and (B-2), and the current predicted point is calculated by equations (A-1) to (A-3). . Note that P is a covariance matrix of errors of prediction points.

First, when an appropriate initial value is given to P _{t + 1} , the Kalman gain K can be calculated by equation (A-1). When the observation point is detected, the current prediction point is obtained from the prediction point of the next cycle to which an appropriate initial value is given and the Kalman gain K, according to the equation (A-2). Further, the covariance matrix Pt of the error at the current prediction point is updated by the equation (A-3).

  Then, the predicted point of the next cycle is calculated from the current predicted point. The value of the predicted point of the next cycle is calculated from the current predicted point and the state equation by equation (B-1). The equation (B-2) is used to calculate the error covariance matrix Pt + 1 of the prediction point of the next cycle. The prediction point of the next cycle of Formula (B-1) is a prediction point used for warning determination. The values calculated by the equations (B-1) and (B-2) are used to calculate the current predicted point by the equations (A-1) to (A-3) for the next observation point.

  The tracking filter 223 calculates a smooth point of the current object using a plurality of past prediction points. The smooth point is the position of the object at time t, and the predicted point is the position of the object at time t + 1. By obtaining a smooth point, it is possible to estimate the state more stably than the predicted point.

  FIG. 6 is an example of a diagram illustrating object data of a certain object. Similar time-series data is recorded for each object. The prediction point and the smooth point are calculated by the tracking filter 223. The smooth point is the value at time t, but the predicted point is the value at time t + 1 predicted at time t. The relative speed is a relative speed in the x direction and y direction obtained from a past prediction point. The reflection intensity is the intensity of power at the observation point. The stationary object / moving object indicates whether the object is determined to be a stationary object or a moving object at each time. In addition, when the warning part 23 evaluates variation based on an observation point, you may include the observation point paired with target object data.

[Operation procedure]
7 and 8 are examples of flowcharts illustrating a procedure in which the object detection device 100 detects an object and issues an alarm. 9 and 10 are examples of diagrams for schematically explaining the prediction points, observation points, and the like of the object. The procedure of FIG. 7 is repeatedly executed while the radar apparatus 11 is operating.
S10: When the radar apparatus 11 outputs the target information, the observation point calculation unit 212 calculates the observation point. FIG. 9A schematically shows observation points. A bicycle runs in the detection range of the radar device 11, and a guardrail is laid along the traveling direction of the vehicle. Here, three observation points 1 to 3 are detected. The three observation points 1 to 3 are scattered from the bicycle and do not always correspond to the actual object. It may also be affected by guardrails.

Next, the tracking part 22 repeats the process of S20-S60 about all the prediction points.
S20: The pairing unit 221 determines whether or not there is an observation point within the pairing area of the prediction points. FIG. 9B is an example for explaining the pairing area. In addition, since the prediction point one cycle before has already been calculated, if the current time is t, the prediction point calculated in FIG. 9B is that at time t-1. A total of three prediction points t-1 are calculated, two for the guardrail and one for the bicycle. The pairing area is, for example, an area within a predetermined distance from each prediction point. FIG. 9B shows an example in which there are observation points in the pairing area.

  S30: When there is an observation point in the pairing area, the pairing unit 221 determines whether or not there are a plurality of observation points in the pairing area. As shown in FIGS. 9B and 9C, a plurality of observation points 1 to 3 in the pairing area are detected.

S40: When there are a plurality of observation points, the pairing unit 221 determines whether or not the evaluation value is larger than the threshold value.
The evaluation value = α × own vehicle speed + β × predicted point relative speed evaluation value is a value for evaluating the own vehicle speed and the relative speed between the own vehicle speed and the object. α and β are weightings. When the host vehicle speed or the relative speed is large, the evaluation value tends to increase.

  When the evaluation value is small (the own vehicle speed is slow and the relative speed is small), since there are many reflection points from a stationary object, the prediction point closest to the own lane is trusted to be paired. Such a pairing policy is referred to as “own lane direction priority”. On the other hand, if the evaluation value is large (the vehicle speed is fast and the relative speed is large), a prediction point with a large relative speed is paired by using the fact that there is a reflection point from a stationary object but there is a significant difference in the relative speed. The target of. Such a pairing policy is referred to as “moving object priority”.

  S50: When the evaluation value is less than or equal to the threshold value, the pairing unit 221 pairs the predicted point and the observation point with “own lane direction priority”. FIG. 9D is an example of a diagram illustrating an example of pairing with priority in the own lane direction. The predicted point of the bicycle at time t-1 is paired with observation point 1. That is, it is paired with the observation point 1 closest to the own lane.

  S60: When the evaluation value is larger than the threshold value, the pairing unit 221 performs “moving object priority” and pairs the observation point having the highest relative speed with the predicted point.

  S70: Next, the duplication determination unit 222 determines whether there is an observation point paired with a plurality of prediction points. FIG. 10A is a diagram illustrating an example of observation points paired with a plurality of prediction points. Since the observation point 1 is paired not only with the prediction point but also with the prediction point of the guardrail, it is determined that the observation point 1 is paired with a plurality of prediction points.

  S80: When there is an observation point paired with a plurality of prediction points, the overlap determination unit 222 determines whether or not the evaluation value is greater than a threshold value.

  S90: When the evaluation value is less than the threshold value, the overlap determination unit 222 prioritizes the prediction point closest to the own lane based on the “own lane direction priority” policy, and cancels the pairing with other prediction points. FIG. 10B is an example of a diagram illustrating cancellation of pairing. Here, prediction points that are eliminated based on “own lane direction priority” are shown. Observation point 1 has been paired with the predicted point of the bicycle and the predicted point of the guardrail, but the pairing with the predicted point of the guardrail is eliminated, leaving the predicted point of the bicycle closest to the own lane.

  S100: When the evaluation value is larger than the threshold value, the duplication determination unit 222 cancels the pairing with other prediction points, leaving the prediction point with the highest relative speed based on the “moving object priority” policy.

  S110: The duplication determination unit 222 creates a pairing set of prediction points and observation points without duplication.

  S120: Next, the tracking filter 223 calculates a prediction point and a smooth point. When it is determined in step S20 that there is no observation point in the past pairing area, a set of the prediction point and the prediction tracking history is input to the tracking filter 223. The prediction tracking history is a past prediction point or a smooth point of the object data in FIG.

  S130: A new prediction point at time t is calculated by the tracking filter 223. This prediction point t is paired with the observation point in the next cycle.

  S140: Also, a new smooth point is calculated by the tracking filter 223. FIG. 10C is an example of a diagram for explaining the prediction points and the smooth points. From the prediction point at time t-1 and the observation point 1, the prediction point t at the current time t and the smooth value at the current time t are calculated. The smooth point t is indicated by a star.

  Next, the warning unit 23 repeats the processes of S20 to S60 for all the prediction points. In addition, although it demonstrates as determining whether it becomes a warning object by BSM or LCA based on a smooth point, you may determine based on a prediction point.

  S150: The application switching unit 231 switches the application to LCA. Thereby, a smooth point in the LCA area becomes an alarm target.

  S160: The moving object / stationary object determination unit 232 determines whether or not the current position is near the curve, using the current position as the vehicle information.

  S170: When the current position is not near the curve, the alarm determination unit 233 issues an alarm according to the alarm determination condition of the LCA application.

  S180: When the current position is near the curve, the moving object / stationary object discrimination unit 232 records that tracking has started in the LCA area at a smooth point. Thereby, when the application is switched to BSM, it can be determined whether or not smooth points are continuously detected from the LCA area to the BSM area. The moving object / stationary object discriminating unit 232 prohibits an alarm for the object.

  S190: Next, the application switching unit 231 switches the application to BSM. Thereby, a smooth point in the BSM area becomes an alarm target. Note that S150 and S190 are executed according to the timing at which BSM and LCA are switched.

  S200: The moving object / stationary object discriminating unit 232 performs appropriate determination for each location. For this reason, it is determined first whether it is a junction.

  S210: When the current position is a merging point, the moving object / stationary object discriminating unit 232 uses the variation of observation points / smooth points, continuity, and clarity of reflection as features, and whether the smooth point is a moving object or a stationary object. Is determined. The moving object / stationary object discrimination unit 232 determines whether the object is a moving object or a stationary object based on the determination conditions (i) to (iii).

  S220: In the case of a stationary object, the alarm unit 23 does not alarm.

  S230: If the object is not a stationary object, the alarm unit 23 issues an alarm according to the alarm determination condition of the BSM application.

  S240: If the current position is not a merging point, the moving object / stationary object determination unit 232 determines whether a smooth point in the BSM area has been tracked from the LCA. That is, it is determined whether or not the tracking start in the LCA area is recorded at the smooth point. Here, it may be determined again whether or not the current position is near the curve.

  S230: If a smooth point is tracked from the LCA and further detected in the BSM area, it can be estimated as a moving object, so the alarm unit 23 warns based on the BSM application.

  S250: When the object is not tracked from the LCA, it may or may not be detected in the BSM area. If it is not detected in the BSM area, it will not be alarmed in the BSM area. For example, even if a wall is detected as a moving object, it does not enter the BSM area, so there is no alarm. Moreover, since the warning by LCA is prohibited, it can suppress that a stationary object is judged to be a moving object and alarmed. If an object that is not tracked from the LCA is detected in the BSM area, an alarm may be issued under the BSM alarm determination condition.

〔Concrete example〕
FIG. 11 is an example of a diagram illustrating application switching at a junction. It is assumed that the application switching unit 231 has already switched to the BSM application. In FIG. 11A, the wall is determined to be a stationary object. However, the determination is made that the stationary object is a moving object due to the joining of the joining vehicle.

Step i
The object detection device 100 detects the wall as a stationary object. Moreover, since the target object detection apparatus 100 has continuously acquired the position information of the host vehicle, it has been detected that the object detection device 100 is traveling near the junction.

Step ii
When there is no wall at the merge point, the radar apparatus 11 does not detect an observation point that can be paired with a prediction point (or a smooth point). In this case, the tracking filter 223 simply interpolates the observation points on the assumption that the prediction points are moving at a constant linear velocity. During this time, the parallel-running vehicles join the junction. The moving object / stationary object discriminating unit 232 determines whether the object is a stationary object or a moving object based on the discontinuity time of the predicted tracking in which the observation points are not continuously detected.

Step iii
From the information of step i and step ii, the moving object / stationary object determination unit 232 determines the object and the moving object that have been determined to be a stationary object. Thereby, the warning unit 23 can output a warning to the joining vehicle.

  Similarly in FIG. 11B, although the wall is determined to be a stationary object, it is switched to a determination that it is a moving object, but the determination is switched based on variations in position information and prediction points.

Step i
Similar to FIG. 11A, the object detection device 100 detects the wall as a stationary object and has detected that the vehicle is traveling near the junction.

Step ii
If the wall at the merge point disappears, the radar apparatus 11 detects the observation point, but the error increases, and the variation of the predicted point increases. During this time, the parallel-running vehicles join the junction. In this case, the moving object / stationary object determination unit 232 determines whether the object is a stationary object or a moving object based on the variation in the positional information of the predicted points.

Step iii
From the information of step i and step ii, the moving object / stationary object determination unit 232 determines the object and the moving object that have been determined to be a stationary object. Thereby, the warning unit 23 can output a warning to the joining vehicle.

  FIG. 12 is an example of a diagram illustrating application switching during curve driving. In FIG. 12A, the vehicle behind the host vehicle is not supplemented by the radar device 11, and the wall is supplemented. The application switching unit 231 switches to the LCA application.

Step i
Since the object detection device 100 continuously acquires the position information of the host vehicle, it has been detected that the vehicle is traveling near the curve. In addition, it is recorded that tracking has started in the LCA area.

Step ii
Moreover, although the target object detection apparatus 100 is detecting the back vehicle in LCA area, since it is near a curve, a warning is temporarily invalidated. As the vehicle advances to the curve exit, the object is switched from the moving object to the wall. However, since this wall is not detected in the BSM area, it is not an object of alarm, and since alarm is prohibited, it is not alarmed by LCA. For this reason, it is possible to suppress the output of an alarm when the moving object is replaced with a stationary object.

  In FIG. 12B, the vehicle behind the host vehicle is traveling along a curve together with the host vehicle. The application switching unit 231 switches to the LCA application.

Step i
This is the same as FIG.

Step ii
Moreover, although the target object detection apparatus 100 is detecting the back vehicle in LCA area, since it is near a curve, a warning is temporarily invalidated.

Step iii
The application switching unit 231 switches to the BSM application. Although the object is detected in the BSM area, the moving object / stationary object determination unit 232 determines that the object is a moving object because the object is continuously tracked from the LCA area. Thereby, the alarm unit 23 can issue an alarm.

  As described above, the object detection device 100 according to the present embodiment switches from a moving object to a stationary object at a predetermined location and changes from a moving object to a stationary object based on the features of the current position and the past position information of the object. It is possible to appropriately control whether or not an alarm is to be detected by detecting the change of the above.

11 Radar device 12 Navi ECU
13 Driving assistance ECU
DESCRIPTION OF SYMBOLS 14 Actuation device 21 Detection part 22 Tracking part 23 Alarm part 100 Object detection apparatus

Claims (5)

An object detection device that detects an object based on radar reflected from the object,
Position recording means for recording position information of the object;
When the current position of the host vehicle acquired by the position information acquisition unit is a predetermined location, it is determined whether the target object is a moving object based on past position information of the same target object recorded by the position recording unit. Determination means to perform,
Alarm control means for controlling the alarm means based on the determination result of the determination means;
The object detection apparatus characterized by having. If the current position of the host vehicle acquired by the position information acquisition means is a merging point,
The determination means has at least one of position information not continuously measured or position information fluctuating by a predetermined value or more with respect to an object that has been determined to be a stationary object based on the position information recorded by the position recording means. When it occurs, it is determined that the object has been switched from a stationary object to a moving object.
The object detection device according to claim 1. The alarm control means warns a moving object existing in a second range closer to the host vehicle than the first range, and a first operation mode for alarming a moving object existing in the first range. A second operating mode;
When the current position of the host vehicle acquired by the position information acquisition unit is a curve,
The determination unit suspends the determination of whether the position information recorded by the position recording unit is a moving object or a stationary object for an object included in the first range;
The alarm control means does not issue an alarm even if there is a moving object in the first range.
The object detection device according to claim 1. The determination means includes
For an object that has been determined as a moving object in the first range based on the position information recorded by the position recording means, if the current position information is included in the second range, the object is a moving object. And
The alarm control means warns the object in the second range;
The object detection device according to claim 3. An object detection device that detects an object based on radar reflected from the object,
Position detecting means for detecting the position of the host vehicle;
Position recording means for recording position information of the object;
If the current position of the host vehicle acquired by the position detection means is a predetermined place, whether or not the object is a moving object is determined based on past position information of the same object recorded by the position recording means. Determination means for determining;
An alarm means for outputting an alarm;
Alarm control means for controlling the alarm means based on the determination result of the determination means;
The object detection apparatus characterized by having.

Images