WO2014057621A1 - Object detecting device - Google Patents

Abstract

An object detecting device (1), wherein a reflection point estimator (S15) estimates a reflection point (4) on an object for a reflected wave using the history of the detection distance of an object and the history of the sensor position. A reflection point corrector (S19 to S22) corrects a low-precision reflection point (41) to a normal reflection point (42) near a corner of the object or a location corresponding to the direction in which the reflected wave arrives in a normal reflection point when an acute historical shape (40) is contained within an historical shape of the reflection point in the vicinity of a corner of the object.

Description

Object detection device Cross-reference of related applications

This disclosure is based on Japanese Patent Application No. 2012-224606 filed on October 9, 2012, the contents of which are incorporated herein by reference.

The present disclosure relates to an object detection device that detects an object existing on the side of a vehicle.

Conventionally, in order to detect a parking space, an object detection device that detects an object such as a parked vehicle existing on the side of the vehicle is known (see, for example, Patent Document 1). In the object detection device of Patent Document 1, a distance sensor such as an ultrasonic sensor is used to sequentially detect the distance to an object when the vehicle is moving along a side path of the parked vehicle (object). Further, the position (sensor position) of the distance measuring sensor when the distance is detected from the moving state of the vehicle is calculated. Then, based on the point sequence data (history) of the point (ranging point) that is separated from the sensor position to the side of the vehicle by the detection distance, and the sensor position history, rotation correction of the ranging point is performed. In other words, in the object detection device of Patent Document 1, the arrival direction of the reflected wave is estimated by the principle of triangulation using the history of the detection distance and the history of the sensor position, and the direction from the sensor position to the arrival direction is estimated. A distance measuring point is corrected at a point (reflecting point) separated by a detection distance. And the outline of a parked vehicle and the position of a corner are determined based on the point sequence data (history) of a reflective point.

Japanese Unexamined Patent Publication No. 2008-21039

This disclosure is intended to provide an object detection device capable of detecting a corner of an object with high accuracy even when the surface angle of the object is steep or when the object is square.

The object detection device according to an aspect of the present disclosure includes a distance detection unit, a position calculation unit, a reflection point estimation unit, a shape determination unit, and a reflection point correction unit. The distance detection unit sequentially transmits an exploration wave to an object present on the side of the vehicle when the vehicle is moving, receives a reflected wave reflected by the exploration wave against the object, and receives the reflected The distance to the object is sequentially detected based on the wave. The position calculation unit calculates a sensor position that is a position of the distance detection unit when the distance of the object is detected. The reflection point estimation unit estimates an arrival direction of the reflected wave using a detection distance history and a sensor position history which are distances detected by the distance detection unit, and moves from the sensor position to the arrival direction. A reflection point that is a point separated by the detection distance is estimated. The shape determination unit determines a history shape of the reflection point near a corner of the object. The reflection point correction unit is the reflection point that does not belong to the steep history shape near the corner when the history shape includes a steep history shape that occurs when the surface angle of the object is steep. The low-precision reflection point, which is the reflection point belonging to the steep history shape, is corrected at a position corresponding to the arrival direction at the normal reflection point or the normal reflection point.

In the object detection device, when the history shape of the reflection point includes a steep history shape, the reflection point (accuracy low reflection point) belonging to the steep history shape is corrected. The arrival direction of the reflected wave at the normal reflection point is similar to the true arrival direction of the reflected wave in the low accuracy range. Since the reflection point correction unit corrects the low accuracy reflection point at a position corresponding to the normal reflection point or the arrival direction of the reflected wave at the normal reflection point, the low accuracy reflection point can be corrected to a position close to the true reflection point. Therefore, even when the surface angle of the object is steep, the position of the corner of the object can be detected with high accuracy.

An object detection device according to another aspect of the present disclosure includes a distance detection unit, a position calculation unit, a reflection point estimation unit, a corner determination unit, and a shape determination unit. The distance detection unit sequentially transmits an exploration wave to an object present on the side of the vehicle when the vehicle is moving, and sequentially receives and receives a reflected wave reflected by the exploration wave hitting the object. The distance to the object is sequentially detected based on the reflected wave. The position calculation unit calculates a sensor position that is a position of the distance detection unit when the distance of the object is detected. The reflection point estimation unit estimates the arrival direction of the reflected wave using the history of the detection distance that is the distance detected by the distance detection unit and the history of the sensor position, and moves from the sensor position to the arrival direction. A reflection point that is a point separated by the detection distance is estimated. The corner determination unit determines a corner of the object based on the history of the reflection points. The shape determination unit determines a history shape near the corner of the object. The corner determining unit determines the position of the corner based on the detection distance history instead of the reflection point history when the history shape is a straight line.

According to the object detection apparatus, when the history shape is a straight line, the position of the corner is determined based on the history of the detection distance instead of the history of the reflection point. Therefore, in the case of a square object, the corner detection accuracy can be improved as compared with the case of using the reflection point history. In addition, the history shape is not intended to mean only a complete straight line, but also to include a shape similar to a straight line.

The above and other objects, configurations, and advantages of the present disclosure will become more apparent from the following detailed description with reference to the following drawings. In the drawing
FIG. 1 is a block diagram showing a schematic configuration of the object detection apparatus. FIG. 2 is a diagram illustrating an example of a detection situation of a parking space. FIG. 3 is a flowchart of parking space detection processing in the first to fifth embodiments. FIG. 4 is a diagram for explaining a method of calculating the arrival direction of the reflected wave and the reflection point. FIG. 5 is a diagram in which the triangle of FIG. 4 is extracted. FIG. 6 is a detailed flowchart of S16 of FIG. 3 in the first embodiment. FIG. 7 is a diagram illustrating a scene where the distance to the parked vehicle is sequentially detected. FIG. 8 is an enlarged view of part A of FIG. 7 and is a diagram for explaining that the history of reflection points is folded around the corner. FIG. 9 is an enlarged view of a portion B in FIG. 7 and shows a state in which the history of reflection points is folded around the corner. FIG. 10 is a diagram showing a detection scene near the corner of the second parked vehicle. FIG. 11 is an enlarged view of a portion C in FIG. 10 and shows a state in which the history of reflection points is folded around the corner. FIG. 12 is a detailed flowchart of S19 of FIG. 3 in the first embodiment. FIG. 13 is a diagram for explaining a reflection point correction method according to the first embodiment. FIG. 14 is a diagram for explaining a method of invalidating an excessively corrected reflection point. FIG. 15 is a detailed flowchart of S22 of FIG. FIG. 16 is an enlarged view of a portion B in FIG. 7 and is a diagram illustrating a method of adding a reflection point when the correction of the reflection point is insufficient. FIG. 17 is a detailed flowchart of S23 of FIG. FIG. 18 is a diagram showing a parabola that approximates the history of distance measurement points. FIG. 19 is a diagram showing a spline curve that approximates the history of distance measurement points. FIG. 20 is a flowchart following S18: No in FIG. FIG. 21 is an enlarged view of part B of FIG. 7 and is a diagram for explaining the processing of FIG. FIG. 22 is an enlarged view of a portion B in FIG. 7 in the second embodiment, and shows a state of a history of reflection points near the corner. FIG. 23 is a detailed flowchart of S16 of FIG. 3 in the second embodiment. FIG. 24 is a diagram for explaining the relationship between the incident angle α of the exploration wave and the reflection angle β of the reflected wave. FIG. 25 is a diagram for explaining a reflection point correction method according to the third embodiment. FIG. 26 is a detailed flowchart of S20 of FIG. 3 in the third embodiment. FIG. 27 is a detailed flowchart of S20 of FIG. 3 in the fourth embodiment. FIG. 28 is an enlarged view of a portion B in FIG. 7 and shows a state of correction of the reflection point in the fourth embodiment. FIG. 29 is a detailed flowchart of S20 of FIG. 3 in the fifth embodiment. FIG. 30 is an enlarged view of a portion B in FIG. 7, and shows a state of correction of the reflection point in the fifth embodiment. FIG. 31 is a diagram showing a detection scene of a parked vehicle as an object having a round corner. FIG. 32 is a diagram showing a detection situation of a prism as a rectangular object near the corner. FIG. 33 is a flowchart of parking space detection processing in the sixth embodiment. FIG. 34 is a diagram illustrating a detection scene of a prism that is inclined obliquely with respect to the moving direction of the vehicle. FIG. 35 is a flowchart of parking space detection processing in the seventh embodiment.

Prior to describing the embodiments of the present disclosure, findings by the present inventors will be described. The estimation of the reflection point based on the principle of triangulation is based on the assumption that the exploration wave from the distance measuring sensor is reflected at the same point of the object between adjacent sensor positions. In a situation where the assumption is greatly deviated, the accuracy of the estimated reflection point decreases. This will be described by taking the vicinity of a corner of a parked vehicle as an example. FIG. 7 shows a scene in which the distance sensor 2 sequentially detects the distance to the parked vehicle 7 when the vehicle 6 is moving along the side path 100 of the parked vehicle 7 in the direction P1. In FIG. 7, the vehicle 6 at a certain time point t1 is indicated by reference numeral “6a”, the distance measurement sensor 2 is indicated by reference numeral “2a”, and the detection range 20 of the distance measurement sensor 2 is indicated by reference numeral “20a”. The vehicle 6 is indicated by reference numeral “6b”, the distance measuring sensor 2 is indicated by reference numeral “2b”, the detection range 20 of the distance measuring sensor 2 is indicated by reference numeral “20b”, and the vehicle 6 at the next time point t3 is indicated by reference numeral “6c”. The distance sensor 2 is indicated by reference numeral “2c”, and the detection range 20 of the distance measuring sensor 2 is indicated by reference numeral “20c”. FIG. 8 is an enlarged view of part A in FIG. 7 and shows the vicinity of the corner 7 a of the parked vehicle 7.

As shown in FIG. 8, since the true reflection point 3a at the time point t1 where the surface angle is gentle and the true reflection point 3b at the time point t2 are close, the above assumption is almost satisfied. Therefore, the reflection point 4a estimated from the sensor positions 2a and 2b and the detection distances La and Lb at the time points t1 and t2 substantially coincides with the true reflection points 3a and 3b. The reflection point 4a is an intersection of an arc 5a having the sensor position 2a as the center and a detection distance La as the radius, and an arc 5b having the sensor position 2b as the center and the detection distance Lb as the radius. On the other hand, at the time points t2 and t3 when the surface angle of the corner 7a becomes steep, the change amount (Lc−Lb) of the detection distance becomes larger than the movement amount d1 of the distance measuring sensor 2. For this reason, the above assumption is broken, and the reflection point 4b estimated from the sensor positions 2b and 2c and the detection distances Lb and Lc at the time points t2 and t3 is far away from the true reflection point 3c. The reflection point 4b is an intersection of an arc 5b centered on the sensor position 2b and a radius of the detection distance Lb and an arc 5c centered on the sensor position 2c and the radius of the detection distance Lc.

Thus, in the case of a round shape such as the vicinity of a corner of a parked vehicle and a steep surface angle of an object, the principle of triangulation is not established, and the accuracy of the estimated reflection point decreases. When the accuracy of the reflection point decreases, the corner detection accuracy decreases.

On the other hand, as shown in FIG. 32, in the case of detecting a rectangular object 8 such as a prism, the corner 8a of which is not rounded, the right and left side surfaces 82 of the object are not directed to the distance measuring sensor 2 side. Even if the exploration wave 20 transmitted from the sensor 2 hits the left and right side surfaces 82, the reflected wave does not return to the distance measuring sensor 2. Therefore, the history range of the detection distance (the history range of the distance measuring point 9) substantially matches the range of the object 8 (the range of the front surface 81 of the object 8). In this case, if the position of the corner 8a of the object 8 is detected based on the reflection point, the detection accuracy of the corner 8a is lowered. This is because there is an error in the detection distance, and the reflection point estimated based on the detection distance having the error is easily detected at a position shifted from the true reflection point. In addition, when the width of the object 8 is small, there may be a case where the number of data of the detection distance necessary for estimating the reflection point cannot be obtained and the corner 8a cannot be detected.
(First embodiment)
Hereinafter, a first embodiment of an object detection device according to the present disclosure will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an object detection device 1 of the present embodiment. The object detection device 1 is mounted on a vehicle 6 (see FIG. 2). The object detection device 1 includes a distance measuring sensor 2, a vehicle speed sensor 12, a steering angle sensor 13, and an ECU 11 connected thereto. The distance measuring sensor 2 is attached to the side surface (left side surface, right side surface) of the vehicle 6 toward the side of the vehicle 6, for example. The distance measuring sensor 2 transmits an exploration wave such as an ultrasonic wave at predetermined intervals (for example, every 100 milliseconds) in the front direction of the distance measuring sensor 2 (side of the vehicle 6) based on an instruction from the ECU 11. . The distance measuring sensor 2 receives the reflected wave reflected by the transmitted exploration wave hitting an object existing on the side of the vehicle. The distance measuring sensor 2 calculates the distance to the object based on the transmission timing of the exploration wave and the reception timing of the reflected wave. Detection information (detection distance) detected by the distance measuring sensor 2 is input to the ECU 11. Note that the ECU 11 may calculate the detection distance. The distance measuring sensor 2 may be any sensor that transmits an exploration wave and receives a reflected wave of the exploration wave, and uses a radio wave regardless of whether it uses a sound wave or a light wave. May be. As the distance measuring sensor 2, for example, a sensor such as an ultrasonic sensor, a laser radar, or a millimeter wave radar can be used.

FIG. 2 illustrates an object detection range 20 (search wave transmission range) by the distance measuring sensor 2. The directivity φ of the detection range 20 is, for example, about 70 to 120 degrees. The maximum detection distance that can be detected by the distance measuring sensor 2 (the distance between the tip of the detection range 20 and the distance measuring sensor 2) is, for example, about 4 m to 10 m. The center line of the detection range 20 (the front direction of the distance measuring sensor 2) is, for example, substantially parallel to the vehicle width direction (left-right direction) of the vehicle 6. Note that the center line may be inclined to the vehicle movement direction side or the opposite side up to, for example, about 20 degrees with respect to the vehicle width direction.

The vehicle speed sensor 12 is a sensor that detects the vehicle speed of the vehicle 6. Detection information (vehicle speed) detected by the vehicle speed sensor 12 is input to the ECU 11. The steering angle sensor 13 is a sensor that detects the steering angle of the steering of the vehicle 6. Detection information (steering angle) detected by the steering angle sensor 13 is input to the ECU 11.

The ECU 11 is composed mainly of a microcomputer composed of a CPU, a ROM, a RAM, and the like, and supports parking of the vehicle 6 based on each detection information input from the distance measuring sensor 2, the vehicle speed sensor 12, and the steering angle sensor 13. Perform various processes. Specifically, the ECU 11 detects the position of an object such as a parked vehicle that exists on the side of the vehicle 6 (for example, the position of a corner), and determines whether there is a parking space based on the detected position of the parking space. Execute the process. Moreover, ECU11 performs the automatic parking process which parks the vehicle 6 automatically in the parking space detected by the parking space detection process.

First, with reference to FIG. 2, a basic concept of a parking space detection method by a parking space detection process executed by the ECU 11 will be described. FIG. 2 is a diagram illustrating an example of a detection situation of a parking space. Specifically, in FIG. 2, while the vehicle 6 moves along the side path 100 of two parked vehicles 71 and 72 parked in parallel (hereinafter, reference numeral 7 is used when the parked vehicles 71 and 72 are not distinguished), The scene which detects the parking space 101 between these parked vehicles 71 and 72 is shown. In FIG. 2, the vehicle 6 is moving in the direction of the arrow P <b> 1 (the direction from the bottom to the top in FIG. 2).

When the ECU 11 tries to detect the parking space, the ECU 11 instructs the distance measuring sensor 2 to sequentially detect the distances to the objects (parked vehicles 71 and 72) existing on the side of the vehicle. At this time, if the vehicle 6 is moving, a history of detection distances corresponding to each position of the vehicle 6 (ranging sensor 2) is obtained. In FIG. 2, the history of the detection distance is indicated by reference numeral 9 (dots). Specifically, each detection distance 9 is the detection distance only in the front direction of the distance measurement sensor 2 (side of the vehicle 6) from the position of the distance measurement sensor 2 (hereinafter referred to as the sensor position) when the detection distance is detected. It is a distant point (hereinafter referred to as a distance measuring point). Since the detection range 20 has a certain extent, the distance detection by the distance measuring sensor 2 is started before the vehicle 6 reaches the front position of the parked vehicle 7, and for a while after the vehicle 6 passes the parked vehicle 7. During this period, distance detection by the distance measuring sensor 2 continues. That is, as shown in FIG. 2, the history range of the distance measuring point 9 is wider than the width of the parked vehicle 7. The distance measuring points 900 plotted outside the parked vehicle 7 are plotted in the depth direction from the surface of the parked vehicle 7 by the distance the distance measurement sensor 2 is away from the parked vehicle 7. As a result, the history shape of the distance measuring point 9 is substantially parabolic.

Thus, since the history range of the distance measuring point 9 is wider than the width of the parked vehicle 7, when attempting to detect the parking space based on the history of the distance measuring point 9, a parking space narrower than the actual parking space 101 is obtained. It will be detected. Therefore, the ECU 11 estimates the reflection point of the reflected wave on the parked vehicle 7 based on the principle of triangulation using the history of the detection distance and the history of the sensor position. The details of the reflection point estimation method will be described later. FIG. 2 shows the history of the estimated reflection point 4 (black dot). The reflection point 4 is plotted at a position substantially coinciding with the surface of the parked vehicle 7. Therefore, for example, a space between the reflection point 410 at the end of the first parked vehicle 71 and the reflection point 420 at the end of the second parked vehicle 72 can be detected as a parking space. Thereby, the parking space of the width | variety equivalent to the actual parking space 101 is detectable. The above is the basic concept of the parking space detection method.

However, as described above, in the vicinity of the corner of the parked vehicle, the plane angle of the parked vehicle becomes large, which may fall outside the scope of the triangulation principle. That is, as shown in FIG. 8, the reflection point 4 b estimated based on the reflected wave (detection distance) from the reflection surface having a large surface angle is greatly separated from the true reflection point 3. Specifically, the reflection point 4b is detected in a direction P3 opposite to the history direction P2 of the true reflection point 3. In the example of FIG. 8, the history direction P2 is an upper left diagonal direction in the paper surface direction of FIG. 8, and the direction P3 from the reflection point 4a to the reflection point 4b is an oblique lower right direction in the paper surface direction of FIG. Yes. FIG. 9 is an enlarged view of a portion B in FIG. 7 and shows a history of the reflection point 4 in the vicinity of the corner 71 a of the parked vehicle 7. As shown in FIG. 9, in the vicinity of the corner 71a, the history direction of the reflection point 4 is as indicated by an arrow P4. That is, the history of the reflection point 4 includes the opposite direction to the moving direction P1 of the vehicle 6 (see FIG. 7) from the end point 421 of the reflection point 42 detected at a position substantially coincident with the parked vehicle 7 and on the side. There is a range 40 (hereinafter referred to as a low accuracy range) that is folded back toward the direction path 100 (see FIG. 7). If the history of the reflection point 4 including the low accuracy range 40 is used, for example, the end point 421 is detected as a corner of the parked vehicle 7. In this case, the width of the detected parked vehicle (the width between corners) is narrower than the actual parked vehicle 7. Therefore, the ECU 11 corrects the reflection point 41 in the low accuracy range 40 (hereinafter referred to as the low accuracy reflection point).

Hereinafter, the details of the parking space detection process executed by the ECU 11 will be described, including the process for correcting the low accuracy reflection point. FIG. 3 shows a flowchart of the parking space detection process. The processing in FIG. 3 is started when a switch (not shown) for instructing detection of the parking space is operated by a passenger of the vehicle 6, for example. During execution of the processing of FIG. 3, the ECU 11 instructs the distance measuring sensor 2 to sequentially detect the distance to the object (parked vehicle 7) that exists on the side of the vehicle 6. When the processing of FIG. 3 is started, first, various parameters used in the subsequent processing are initialized (S11). Specifically, the time t (n) when the distance measuring sensor 2 performs distance detection is set to zero (S11). In addition, the number of times the distance measurement sensor 2 tries to detect the distance (measurement count) n is set to 1 (S11). Further, the measurement count Car1Start (hereinafter referred to as the first vehicle start point) when the detection of the first parked vehicle 71 (see FIG. 2) is started is set to zero (S11). Further, the measurement count Car1End (hereinafter referred to as the first vehicle end point) when the detection of the first parked vehicle 71 is completed is set to zero (S11). Further, the measurement count Car2Start (hereinafter referred to as the second vehicle start point) when the detection of the second vehicle 72 (see FIG. 2) is started is set to zero (S11). Further, the measurement count Car2End (hereinafter referred to as the second vehicle end point) when the detection of the second parked vehicle 72 is completed is set to zero (S11). The ECU 11 measures time t (n) (time based on t (1)) at each measurement count n.

Next, the detection distance L (n) at each measurement count n is acquired from the distance measurement sensor 2 (S12). When the ECU 11 itself is configured to calculate the detection distance L, in S12, the reception timing of the reflected wave is acquired from the distance measurement sensor 2, and the detection distance L (n) is calculated based on the transmission timing and the reception timing. calculate. When no object is present in the detection range 20 (see FIG. 2) of the distance measuring sensor 2, that is, in the case of non-detection, the detection distance L (n) = 0. Next, the sensor position Attd at each measurement count n is calculated (S13). Specifically, as shown in FIG. 2, the position of the distance measuring sensor 2 at the time of starting the processing of FIG. 3 is the origin O, the moving direction of the vehicle 6 at that time is the X axis, and the direction perpendicular to the X axis Set the coordinate system with Y as the Y axis. The sensor position Attd is calculated as coordinates (AttdX (n), AttdY (n)) in the coordinate system. At this time, the moving distance of the vehicle 6 (ranging sensor 2) from the previous sensor position Attd (measurement count n-1) can be calculated from the vehicle speed input from the vehicle speed sensor 12 and the time t (n). From the steering angle input from the steering angle sensor 13, the moving direction of the vehicle 6 (ranging sensor 2) from the previous sensor position Attd can be calculated. The current sensor position Attd is calculated from the moving distance and moving direction. The detected distance L (n) acquired in S12 is stored in a work memory such as a RAM in association with the sensor position Attd (AttdX (n), AttdY (n)).

Thus, as shown in FIG. 2, as the vehicle 6 moves, the history of the distance measuring point 9 (history of the detected distance) can be plotted in the set coordinate system. Next, based on the history of the detection distance (ranging point 9), the range of the first parked vehicle 71 and the range of the second parked vehicle 72 are extracted (S14). Specifically, the first vehicle start point Car1Start and the first vehicle end point Car1End are specified as the range of the parked vehicle 71. For example, in the history of the detection distance, the measurement count when the value of the detection distance changes abruptly is set as the first vehicle start point Car1Start and the first vehicle end point Car1End. In the example of FIG. 2, the measurement count at the distance measuring point 91 is the first vehicle start point Car1Start. Further, the measurement count at the distance measuring point 92 is defined as the first vehicle end point Car1End.

Similarly, the second vehicle start point Car2Start and the second vehicle end point Car2End are specified as the range of the parked vehicle 72. In the example of FIG. 2, the measurement count at the distance measuring point 93 is a second start point Car2Start. In the example of FIG. 2, since the vehicle 6 has not yet passed the second parked vehicle 72, the second vehicle end point Car2End is the current measurement count n. When the distance detection of the second parked vehicle 72 has not yet started (when the vehicle 6 has not yet reached the position of the parked vehicle 72), only the range of the first parked vehicle 71 is displayed in S14. Will be extracted. In this case, Car2Start = 0 and Car2End = 0 are maintained.

Next, by using the history of the detection distance L (n) and the history of the sensor position Attd stored in the work memory, the arrival direction RflDirection (n) of the reflected wave at the measurement count n is calculated according to the principle of triangulation. Calculate (S15). Then, a point away from the sensor position Attd by the detection distance L (n) in the arrival direction RflDirection (n) is calculated as the reflection point Rflt (S15). This reflection point Rflt is calculated as coordinates (RfltX (n), RfltY (n)) in the coordinate system of FIG.

Here, FIG. 4 is a diagram for explaining a method of calculating the arrival direction RfltDirection (n) and the reflection point Rflt (n). Specifically, FIG. 4 shows a position 6b of the vehicle 6 at the measurement count n (shown by a solid line), a position 6a of the vehicle 6 at the previous measurement count n-1 (shown by a broken line), a parked vehicle 7, The figure which looked at from the top is shown. As shown in FIG. 4, the sensor position Attd (n) and the detection distance L (n) at the measurement count n, and the sensor position Attd (n−1) and the detection distance L (n−1) at the measurement count n−1. Consider a triangle 53 consisting of FIG. 5 shows a diagram in which only the triangle 53 is extracted. Based on the coordinate components of the sensor positions Attd (n) and Attd (n−1), the vector components (AttdBktX, AttBktY) of the side 531 between these positions and the absolute value AttDBkt of the vector are calculated. Further, assuming that the angle between the side 531 and the side 532 of the detection distance L (n) is θ (n), the angle θ (n) is calculated by the following equation 1. This angle θ (n) corresponds to the arrival direction RflDirection (n) of the reflected wave.

Then, the angle θ (n), the above vector components (AttdBktX, AttdBktY), and the absolute value AttDBkt are substituted into the following formulas 2 to 6, and the coordinates of the vertex of the triangle 53, that is, the X coordinate RfltX ( n) (Expression 5) and Y coordinate RfltY (n) (Expression 6) are calculated.

As a result, as shown in FIG. 2, the history of the reflection point 4 can be plotted at a position substantially coincident with the surface of the parked vehicle 7 as the vehicle 6 moves. Next, the low accuracy range 40 (see FIG. 9) in which the history of the reflection point 4 is folded is extracted from the history of the reflection point 4 (S16). Here, FIG. 6 shows a detailed flowchart of S16. If it transfers to the process of FIG. 6, it will be judged first whether either the low accuracy range in the 1st parked vehicle 71 or the low accuracy range in the 2nd parked vehicle 72 is extracted (S101). Specifically, for example, it is determined whether the second vehicle start point Car2Start = 0, that is, whether the distance detection of the second parked vehicle 72 has not yet started (S101). When the second vehicle start point Car2Start = 0 (S101: Yes), the low accuracy range of the first vehicle 71 is extracted, and the process proceeds to S102.

In S102, the X coordinate RfltX (Car1End) of the reflection point at the first end point Car1End is smaller than the X coordinate RfltX (Car1End-1) of the reflection point at the previous measurement count (Car1End-1). It is determined whether or not (S102). As shown in FIG. 9, when the low accuracy range 40 exists, in S102, the X coordinate of the reflection point 411 located at the tip of the low accuracy range 40 and the X of the previous reflection point 413 are shown. The coordinates are compared. FIG. 9 shows an X axis similar to the X axis in FIG. If RfltX (Car1End) is larger than RfltX (Car1End-1) (S102: No), it is determined that there is no low accuracy range (S108). In this case, since the vehicle 6 has not yet reached the vicinity of the corner 71a (see FIG. 9) of the parked vehicle 71 or has reached, but the surface angle of the corner 71a is gentle, the reflection point 4 Is assumed to be the same as the moving direction P1 of the vehicle 6. After S108, the process of the flowchart of FIG.

In S102, when RfltX (Car1End) is smaller than RfltX (Car1End-1) (S102: Yes), it is determined that a low accuracy range exists, and the process proceeds to S103. In S103, the target measurement count value j is set to the first vehicle end point Car1End (S103). Next, the X coordinate RfltX (j) of the reflection point at the measurement count j is compared with the X coordinate RfltX (j−1) of the reflection point at the previous measurement count j−1 (S104). When RfltX (j−1) is larger than RfltX (j) (S104: No), the process proceeds to S105, and it is determined whether or not the measurement count j has reached the first start point Car1Start. . If the measurement count j has not yet reached the first start point Car1Start (S105: No), the process proceeds to S106, and the value of the measurement count j is set to the previous value (j = j−1). ). After that, returning to S104, the X coordinate RfltX (j) of the reflection point at the measurement count j to be newly focused, and the X coordinate RfltX (j−1) of the reflection point at the previous measurement count j−1 Compare

In this way, by the processing of S104, S105, and S106, paying attention to two reflection points 4 that are adjacent in order from the reflection point 411 in FIG. 9 in the opposite direction to the history direction P4, the X coordinates of these two reflection points 4 A size comparison is performed. In S104, if RfltX (j) is larger than RfltX (j-1) (S104: Yes), the process proceeds to S107. In S107, a measurement count AreaStart (hereinafter referred to as a range start point) corresponding to the start point of the low accuracy range is set to the measurement count j + 1. Further, a measurement count AreaEnd (hereinafter referred to as a range end point) corresponding to the end point of the low accuracy range is set to the first end point Car1End (S107). Also, a measurement count CorrectEdge (hereinafter referred to as an end point) of a reflection point adjacent to the reflection point at the end of the low accuracy range (reflection point at the range start point) is set as the measurement count j (S107). That is, in S107, the range from the measurement count j + 1 to Car1End is extracted as the low accuracy range. In the example of FIG. 9, the range from the reflection point 412 adjacent to the reflection point 421 positioned just at the turning point to the reflection point 411 is extracted as the low accuracy range 40. Further, the reflection point 421 is extracted as an end point. Hereinafter, the reflection point 4 that is not included in the low accuracy range 40 may be referred to as a normal reflection point, and the normal reflection point is denoted by reference numeral 42. After S107, the process of the flowchart of FIG.

In S105, when the measurement count j reaches the first start point Car1Start (S105: Yes), the process proceeds to S108. In this case, although the history direction of the reflection point is opposite to the moving direction of the vehicle 6, the turning point is not found. Since this situation is an unexpected situation, it is determined that there is no low accuracy range (S108). Thereafter, the process of the flowchart of FIG.

On the other hand, in S101, if the second vehicle start point Car2Start ≠ 0 (S101: No), that is, if detection of the second parked vehicle 72 has already started, the low accuracy range of the parked vehicle 72 is reduced. As extracting, it transfers to S109. Note that the first parked vehicle 71 and the second parked vehicle 72 are detected in the order of (1) low accuracy range → (2) normal reflection point → (3) low accuracy range. In the process of FIG. 6 for the first parked vehicle 71, the accuracy is low while reversing the low accuracy range of (3) from (3) to (2) (reversing the history direction P4 of FIG. 9). A range search is in progress. The vicinity of (3) corresponds to the vicinity of the second corner (back side corner) through which the vehicle 6 passes among the two corners on the side path 100 side of the first parked vehicle 71. On the other hand, in the process of FIG. 6 for the second parked vehicle 72, the low accuracy range search is performed while proceeding from (1) to (2) in the vicinity of the low accuracy range of (1). The vicinity of (1) corresponds to the vicinity of the first corner (front side corner) through which the vehicle 6 passes among the two corners on the side path 100 side of the second parked vehicle 72. Here, FIGS. 10 and 11 are diagrams illustrating the history direction of the reflection point in the second parked vehicle 72. Specifically, FIG. 10 is a view similar to FIG. 7, in which the vehicle 6 is a corner 72a of the second parked vehicle 72 (a corner on the first parked vehicle 71 side, a corner adjacent to the parking space 101). ) Shows a scene where a nearby reflection point is being detected. FIG. 11 is an enlarged view of a portion C in FIG. 10 and shows a state of the reflection point 4 in the vicinity of the corner 72a (near (1) → (2)). In FIG. 10, the vehicle 6 is moving in the direction P1 from the right side to the left side in FIG. FIG. 11 shows the direction P1 and the X axis.

As shown in FIG. 11, the history direction of the reflection point 4 in the second parked vehicle 72 is the direction indicated by the arrow P5. That is, first, the reflection point 411 is detected. The reflection point 411 is detected at a position advanced to the moving direction P1 side from the actual position of the corner 72a and a position on the vehicle 6 side. Thereafter, the reflection point 4 is detected from the reflection point 411 in the direction opposite to the moving direction P1 and closer to the corner 72a (upward and diagonally to the right in the paper direction of FIG. 11). Thereafter, the reflection point 4 is detected in the same direction as the movement direction P1 at a position substantially coincident with the surface of the parked vehicle 72 with a certain reflection point 421 as a boundary. Note that the measurement count at the reflection point 411 is the second start point Car2Start.

Based on the above, the processing after S109 in FIG. 6 will be described. In S109, the X coordinate RfltX (Car2Start) of the reflection point at the second start point Car2Start is compared with the X coordinate RfltX (Car2Start + 1) of the reflection point in the next measurement count (Car2Start + 1) (S109). . In the example of FIG. 11, the X coordinates of the reflection point 411 and the next detected reflection point 414 are compared. When RfltX (Car2Start) is smaller than RfltX (Car2Start + 1) (S109: No), it is determined that there is no low accuracy range (S116). In this case, for example, since the surface angle of the corner 72a is gentle, it is assumed that the history of the reflection point 4 does not return. After S116, the process of the flowchart of FIG.

In S109, when RfltX (Car2Start) is larger than RfltX (Car2Start + 1) (S109: Yes), it is determined that the low accuracy range exists, and the process proceeds to S110. In S110, the target measurement count value j is set to the second start point Car2Start (S110). Next, the X coordinate RfltX (j) of the reflection point at the measurement count j is compared with the X coordinate RfltX (j + 1) of the reflection point at the next measurement count j + 1 (S111). When RfltX (j) is larger than RfltX (j + 1) (S111: No), the process proceeds to S112, and it is determined whether or not the measurement count j has reached the second end point Car2End. When the measurement count j has not yet reached the second end point Car2End (S112: No), the process proceeds to S113, and the value of the measurement count j is set to the next value (j = j + 1). Thereafter, returning to S111, the X coordinate RfltX (j) of the reflection point at the newly focused measurement count j is compared with the X coordinate RfltX (j + 1) of the reflection point at the next measurement count j + 1.

In this way, by the processing of S111, S112, and S113, paying attention to two reflection points 4 that are adjacent in order from the reflection point 411 in FIG. 11 in the same direction as the history direction P5, the X coordinates of these two reflection points 4 A size comparison is performed. In S111, when RfltX (j + 1) is larger than RfltX (j) (S111: Yes), the process proceeds to S114. In S114, the range start point AreaStart is set to the second start point Car2Start. Further, the range end point AreaEnd is set to the measurement count j−1 (S114). Further, the end point CorrectEdge is set to the measurement count j (S114). That is, in S114, the range from the measurement count Car2Start to j−1 is extracted as the low accuracy range in the second parked vehicle 72. In the example of FIG. 11, the range from the reflection point 412 adjacent to the reflection point 421 positioned just at the turning point to the reflection point 411 is extracted as the low accuracy range 40. Further, the reflection point 421 is extracted as an end point. After S114, the process of the flowchart of FIG.

In S112, when the measurement count j reaches the second vehicle end point Car2End (S112: Yes), the process proceeds to S115. In this case, although the history direction of the reflection point is opposite to the moving direction of the vehicle 6, the turning point is not found. Since this situation is an unexpected situation, it is determined that there is no low accuracy range (S115). Thereafter, the process of the flowchart of FIG.

3, returning to the description of FIG. 3, after S <b> 16, the process proceeds to S <b> 17 to determine whether there is a low accuracy range. When the low accuracy range does not exist (in the case of S108, S115, and S116 in FIG. 6, S17: No), the process proceeds to S25. On the other hand, when the low accuracy range exists (in the case of S107 and S114 in FIG. 6, S17: Yes), the process proceeds to S18, and depending on whether or not the parameter Method set in the ROM is 1, S19 It is determined whether the subsequent processing is executed or the processing of FIG. If Method = 1 (S18: Yes), the subsequent processing is executed in S19. That is, the target measurement count j is set to the range start point AreaStart set in S107 or S114 of FIG. 6 (S19). That is, attention is paid to the reflection point 412 in the example of FIG. 9, and attention is paid to the reflection point 411 in the example of FIG. Next, the position of the reflection point Rflt (j) at the measurement count j is corrected (S20). That is, the position of the reflection point 41 (low accuracy reflection point) in the low accuracy range 40 of FIGS. 9 and 11 is corrected.

Here, FIG. 12 is a detailed flowchart of S20. FIG. 13 is a diagram for explaining the concept of correction by S20, and corresponds to an enlarged view of part A in FIG. In FIG. 13, the reflection point 421 (end point) adjacent to the end of the low accuracy range, the sensor position 21 when the reflection point 421 is detected, the low accuracy reflection point 41 to be corrected, and the low accuracy reflection thereof. The sensor position 22 when the point 41 is detected is illustrated. As a premise for correction, it is assumed that the reflection surface corresponding to the low accuracy range has a surface angle at the reflection point 421. In this case, the arrival direction θ2 of the reflected wave in the low accuracy range can be approximated to be the same as the arrival direction θ1 at the reflection point 421 (the arrival direction RflDirection (CorrectEdge) at the end point CorrectEdge) (θ2 = θ1). Therefore, in S121 of FIG. 12, the position of the reflection point Rflt (j) is corrected using the arrival direction RflDirection (CorrectEdge). More specifically, the arrival direction θ1 at the reflection point 421 (calculated in S15 in FIG. 3), the sensor position 22, and the detection distance L1 at the sensor position 22 are read from the work memory. Then, the low precision reflection point 41 is corrected to the position of the point 431 that is separated from the sensor position 22 in the arrival direction θ1 by the detection distance L1.

Thus, the history direction of the corrected reflection point 431 (corrected reflection point) from the reflection point 421 can be made the same as the moving direction of the vehicle 6. As a result, the corrected reflection point 431 can be arranged near the corner 7a. After S121, the process of the flowchart of FIG.

3, after S20, the process proceeds to S21, and it is determined whether or not the measurement count j has reached the range end point AreaEnd set in S107 or S114 of FIG. If not reached yet (S21: No), the process proceeds to S22, and the measurement count j is updated to the next value (j = j + 1). Thereafter, the process returns to S20 to correct the position of the reflection point Rflt (j) at the updated measurement count j. In this way, by repeating S20, S21, and S22, correction is performed in order from the reflection point 412 to the reflection point 411 in the example of FIG. 9, and in the example of FIG. 11, from the reflection point 411 to the reflection point 412 in order. Correction is performed. That is, all the low precision reflection points 41 included in the low precision range 40 are corrected.

In S21, when the measurement count j reaches the range end point AreaEnd (S21: Yes), the process proceeds to S23. Here, the reason why the process of S23 is performed will be described with reference to FIG. FIG. 14 is an enlarged view of a portion B in FIG. 7 and illustrates the reflection point 4 before and after the correction of S20 in the vicinity of the corner 7a. That is, FIG. 14 shows a state in which the reflection point 41 (reflection point before correction) in the low accuracy range 40 is corrected to the position indicated by reference numeral 43 by the correction of S20. As shown in FIG. 14, by performing the correction of S20, the return of the history of the reflection point 4 can be eliminated. However, if excessive correction is performed, there is a possibility that some of the corrected reflection points 432 out of the correction reflection points 43 may protrude from the corner 7a. Therefore, in S23, the corrected reflection point 432 that may protrude from the corner 7a is determined, and the determined corrected reflection point 432 is invalidated. FIG. 15 is a detailed flowchart of S23.

15, first, it is determined whether to process the first parked vehicle 71 or the second parked vehicle 72 (S131). Specifically, for example, it is determined whether the second vehicle start point Car2Start = 0, that is, whether the distance detection of the second parked vehicle 72 has not yet started (S131). When the second vehicle start point Car2Start = 0 (S131: Yes), the first vehicle parked 71 is assumed to be processed, and the process proceeds to S132.

In S132, the parameter JStart used in the subsequent processing is set to the first start point Car1Start (S132). Further, the parameter JEnd used in the subsequent processing is set to the first vehicle end point Car1End (S132). Hereinafter, the parameter JStart is referred to as a parked vehicle start point, and the parameter JEnd is referred to as a parked vehicle end point.

Next, the target measurement count j is set to the parked vehicle start point JStart (S134). In addition, when the depth direction of the parked vehicle is the Y-axis direction in FIG. 2, the Y coordinate MinRfltY (hereinafter simply referred to as the minimum value) of the reflection point having the smallest position in the depth direction is represented by the reflection point at the parked vehicle start point JStart. The Y coordinate RfltY (JStart) is set (S134). Further, the Y coordinate MaxRfltY (hereinafter simply referred to as the maximum value) of the reflection point having the maximum position in the depth direction is set to the Y coordinate RfltY (JStart) of the reflection point at the parked vehicle start point JStart (S134).

Next, it is determined whether or not the Y coordinate RfltY (j) of the reflection point in the target measurement count j is smaller than the minimum value MinRfltY (S135). If it is smaller (S135: Yes), the minimum value MinRfltY is updated to the RfltY (j) (S136). Thereafter, the process proceeds to S137. When RfltY (j) is larger than MinRfltY (S135: No), S136 is skipped and the process proceeds to S137. In this case, the current minimum value MinRfltY is maintained.

In S137, it is determined whether RfltY (j) is larger than the maximum value MaxRfltY (S137). When it is larger (S137: Yes), the maximum value MaxRfltY is updated to the RfltY (j) (S138). Thereafter, the process proceeds to S139. When RfltY (j) is smaller than MaxRfltY (S137: No), S138 is skipped and the process proceeds to S139. In this case, the current maximum value MaxRfltY is maintained.

In S139, it is determined whether or not the measurement count j has reached the parked vehicle end point JEnd (S139). If not reached yet (S139: No), the process proceeds to S140, and the measurement count j is updated to the next value (J = J + 1). Thereafter, the process returns to S135, and the processes of S135 to S139 described above are executed for the updated measurement count j. In this manner, in S135 to S140, the reflection point having the minimum position in the depth direction (Y coordinate) and the maximum reflection point are extracted from the reflection points from the parked vehicle start point JStart to the parked vehicle end point JEnd. Yes.

In S139, when the measurement count j reaches the parked vehicle end point JEnd (S139: Yes), the process proceeds to S141. Then, it is determined whether or not the difference between the maximum value MaxRfltY and the minimum value MinRfltY (MaxRfltY−MinRfltY) is greater than a predetermined threshold th1 (S141). In the example of FIG. 14, in S141, it is determined whether or not the difference ΔY1 in the Y-axis direction between the reflection point 441 having the smallest position in the depth direction and the reflection point 442 having the largest position is larger than the threshold th1. . In the example of FIG. 14, the difference ΔY1 is larger than the threshold th1.

In S141, when the difference (MaxRfltY−MinRfltY) is smaller than the threshold th1 (S141: No), it is determined that there is no overcorrected reflection point, and the process of the flowchart in FIG. On the other hand, when the difference (MaxRfltY−MinRfltY) is larger than the threshold th1 (S141: Yes), it is determined that there is an overcorrected reflection point, and the processes after S142 are performed. That is, in S142, the target measurement count j is reset to the parked vehicle start point JStart (S142). Next, it is determined whether or not the difference (RfltY (j) −MinRfltY) between the Y coordinate RfltY (j) of the reflection point at the measurement count j and the minimum value MinRfltY is larger than the threshold th1 (S143). If it is larger (S143: Yes), the reflection point Rflt (j) is invalidated as an overcorrection reflection point (S144). Thereafter, the process proceeds to S145. On the other hand, when the difference (RfltY (j) −MinRfltY) is smaller than the threshold th1 (S143: No), S144 is skipped and the process proceeds to S145.

In S145, it is determined whether or not the measurement count j has reached the parked vehicle end point JEnd. If not reached yet (S145: No), the process proceeds to S146, and the measurement count j is updated to the next value. Thereafter, the process returns to S143, and the processes of S143 to S145 described above are executed for the updated measurement count j. Referring to the example of FIG. 14, in S142 to S146, whether each reflection point 4 exceeds the threshold value th1 line 700 from the reflection point 4 at the parked vehicle start point JStart to the reflection point 4 at the parked vehicle end point in order. Determine whether. The reflection point 4 exceeding the line 700 is invalidated. In the example of FIG. 14, the reflection point 432 is invalidated.

In S145, when the measurement count j reaches the parked vehicle end point JEnd (S145: Yes), the process of the flowchart in FIG.

On the other hand, in S131, when the second vehicle start point Car2Start ≠ 0 (S131: No), the processing for the second parking vehicle 72 is performed, and the process proceeds to S147. Then, the parked vehicle start point JStart is set to the second vehicle start point Car2Start (S147). Further, the parked vehicle end point JEnd is set to the second vehicle end point Car2End (S147). Thereafter, in the processes after S134 described above, the presence / absence of an overcorrected reflection point is determined. If there is an overcorrected reflection point, the overcorrected reflection point is invalidated.

As described above, in FIG. 15, the overcorrected reflection point is determined based on the distance from the minimum reflection point in the Y-axis direction. This is because the value of the X coordinate increases as the reflection point is further away from the minimum reflection point in the Y-axis direction and may protrude from the corner 7a. By determining the overcorrected reflection point using the Y coordinate, it is possible to determine the reflection point that may protrude from the corner 7a with high accuracy regardless of the width of the parked vehicle in the X-axis direction.

Returning to the explanation of FIG. 3, after S23, the process proceeds to S24. Here, the reason why the process of S24 is performed will be described with reference to FIG. FIG. 16 is an enlarged view of a portion B in FIG. 7 and illustrates various points such as the reflection point 4 near the corner 7a. The reflection point 4 (illustrated by a black circle) in FIG. 16 includes the corrected reflection point corrected in S20. As shown in FIG. 16, even after the correction is performed, the end point 442 of the history of the reflection point 4 may not reach the end surface of the corner 7a. This is because distance detection could not be performed satisfactorily because the detection range of the distance measuring sensor 2 was small or the wind was strong. In other words, this is because the number of detection distance data used for calculating the reflection point 4 is small. FIG. 16 shows the history of the distance measuring point 9 (illustrated by a white triangle), but the distance measuring point 9 does not reach the end face of the corner 7a (usually, as shown in FIG. The history of the distance point 9 protrudes from the parked vehicle 7). Therefore, in S24, it is determined whether or not there is a possibility of reaching the end face of the corner 7a (determining whether or not the correction amount in S20 is insufficient). If the correction is insufficient, an additional reflection point is calculated so as to reach the end face of the corner 7a. FIG. 17 is a detailed flowchart of S24.

When the process proceeds to FIG. 17, first, for example, by determining whether or not the second start point Car2Start = 0, whether the process is performed on the first parked vehicle 71 or the second parked vehicle 72. It is determined whether or not processing is to be performed (S151). When the second vehicle start point Car2Start = 0 (S151: Yes), the first vehicle parked 71 is assumed to be processed, and the process proceeds to S152.

In S152, various parameters used in the subsequent processes are set (S152). Specifically, the parameter JStart1 (hereinafter referred to as a curve approximation start point) is set as the first vehicle start point Car1Start (S152). In addition, the parameter JEnd1 (hereinafter referred to as a curve approximation end point) is set to the first end point Car1End (S152). Also, the parameter JStart2 (hereinafter referred to as the distance approximation start point) is set to the measurement count (Car1End + 1) immediately after the first end point Car1End (S152). In addition, the parameter JEnd2 (hereinafter referred to as the distance approximation end point) is set to the measurement count (Car2Start-1) immediately before the second unit start point Car2Start (S152). Further, the parameter dJ is set to 1 (S152).

Next, it is determined whether or not the difference (MaxRfltY−MinRfltY) between the maximum value MaxRfltY and the minimum value MinRfltY calculated in the process of FIG. 15 is smaller than a predetermined threshold th2 (S153). This threshold th2 may be the same as the threshold th1 used in FIGS. 14 and 15, or may be a value smaller than the threshold th1. In the example of FIG. 16, in S153, it is determined whether or not the difference ΔY2 in the Y-axis direction between the reflection point 441 having the minimum position in the depth direction and the reflection point 442 having the maximum depth is smaller than the threshold th2. . In the example of FIG. 16, the difference ΔY2 is smaller than the threshold th2.

In S153, when the difference (MaxRfltY−MinRfltY) is larger than the threshold th2 (S153: No), the process of the flowchart in FIG. On the other hand, when the difference (MaxRfltY−MinRfltY) is smaller than the threshold th2 (S153: Yes), it is determined that the correction is insufficient, and the processing after S154 is performed. That is, in S154, it is determined whether or not the parameter Mode indicating the type of approximate curve is set to “1” (S154). Whether Mode = 1 is preset in the ROM of the ECU 11.

When Mode = 1 (S154: Yes), the process proceeds to S155, and a parabola (quadratic curve) is selected as an approximate curve. On the other hand, if Mode ≠ 1 (S154: No), the process proceeds to S156, and a spline curve is selected as the approximate curve. A spline curve is a smooth curve that passes through a plurality of given points, and is a curve that uses individual polynomials for each section sandwiched between adjacent points. After S155 or S156, the process proceeds to S157. In S157, an approximate curve for the history of the detected distance (ranging point) from the curve approximation start point JStart1 (= Car1Start) to the curve approximation end point JEnd1 (= Car1End) is calculated (S157). At this time, the approximate curve of the type selected in S155 or S156 is calculated (S157). When a parabola is selected in S155, a parabola 301 approximating the history of the distance measuring point 9 is obtained as shown in FIG. As shown in FIG. 2, the history shape of the distance measuring point 9 at the normal time is substantially a parabolic shape. Therefore, by approximating the history of the distance measuring point 9 with a parabola, a curve similar to the normal history shape of FIG. 2 can be obtained.

If a spline curve is selected in S156, a spline curve 302 approximating the history of the distance measuring point 9 is obtained as shown in FIG. In FIG. 19, the distance measuring points 94 used for curve approximation are indicated by solid triangles, and the other distance measuring points 9 are indicated by dashed triangles. Thus, by approximating with the spline curve, even if the history shape of the distance measuring point 9 is a shape away from the parabolic shape, an accurate approximate curve reflecting each distance measuring point 9 can be obtained. it can. Note that all of the distance measuring points 9 may be used for curve approximation, or some of the distance measuring points 94 may be used as shown in FIG. FIG. 16 illustrates the approximate curve 300 (parabola 301 in FIG. 18 or spline curve 302 in FIG. 19) obtained in S157.

Next, the target measurement count j is set to the distance approximation start point JStart2 (S158). Next, an approximate value ApproximateL (j) of the detection distance at the measurement count j is calculated using the approximate curve calculated in S157 (S159). Specifically, when the approximate curve is considered to be an X function ApproximateL (X) with respect to the detected distance, the X coordinate AttdX (j) of the sensor position at the measurement count j calculated in S13 of FIG. 3 is obtained from the work memory. read out. Then, the AttdX (j) is substituted for X of the function ApproximateL (X). Thereby, an approximate value of the detection distance at the sensor position AttdX (j) can be obtained. FIG. 16 illustrates an estimated distance measuring point 96 obtained from the approximate curve 300.

Next, based on the principle of triangulation (by the above formulas 1 to 6), the reflection point is calculated using the approximate value ApproximateL (j) of the detection distance and the adjacent detection distance (or approximate value) and their sensor positions. Rflt (j) is calculated (S160). In FIG. 16, the reflection point 45 calculated in S160 is shown by a white circle.

Next, it is determined whether or not the difference (RfltY (j) −MinRfltY) between the Y coordinate RfltY (j) of the reflection point Rflt (j) calculated in S160 and the minimum value MinRfltY is larger than the threshold th2 (S161). If it is smaller (S161: No), the process proceeds to S162, and it is determined whether or not the measurement count j has reached the distance approximation end point JEnd2. If not reached yet (S162: No), the parameter dJ (= 1) is added to the measurement count j, and the measurement count j is updated (S163). Thereafter, the process returns to S159, and a reflection point is added to the updated measurement count j (S159, S160).

In this manner, by repeating the processes of S159 to S163, the reflection point is repeatedly added until the difference between the maximum reflection point in the depth direction (Y-axis direction) and the minimum reflection point in the Y-axis direction exceeds the threshold th2. Done. In S161, when the difference (RfltY (j) −MinRfltY) is larger than the threshold th2 (S161: Yes), the processing of the flowchart of FIG. 17 ends. In S162, when the measurement count j reaches the distance approximation end point JEnd2 (S162: Yes), the process of the flowchart in FIG.

On the other hand, in S151, when the second vehicle start point Car2Start ≠ 0 (S151: No), the processing for the second parked vehicle 72 is performed and the process proceeds to S164. In S164, various parameters used in the subsequent processes are set. Specifically, the curve approximation start point JStart1 is set to the second vehicle start point Car2Start (S164). Further, the curve approximation end point JEnd1 is set to the second vehicle end point Car2End (S164). Further, the distance approximation start point JStart2 is set to the measurement count (Car2Start-1) immediately before the second start point Car2Start (S164). Further, the distance approximation end point JEnd2 is set to the measurement count (Car1End + 1) immediately after the first end point Car1End (S164). Further, the parameter dJ is set to −1 (S164).

Thereafter, it is determined whether or not the correction is insufficient in the processes after S153 (S153). If the correction is insufficient, the reflection point is added until the correction is resolved (until S161 becomes Yes). At this time, since JStart2 = Car2Start−1, JEnd2 = Car1End + 1, and dJ = −1, the reflection point is added in the section between the first parked vehicle 71 and the second parked vehicle 72. It is performed from the parked vehicle 72 side toward the parked vehicle 71 side.

17 is performed, the history of the reflection point 4 can be extended to the vicinity of the end face of the corner 7a as shown in FIG. Further, as in the process of FIG. 15, since insufficient correction is determined based on the Y coordinate of the reflection point, it is possible to determine insufficient correction with high accuracy regardless of the width of the parked vehicle in the X-axis direction.

Returning to the explanation of FIG. 3, after S24, the process proceeds to S25. In S25, the positions of the corner of the first parked vehicle 71 and the corner of the second parked vehicle 72 are estimated based on the history of reflection points (S25). Specifically, for example, the reflection point 410 (see FIG. 2) located at the end of the history of the reflection point 4 with respect to the first parked vehicle 71 is estimated as the position of the corner 71a of the parked vehicle 71. Similarly, the reflection point 420 (see FIG. 2) located at the end of the history of the reflection point 4 for the second parked vehicle 72 is estimated as the position of the corner 72a of the parked vehicle 72. And the space between the estimated corners is set as a parking space (S25).

Next, it is determined whether or not there is an instruction to end the process of FIG. 3 (S26). For example, when the width of the parking space detected in S25 is compared with the width of the vehicle 6 and it is determined that the vehicle 6 can be parked in the parking space, there is an end instruction. When there is no end instruction (S26: No), the process proceeds to S27, and the measurement count n is updated to the next value (n = n + 1). Thereafter, the process returns to S12, and the processes after S12 are performed on the updated measurement count n. In S26, when there is an end instruction (S26: Yes), the process of the flowchart of FIG. 3 is ended.

As described above, according to the present embodiment, the low accuracy range in which the history of the reflection point is folded is extracted, and when Method = 1 in S18, the reflection point in the low accuracy range is corrected. Therefore, it is possible to eliminate the reflection of the reflection point history. Since the corner of the parked vehicle is estimated based on the corrected reflection point history, it is possible to improve the estimation accuracy of the corner having a steep surface angle. As a result, the detection accuracy of a parking space sandwiched between two parked vehicles can be improved.

On the other hand, if Method ≠ 1 in S18, the process proceeds to the flowchart of FIG. The processing of FIG. 20 will be described with reference to FIG. FIG. 21 is an enlarged view of a portion B in FIG. When the processing proceeds to FIG. 20, first, a parabola A that approximates the history shape of the detection distance (ranging point) near the corner with respect to the history of the X coordinate of the sensor position is calculated. FIG. 21 illustrates an approximate parabola 501 for the history of the distance measuring point 9 calculated in S311.

Next, the detection distance L1 at the vertex of the parabola A calculated at S311 and the detection distance L2 at the point on the parabola A that is a predetermined amount (for example, about 1 mm to 1 cm) away from the vertex are calculated (S312). If the process of S312 is demonstrated with reference to FIG. 21, the vertex 502 of the parabola 501 will be extracted. At this time, the Y coordinate of the vertex 502 is the approximate value L1 of the detection distance, and the X coordinate is the X coordinate of the sensor position 503 with respect to L1. However, the sensor position 503 is not stored because it is not the sensor position at the moment when the distance is actually detected. Therefore, it is necessary to estimate and obtain the Y coordinate of the sensor position 503. Therefore, the sensor position 505 closest to the X coordinate of the sensor position 503 is extracted from the history 50 stored as the X coordinate of the sensor position. Further, a sensor position 506 adjacent to the sensor position 505 is extracted in the direction of the sensor position 503. The Y coordinate of the sensor position 503 on the straight line connecting the Y coordinates of the sensor position 505 and the sensor position 506 is obtained. Further, an approximate value L2 of the detection distance corresponding to the sensor position 505 is obtained from the X coordinate of the sensor position 505 and the parabola A.

Next, using the detection distances L1 and L2 calculated in S312 and the sensor positions corresponding to the detection distances, a point B determined by triangulation (the above formulas 1 to 6) is calculated as a vertex reflection point (S313). . In the example of FIG. 21, the intersection 47 of an arc (not shown) centered on the sensor position 503 with a detection distance L1 as a radius and an arc 47 (not shown) centered on the sensor position 505 and a detection distance L2 as a radius is reflected at the vertex. Calculate as a point.

Next, a parabola C passing through the reflection point Rflt (CorrectEdge) of the end point CorrectEdge (calculated in S107 or S114 in FIG. 6) is calculated using the vertex reflection point B calculated in S313 as a vertex (S314). In the example of FIG. 21, the parabola 510 passing through the reflection point 421 is calculated with the vertex reflection point 47 as the vertex.

Next, a point where the difference in the depth direction (Y direction) from the vertex reflection point B is a predetermined threshold th3 is determined as a corner (S315). This threshold th3 corresponds to the threshold th1 (see FIG. 14) used in the processing of S23 in FIG. 3 and the threshold th2 (see FIG. 16) used in the processing of S24. In the example of FIG. 21, a point 48 that is separated from the vertex reflection point 47 by a threshold th3 in the depth direction is determined as a corner position.

Next, a space between the corner of the first parking vehicle 71 determined in S315 and the corner of the second parking vehicle 72 is determined as a parking space (S316). Thereafter, the process of FIG. 20 is terminated, and the process proceeds to S25 of FIG.

As described above, when Method ≠ 1 in S18, the corner is determined directly without passing through the low-reflective point correction, overcorrection, and undercorrection processing (S20, S23, S24). Therefore, the correction of the low reflection point, the overcorrection, and the undercorrection process can be omitted, and the calculation load can be reduced. Further, since there are only two reflection points used for corner detection, the vertex reflection point 47 and the reflection point 421, calculation of the reflection point between the vertex reflection point 47 and the reflection point 421 can be omitted.

(Second Embodiment)
Next, a second embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above embodiment. The configuration of the object detection device of the present embodiment is the same as the configuration of FIG. Further, the ECU 11 executes the processing of the flowchart of FIG. At this time, the details of the processing of S16, that is, the extraction method of the low accuracy range is different from the first embodiment.

First, with reference to FIG. 22, the concept of an extraction method with a low accuracy range in this embodiment will be described. FIG. 22 is an enlarged view of a portion B in FIG. 7, and shows a history of the reflection point 4 in the vicinity of the corner 7a. In FIG. 22, the reflection surface 74 on the left side of the line 400 is a reflection surface having a steep surface angle. As described with reference to FIG. 8, since the amount of change in the detection distance with respect to the movement amount of the distance measuring sensor 2 is large on the reflection surface with a steep surface angle, the accuracy of the reflection point estimated by the triangulation principle is deteriorated. . Therefore, as shown in FIG. 22, the reflection point 41 on the left side of the line 400 has a larger error with respect to the reflection surface 74 than the reflection point 4 on the right side. Therefore, in S16, the angle of the reflection surface 73 obtained from the history of the reflection point 4 is calculated. And the range 40 of the reflective point 41 which comprises the reflective surface 731 whose angle becomes more than a threshold value is extracted as a low precision range.

Hereinafter, the details of the processing of S16 will be described with reference to FIG. FIG. 23 is a detailed flowchart of S16. 23, first, for example, by determining whether or not the second start point Car2Start = 0, whether to process the first parked vehicle 71 or not To determine whether to perform the process (S171). When the second vehicle start point Car2Start = 0 (S171: Yes), the first parking vehicle 71 is assumed to be processed, and the process proceeds to S172.

In S172, the reflection surface angle RfltAngle (Car1End) at the first vehicle end point Car1End is calculated. Specifically, a reflection surface is obtained that is composed of a reflection point Rflt (Car1End) at the first end point Car1End and a reflection point Rflt (Car1End-1) at the previous measurement count (Car1End-1). Then, the angle of the reflecting surface with respect to the X-axis direction (the moving direction of the vehicle 6) is calculated as RfltAngle (Car1End). In the example of FIG. 22, when the reflection point 411 is the reflection point Rflt (Car1End), the reflection surface 731a composed of the reflection point 411 and the adjacent reflection point 413, and the reference line P6 corresponding to the X-axis direction The angle γ is calculated as the reflection surface angle RfltAngle (Car1End).

Next, it is determined whether or not the reflection surface angle RfltAngle (Car1End) is larger than a predetermined threshold th3 (S173). If it is smaller (S173: No), the process proceeds to S180, and it is determined that there is no low accuracy range. In this case, a situation in which the vehicle 6 has not yet reached the corner or a situation in which the vehicle 6 has reached but the corner surface angle is gentle is assumed. After S180, the process of the flowchart of FIG.

In S173, when the reflection surface angle RfltAngle (Car1End) is larger than the threshold th3 (S173: Yes), the process proceeds to S174. In S174, the target measurement count j is set to the first vehicle end point Car1End. Next, the reflection surface angle RfltAngle (j) at the measurement count j is calculated (S175). Specifically, similar to S172, using the reflection point Rflt (j) at the measurement count j and the reflection point Rflt (j-1) at the previous measurement count j-1, the reflection surface angle RfltAngle (j ) Is calculated.

Next, it is determined whether or not the reflection surface angle RfltAngle (j) is smaller than the threshold th3 (S176). When it is larger (S176: No), the process proceeds to S177, and it is determined whether or not the measurement count j has reached the first start point Car1Start. If not reached yet (S177: No), the process proceeds to S178, and the measurement count j is updated to the previous value (j = j-1). Thereafter, the process returns to S175, and the processes of S175 and S176 described above are performed on the updated measurement count j. As described above, by repeating the processes of S175 to S178, a range of reflection points having a reflection surface angle larger than the threshold th3 is extracted from the first vehicle end point Car1End toward the first vehicle start point Car1Start. .

In S176, when the reflection surface angle RfltAngle (j) is smaller than the threshold th3 (S176: Yes), the process proceeds to S179. Then, the range start point AreaStart (measurement count corresponding to the start point of the low accuracy range) is set to the measurement count j + 1 (S179). Further, the range end point AreaEnd (measurement count corresponding to the end point of the low accuracy range) is set to the first end point Car1End (S179). Further, the end point CorrectEdge is set to the measurement count j (S179). That is, in S179, the range from the measurement count j + 1 to Car1End is extracted as the low accuracy range. In the example of FIG. 22, the range 40 on the left side of the line 400 is extracted as the low accuracy range. After S179, the process of the flowchart in FIG.

In S177, when the measurement count j reaches the first start point Car1Start (S177: Yes), the process proceeds to S180, and it is determined that there is no low accuracy range. This is because the reflection surface angle is larger than the threshold th3 over the entire range from the first vehicle end point Car1End to the first vehicle start point Car1Start, which is an unexpected situation. It is. Thereafter, the process of the flowchart of FIG.

On the other hand, in S171, when the second vehicle start point Car2Start ≠ 0 (S171: No), the processing for the second parked vehicle 72 is performed and the process proceeds to S181. In the processing after S181, similar to the processing in S172 to S180 described above, the range of the reflection point having the reflection surface angle larger than the threshold th3 is extracted as the low accuracy range. At this time, the low accuracy range is extracted from the second vehicle start point Car2Start toward the second vehicle end point Car2End.

That is, in S181, using the reflection point Rflt (Car2Start) at the second start point Car2Start and the reflection point Rflt (Car2Start + 1) at the next measurement count (Car2Start + 1), the reflection surface angle at the second start point Car2Start RfltAngle (Car2Start) is calculated.

Next, it is determined whether or not the reflection surface angle RfltAngle (Car2Start) is larger than the threshold th3 (S182). When it is smaller (S182: No), the process proceeds to S189, and it is determined that there is no low accuracy range. Thereafter, the process of the flowchart of FIG. When the reflection surface angle RfltAngle (Car2Start) is larger than the threshold th3 (S182: Yes), the process proceeds to S183. In S183, the target measurement count j is set to the second start point Car2Start. Next, similarly to S175, the reflection surface angle RfltAngle (j) at the measurement count j is calculated (S184). Next, it is determined whether or not the reflection surface angle RfltAngle (j) is smaller than the threshold th3 (S185). When it is larger (S185: No), the process proceeds to S186, and it is determined whether or not the measurement count j has reached the second vehicle end point Car2End. If not reached yet (S186: No), the process proceeds to S187, and the measurement count j is updated to the next value (j = j + 1). Thereafter, the process returns to S184, and the processes of S184 and S185 described above are performed on the updated measurement count j.

When the reflection surface angle RfltAngle (Car2Start) is smaller than the threshold th3 (S185: Yes), the process proceeds to S188. Then, the range start point AreaStart is set to the second vehicle start point Car2Start, and the range end point AreaEnd is set to the measurement count j−1 (S188). Further, the end point CorrectEdge is set to the measurement count j (S179). Thereafter, the process of the flowchart of FIG. In S186, when the measurement count j reaches the second end point Car2End (S186: Yes), the process proceeds to S189, and it is determined that there is no low accuracy range. Thereafter, the process of the flowchart of FIG.

Returning to the description of FIG. 3, when the low accuracy range is extracted in S16 (S17: Yes), the reflection point in the low accuracy range is corrected when Method = 1 in S18 as in the first embodiment (S18). S19 to S22). At this time, each reflection point 41 in the low accuracy range 40 is corrected using the arrival direction of the reflected wave at the reflection point 421 (reflection point adjacent to the end of the low accuracy range 40) in FIG. Other processes in FIG. 3 are the same as those in the first embodiment.

As described above, according to this embodiment, the same effect as that of the first embodiment can be obtained. Furthermore, in the present embodiment, since the range of the reflection point having a surface angle larger than the threshold is extracted as the low accuracy range, the reflection point is not generated or has not occurred before the reflection point history occurs. Can be corrected.

(Third embodiment)
Next, a third embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above embodiment. The configuration of the object detection apparatus of the present embodiment is the same as the configuration of FIG. Further, the ECU 11 executes the processing of the flowchart of FIG. At this time, the details of the processing of S20, that is, the correction method of the low accuracy reflection point is different from the above embodiment.

First, with reference to FIG. 24 and FIG. 25, the concept of the method of correcting the low reflection point accuracy in this embodiment will be described. As shown in FIG. 24, it is assumed that the exploration wave 231 transmitted from the distance measuring sensor 2 is incident on the plane 500 at an incident angle α. At this time, according to the law of reflection, the reflection angle β of the reflected wave 232 is equal to the incident angle α. Therefore, the distance measuring sensor 2 theoretically cannot receive the reflected wave 232 unless it is positioned in the direction of the reflection angle β, but in reality, it is reflected when the incident angle α is small to some extent (when the surface angle is gentle). Wave 232 can be received. However, when the incident angle α is large (when the surface angle is steep), the distance measuring sensor 2 cannot receive the reflected wave 232. In the present embodiment, near the corner of the parked vehicle, a minute plane having a limit surface angle at which the distance measuring sensor 2 can receive the reflected wave is continuous, and has a convex curved surface as a whole. Assume.

This assumption and the concept of correction based on this assumption will be further described with reference to FIG. FIG. 25 shows sensor positions 241, 242, and 243 of the distance measuring sensor at three consecutive measurement counts j−2, j−1, and j, and reflection points 422 and 423 detected at the sensor positions 241 and 242. Yes. These reflection points 422 and 423 are normal reflection points that do not require correction or have been corrected. Note that the reflection point 41 (correction target reflection point) at the sensor position 243 (target sensor position) is a low precision reflection point, and FIG. 25 shows a scene where the reflection point 41 is corrected. Assuming that the distance measuring sensor is at the sensor position 242 (adjacent sensor position), the distance measuring sensor receives the reflected wave reflected by the reflecting surface 732 (adjacent reflecting surface) composed of the reflection points 422 and 423. Can be considered. Under the above assumption, the reflection surface 732 has a limit surface angle at which the distance measuring sensor at the sensor position 242 can receive the reflected wave. Under the above assumption, it can be considered that the reflection surface 733 having a limit surface angle at which the sensor position 242 can receive the reflected wave is formed so as to be continuous with the reflection surface 732. The reflecting surfaces 732 and 733 form a convex curved surface. If the distance measuring sensor is at the sensor position 243, it can be considered that the distance measuring sensor has received the reflected wave reflected by the reflecting surface 733. That is, it can be considered that the true reflection point at the sensor position 243 is on the reflection surface 733. Therefore, in the present embodiment, the reflection surface 733 is estimated, and the low-precision reflection point 41 is corrected so as to be on the reflection surface 733.

Hereinafter, details of the correction method of the present embodiment will be described with reference to FIG. FIG. 26 is a detailed flowchart of S20 of FIG. When the process proceeds to FIG. 26, first, the normal direction of the reflection surface RfltAngle (j−1) at the measurement count j−1 is calculated (S191). Specifically, in the example of FIG. 25, a straight line 732 (reflection surface) connecting the reflection point 423 (adjacent reflection point) at the measurement count j-1 and the reflection point 422 at the measurement count j-2 is represented. Set as RfltAngle (j-1). Then, a straight line 151 perpendicular to the reflecting surface 732 is set as a normal line of Rflt Angle (j−1).

Next, an angle dAngle formed by a straight line Line1 (j-1) connecting the reflection point Rflt (j-1) and the sensor position Attd (j-1) at the measurement count j-1 and the normal calculated in S191. (J-1) is calculated (S192). In the example of FIG. 25, an angle δ formed by a straight line 152 connecting the reflection point 423 and the sensor position 242 and a normal line 151 is calculated.

Next, a straight line Line2 (j−1) in which the angle formed with the straight line Line1 (j−1) is −dAngle (j−1) is calculated (S193). In the example of FIG. 25, a straight line 153 having an angle of −δ with the straight line 152 is calculated.

Next, a straight line that passes through the reflection point Rflt (j-1) and is perpendicular to the straight line Line2 (j-1) is calculated as a corrected reflection surface CorrectLine (j) (S194). In the example of FIG. 25, a straight line 733 that passes through the reflection point 423 and is perpendicular to the straight line 153 is calculated as the corrected reflection surface CorrectLine (j). Under the above assumption, the reflection surface RfltAngle (j−1) and the correction reflection surface CorrectLine (j) are the reflection surfaces having the limit surface angle at which the reflected wave can be received at the sensor position Attd (j−1). . In the measurement count j, it can be considered that the reflected wave returns from the corrected reflection surface CorrectLine (j).

Next, an accurate low reflection point Rflt (j) at the intersection of the corrected reflection surface CorrectLine (j) and an arc whose center is the sensor position Attd (j) at the measurement count j and whose radius is the detection distance L (j). Is corrected (S195). In the example of FIG. 25, the low-precision reflection point 41 is corrected to the intersection 433 of the straight line 733 and the arc 154 having the sensor position 243 as the center and the detection distance L (j) as the radius. After S195, the process of the flowchart of FIG.

Returning to the description of FIG. 3, when the measurement count j has not yet reached the range end point AreaEnd in S21 (S21: No), the measurement count j is updated to the next value (S22). Then, the process of S20 is performed on the updated measurement count j. At this time, the corrected reflection point corrected in the previous processing of S20 is set as the reflection point Rflt (j-1) in the measurement count j-1. That is, in the example of FIG. 25, the corrected reflection point 433 is set as the reflection point Rflt (j−1). Then, the reflection surface composed of the corrected reflection point 433 and the previous reflection point 423 is defined as a reflection surface RfltAngle (j−1), and based on the reflection surface RfltAngle (j−1), the current measurement count j The corrected reflection surface CorrectLine (j) is calculated. The other processes in FIG. 3 are the same as those in the first and second embodiments.

As described above, according to this embodiment, the same effects as those of the first and second embodiments can be obtained. Further, in the present embodiment, when the reflection point is corrected, the next accuracy low reflection point is corrected using the corrected reflection point. The normal reflection point closest to the low reflection point can be reflected in the correction.

(Fourth embodiment)
Next, a fourth embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above embodiment. The configuration of the object detection apparatus of the present embodiment is the same as the configuration of FIG. Further, the ECU 11 executes the processing of the flowchart of FIG. At this time, the details of the processing of S20, that is, the correction method of the low accuracy reflection point is different from the above embodiment. Note that the correction method of the present embodiment is suitable for correcting the reflection point 41 in the low accuracy range 40 (see FIG. 9) in the first embodiment. That is, it is suitable for correcting the reflection point in the return range when the return point history is turned back.

The correction method of this embodiment will be described with reference to FIGS. FIG. 27 is a detailed flowchart of S20. FIG. 28 is an enlarged view of a portion B in FIG. 7 and shows a state of the reflection point 4 near the corner 7a. FIG. 28 illustrates a range 40 (low accuracy range) in which the history of the reflection point 4 is folded, and a corrected reflection point 434 after correcting the reflection point 41 in the low accuracy range 40.

When the processing proceeds to FIG. 27, the accuracy low reflection point Rflt (j) is a point that is symmetrical with the accuracy low reflection point Rflt (j) at the measurement count j with respect to the reflection point Rflt (CorrectEdge) at the end point CollectEdge. Is corrected (S201). In the example of FIG. 28, the normal reflection point 421 adjacent to the end of the low accuracy range 40 is defined as a reflection point Rflt (CorrectEdge), and each accuracy low reflection point 41 is symmetric with respect to the normal reflection point 421. The reflection point 41 is corrected. Specifically, for example, the low accuracy reflection point 41a is corrected to the position of the point 434a, the low accuracy reflection point 41b is corrected to the position of the point 434b, and the low accuracy reflection point 41c is corrected to the position of the point 434c. Thereby, the return of the history of the reflection point 4 can be eliminated, and the history of the reflection point 4 can be extended to the vicinity of the end face of the corner 7a. The other processes in FIG. 3 are the same as in the above embodiment.

As described above, according to this embodiment, the same effect as that of the above embodiment can be obtained. Furthermore, it is only necessary to calculate a point-symmetric point, so that the calculation load can be reduced.

(Fifth embodiment)
Next, a fifth embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above embodiments. The configuration of the object detection apparatus of the present embodiment is the same as the configuration of FIG. Further, the ECU 11 executes the processing of the flowchart of FIG. At this time, the details of the processing of S20, that is, the correction method of the low accuracy reflection point is different from the above embodiment.

The correction method of this embodiment will be described with reference to FIGS. FIG. 29 is a detailed flowchart of S20. FIG. 30 is an enlarged view of a portion B in FIG. 7 and shows a state of the reflection point 4 near the corner 7a. When shifting to the processing of FIG. 29, the accuracy low reflection point at the measurement count j is set to a point offset from the reflection point Rflt (correctEdge) at the end point CorrectEdge by a predetermined distance d in the moving direction of the vehicle or the opposite direction thereof. Rflt (j) is corrected (S211). Note that the moving direction of the vehicle in S211 may be the moving direction of the vehicle (that is, the X-axis direction) at the measurement count n = 1, or may be the moving direction of the vehicle at the current time. At this time, as shown in FIG. 30, when the end surface 7a1 of the corner 7a is oriented in the X-axis direction (vehicle movement direction) (when processing is performed on the first parked vehicle), the vehicle is offset in the vehicle movement direction. . On the other hand, as shown in FIG. 11, when the end surface of the corner 72a faces in the direction opposite to the X-axis direction (the direction opposite to the moving direction of the vehicle) (when processing the second parked vehicle), Offset in the direction opposite to the direction of movement. In the example of FIG. 30, the low-accuracy reflection point 41 is corrected to a point 435 that is offset from the normal reflection point 421 as an end point by a distance d in the X-axis direction. In addition, it is only necessary to collect some experimental values of the distance between the reflection point Rflt (CorrectEdge) and the end face of the corner and set the distance d based on the collected experimental values by an actual vehicle test. As a result, the tip of the history of reflection points can be extended to near the X coordinate of the end face of the corner.

Returning to the description of FIG. 3, when the measurement count j has not yet reached the range end point AreaEnd in S21 (S21: No), the measurement count j is updated to the next value (S22). Then, the process of S20 is performed on the updated measurement count j. At this time, the current low-reflection point Rflt (j) is also corrected to a point 435 in FIG. That is, in this embodiment, all the low accuracy reflection points are corrected to the same position. Note that only one of the low accuracy reflection points may be corrected. The other processes in FIG. 3 are the same as in the above embodiment.

As described above, according to this embodiment, the same effect as that of the above embodiment can be obtained. Furthermore, since it is only necessary to calculate the offset position, the calculation load can be reduced.

(Sixth embodiment)
Next, a sixth embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above embodiments. In the case of a rectangular object, this embodiment is an embodiment corresponding to the disclosure in which the position of the corner of the object is determined based on the history of the detection distance (ranging point) instead of the history of the reflection point. The configuration of the object detection apparatus of the present embodiment is the same as the configuration of FIG. However, the parking space detection process (object detection process) executed by the ECU 11 is different from the above embodiment. Hereinafter, the parking space detection process (object detection process) of this embodiment will be described with reference to FIGS. FIG. 31 shows a detection scene of the parked vehicle 7 as an object having a round corner. In addition, in FIG. 31, although the detection scene of the parking vehicle 7 of parallel parking is shown, this embodiment is applicable also to the detection scene of the parking vehicle of parallel parking. FIG. 32 shows a detection scene of the prism 8 (square prism) as an object having a square shape (bent substantially at a right angle) near the corner. In FIG. 32, the vehicle 6 is moving substantially parallel to the front surface 81 of the prism 8. FIG. 33 is a flowchart of the parking space detection process of the present embodiment. ECU11 performs the process of FIG. 33 instead of the process of FIG. The processing of FIG. 33 is started when a switch (not shown) for instructing detection of a parking space is operated by an occupant of the vehicle 6, for example.

33. When the process of FIG. 33 is started, first, various parameters used in the subsequent processes are initialized (S31). Specifically, similarly to S11 of FIG. 3, the time t (n) is set to zero, and the measurement count n is set to 1 (S31). Further, an object flag FlagObj indicating whether the object currently detected is a rectangular object or other object (hereinafter referred to as a round object) is set to a value NonRound indicating that the object is a rectangular object. (S31).

Next, as in S12 of FIG. 3, the detection distance L (n) at the measurement count n is acquired from the distance measuring sensor 2 (S32). Next, as in S13 of FIG. 3, the sensor position Attd at the measurement count n is calculated (S33). Next, a distance measuring point Dtct that is separated from the sensor position at the measurement count n by the detection distance L (n) in the front direction of the distance measuring sensor 2 (side of the vehicle 6) is calculated (S34). The distance measuring point Dtct has the position of the distance measuring sensor 2 at the time of starting the processing of FIG. 33 as the origin O, the moving direction of the vehicle 6 at that time as the X axis, and the direction perpendicular to the X axis as the Y axis. Calculated as coordinates (DtctX (n), DtctY (n)) in the coordinate system (see FIGS. 31 and 32). Accordingly, as the measurement count n is updated in S42 described later (as the vehicle 6 moves), as shown in FIGS. 31 and 32, the surfaces of the objects 7 and 8 existing on the side of the vehicle 6 are displayed. The history (point sequence) of the distance measuring point 9 can be plotted along 7b and 81.

As shown in FIG. 31, when the object to be detected is a parked vehicle 7, the vicinity of the corner 7a is rounded and faces the distance measuring sensor 2 side. Therefore, the reflected wave from the vicinity of the corner 7a is received by the distance measuring sensor 2 for a while after the vehicle 6 (the distance measuring sensor 2) passes through the parked vehicle 7. As a result, there is a distance measuring point 973 that is plotted outside the corner 7a and closer to the depth side than the side surface 7b of the parked vehicle 7. Accordingly, in the vicinity of the corner 7a, the history shape of the distance measuring point 9 is a curved line by the distance measuring point 973 and the distance measuring point 974 plotted along the side surface 7b. On the other hand, in the case of the prism 8 of FIG. 32, as described above, the history range of the distance measuring point 9 substantially matches the range of the front surface 81 of the prism 8. Therefore, in the vicinity of the corner 8a of the prism 8, the history shape of the distance measuring point 9 is linear. In the processing after S35, the method for determining the corner position is changed depending on whether the history shape of the distance measuring point (detected distance) near the corner is curved or linear.

That is, in S35, it is determined whether or not the object flag FlagObj is set to NonRound. Since the initial value of the object flag FlagObj is set to NonRound in S31, it is determined in S35 that the object flag FlagObj is set to NonRound for the first time (S35: Yes). Next, it is determined whether the history shape of the distance measuring points near the corner is curved or linear (S36). Specifically, for example, in a distance measurement point history, a point at which the distance measurement point changes abruptly is an end point, and a history of distance measurement points included in a range near the end point (for example, a range of 1 to 2 m from the end point). Determine the shape. At this time, for example, it is determined which can better approximate the distance measurement point history when the distance measurement point history is approximated by a curve or when the distance measurement point approximation is approximated by a straight line. If the approximation with a curve can be approximated better, it is determined that the history shape of the distance measuring point is a curve. On the other hand, when the approximation with a straight line can be better approximated, it is determined that the history shape of the distance measuring point is a straight line.

In S36, when the history shape of the distance measuring point is a curve (S36: Yes), the process proceeds to S37, and the object flag FlagObj is set to a value Round indicating that the object is a round object. Next, in the same manner as S15 in FIG. 3, using the history of the detection distance L and the history of the sensor position Attd, the reflection point Rflt (RfltX (n ), RfltY (n)) is calculated (S38). Next, the end point RfltEdge of the history of the reflection point Rflt is determined as the position of the corner of the object (S39). In the example of FIG. 31, the end point 46 of the history of the reflection point 4 is determined as the position of the corner 7 a of the parked vehicle 7. Thereby, it is possible to bring the end point 971 of the history of the distance measuring point 9 closer to the actual position of the corner 7a than when the position of the corner 7a is determined.

Next, as in S26 of FIG. 3, it is determined whether there is an instruction to end the processing of FIG. 33 (S41). When there is no end instruction (S41: No), the process proceeds to S42, and the measurement count n is updated to the next value (n = n + 1). Thereafter, returning to S32, the above-described S32 to S41 are executed for the updated measurement count n. At this time, since the object flag FlagObj is set to Round in the previous processing of S37, it is determined that the object flag FlagObj is set to Round in S35 of this time (S35: No). Next, the process proceeds to S38, and the reflection point is calculated. That is, when it is determined that the history shape of the distance measuring point is once curved, the history shape is not determined thereafter, and the reflection point is sequentially calculated as the measurement count n is updated.

On the other hand, when the history shape of the distance measuring points is a straight line in S36 (S36: No), the process proceeds to S40. In S40, the reflection point is not calculated, and the end point DtctEdge of the distance measurement point history is determined as the position of the corner of the object. In the example of FIG. 32, the end point 972 of the history of the distance measuring point 9 is determined as the position of the corner 8a of the prism 8. Thereby, it can be brought closer to the actual position of the corner 8a than when the position of the corner 8a is determined at the reflection point.

Next, the process proceeds to S41, and it is determined whether or not there is an instruction to end the process of FIG. When there is no end instruction (S41: No), the process proceeds to S42, and the measurement count n is updated to the next value (n = n + 1). Thereafter, returning to S32, the above-described S32 to S41 are executed for the updated measurement count n. At this time, since the history shape is determined again in S36, even if it is determined that the history shape is linear until the previous time, if it is determined that the current history shape is curved, The determination method can be switched from a method using a distance measurement point history to a method using a reflection point history. Until this time, it is assumed that the vehicle 6 has not yet reached the vicinity of the corner 7a of the parked vehicle 7, and this time the vehicle 6 has reached the vicinity of the corner 7a.

Thereafter, when there is an end instruction (S41: Yes), the processing of the flowchart of FIG. 33 is ended.

As described above, in the present embodiment, when detecting a rectangular object, the position of the corner is determined from the history of the distance measurement points without calculating the reflection point, so that the calculation load can be reduced, Corner detection accuracy can be improved. As a result, even when the parking space is divided by a prism, such as an indoor parking lot such as a multilevel parking lot and an underground parking lot, the parking space can be detected with high accuracy.

(Seventh embodiment)
Next, a seventh embodiment of the object detection device according to the present disclosure will be described with a focus on differences from the above-described embodiment. The present embodiment is an embodiment related to the sixth embodiment. The configuration of the object detection apparatus of the present embodiment is the same as the configuration of FIG. However, the parking space detection process (object detection process) executed by the ECU 11 is partly different from the sixth embodiment. Hereinafter, the parking space detection process (object detection process) of this embodiment will be described with reference to FIGS. 32, 34, and 35. FIG. 34 shows a detection scene for the prism 8 that is inclined obliquely with respect to the moving direction P <b> 1 of the vehicle 6. FIG. 35 is a flowchart of the parking space detection process executed by the ECU 11 of the present embodiment. In FIG. 35, the same processes as those in FIG. 33 are denoted by the same reference numerals. The process of FIG. 35 differs from the process of FIG. 33 in the case where S36 is No, and is otherwise the same. In the following, among the processes in FIG. 35, the description of the same processes as in FIG. 33 is omitted, and different processes are described.

In S36, when the history shape of the distance measuring points is linear (S36: No), the process proceeds to S43. In S43, an approximate straight line for the history of the detection distance (ranging point) is calculated using the least square method or the like. FIG. 34 shows an approximate straight line 601 for the history of the distance measuring point 9. Next, the inclination of the rectangular object with respect to the moving direction of the vehicle is calculated based on the approximate straight line calculated in S43 (S44). This moving direction may be the moving direction of the vehicle (that is, the X-axis direction) at the measurement count n = 1, or may be the current moving direction of the vehicle. In the example of FIG. 34, the inclination of the approximate straight line 601 with respect to the moving direction P1 (X-axis direction) of the vehicle 6 is calculated. Next, based on the inclination calculated in S44, it is determined whether or not the object is inclined (S45). Specifically, it is determined that the object is inclined when the inclination calculated in S44 is equal to or greater than a predetermined threshold angle (for example, 10 degrees), and is determined that the object is not inclined when the inclination is less than the threshold angle. . For example, it is determined that the prism 8 in FIG. 32 is not tilted, and the prism 8 in FIG. 34 is determined to be tilted.

In S45, when it is determined that the object is not tilted (S45: No), the process proceeds to S46. In S46, the average value Lavg of the history of the detection distance is calculated. Next, the X coordinate DtctEdgeX at the end point DtctEdge of the distance measuring point history is set as the X coordinate of the corner of the object (S47). The average value Lavg is set as the Y coordinate of the corner of the object (S47). In other words, a point separated from the sensor position Attd at the end point DtctEdge by the average value Lavg in the Y-axis direction is determined as the position of the corner of the object (S47). Thereby, even if there is an error in the detection distance, the influence of the error can be reduced by using the average value Lavg. Therefore, the position of the corner can be detected stably. After S47, the process proceeds to S41.

On the other hand, if it is determined in S45 that the object is tilted (S45: Yes), the process proceeds to S48. In S48, the intersection of the line parallel to the Y axis passing through the end point DtctEdge and the approximate straight line calculated in S43 is determined as the position of the corner of the object. In the example of FIG. 34, the intersection 602 of the line 155 passing through the end point 972 and parallel to the Y axis and the approximate straight line 601 is determined as the corner position of the object 8. Thus, by determining the position of the corner at a point on the approximate straight line, even when the object is tilted, it is possible to detect a stable corner with reduced influence of the detection distance error. After S48, the process proceeds to S41.

As described above, according to the present embodiment, in addition to the effects of the sixth embodiment, the position of the corner can be detected stably regardless of whether the object is tilted or not.

Note that the object detection device according to the present disclosure is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the claims. In the first to fifth embodiments, the example in which the present disclosure is applied to detection of a parked parking vehicle (parking space for parallel parking) has been described. However, a parked vehicle (parking space for parallel parking) The present disclosure may be applied to the detection of. In the flowcharts of the first to fifth embodiments, the process for the first parked vehicle and the process for the second parked vehicle are separated. Each time a reflection point near the corner of the vehicle is detected, the low accuracy range extraction and the low accuracy reflection point correction may be performed.

Further, in the sixth and seventh embodiments, the example of the quadrangular prism is given as the rectangular object, but the present disclosure can be applied to a prism having another shape such as a triangular prism as long as the corner is square. . The present disclosure can also be applied to detection of a rectangular object other than a prism (for example, a rectangular vehicle such as a wall or one box).

Also, the configurations of the first to fifth embodiments and the configurations of the sixth and seventh embodiments may be combined. Specifically, for example, when S31 to S36 in FIG. 33 are executed and the history shape of the distance measuring point is a straight line in S36, the calculation of the reflection point is stopped and the corner of the object is determined from the history of the distance measuring point. Is determined (configuration of the sixth and seventh embodiments). On the other hand, when the history shape of the distance measuring point is curved in S36, the processing after S14 in FIG. 3 is executed to extract the low accuracy range and to correct the extracted reflection point in the low accuracy range. (S18: Yes) or determine the corner directly (S18: No).

In the above-described embodiment, an example in which the present disclosure is applied to a parking space detection scene has been described. However, the present disclosure may be applied to support for passing through narrow roads where obstacles such as other vehicles exist around. In this case, the object detection device of the present disclosure detects surrounding obstacles, and based on the detection result, for example, steering assistance (announcement of how much steering is better or automatic steering) so that it can pass through. I do. Moreover, you may apply this indication to the start assistance from a parking state like when starting at the back from the state of forward-facing parking. Also in this case, similarly to the slip-through support, the object detection device of the present disclosure detects an obstacle (such as a wall) existing in the vehicle start direction, and performs, for example, steering support based on the detection result.

In the above embodiment, the distance measuring sensor 2 can function as a distance detection unit. The vehicle speed sensor 12 and the steering angle sensor 13 can function as a position calculation unit. Further, the ECU 11 can function as a position calculation unit when executing the process of S13. The ECU 11 can function as a reflection point estimation unit when executing the process of S15. The ECU 11 can function as a shape determination unit when executing the processes of S16 and S17. The ECU 11 can function as a reflection point correction unit when executing the processes of S19 to S22.

The ECU 11 can function as a first extraction unit when executing the processes of S101 to S116. The ECU 11 can function as an angle calculation unit when executing the processes of S172, S175, and S184. The ECU 11 can function as a second extraction unit when executing the processes of S171 to S189. The ECU 11 can function as a first correction unit when executing the process of S121. The ECU 11 can function as a reflection surface calculation unit when executing the processes of S191 to S194. The ECU 11 can function as a second correction unit when executing the process of S195. The ECU 11 can function as a third correction unit when executing the process of S201. The ECU 11 can function as a fourth correction unit when executing the process of S211.

The ECU 11 can function as a reflection point invalid part when executing the processes of S131 to S147. The ECU 11 can function as a difference determination unit when executing the process of S153. The ECU 11 can function as a curve calculation unit when executing the processes of S154 to S157. The ECU 11 can function as an estimated distance calculation unit when executing the process of S159. The ECU 11 can function as an additional point estimation unit when executing the processes of S160 to S163.

The ECU 11 can function as a first parabola calculating unit when executing the process of S311. The ECU 11 can function as a vertex reflection point calculation unit when executing the processes of S312 and S313. The ECU 11 can function as a second parabola calculating unit when executing the process of S314. The ECU 11 can function as a corner determination unit when executing the process of S315.

The ECU 11 can function as a position calculation unit when executing the process of S33. The ECU 11 can function as a reflection point estimation unit when executing the process of S38. The ECU 11 can function as a corner determination unit when executing the processes of S39, S40, S45, S46, S47, and S48. The ECU 11 can function as an average value calculation unit when executing the process of S46. The ECU 11 can function as a straight line calculation unit when executing the process of S43. The ECU 11 can function as an inclination determination unit when executing the processes of S44 and S45.

Claims (18)

1. When the vehicle (6) is moving, the exploration wave is sequentially transmitted to the objects (7, 71, 72) existing on the side of the vehicle, and the reflected wave reflected by the exploration wave is received. A distance detector (2) for sequentially detecting the distance to the object based on the received reflected wave;
   A position calculation unit (12, 13, S13) for calculating a sensor position which is a position of the distance detection unit when detecting the distance of the object;
   Using the detection distance history and the sensor position history detected by the distance detection unit, the arrival direction of the reflected wave is estimated, and the distance from the sensor position to the arrival direction by the detection distance A reflection point estimation unit (S15) for estimating the reflection point (4),
   A shape determination unit (S16, S17) for determining the history shape of the reflection point near the corners (7a, 71a, 72a) of the object;
   When the history shape includes a steep history shape (40) that occurs when the surface angle of the object is steep, a normal reflection point that is the reflection point that does not belong to the steep history shape near the corner ( 42) or a reflection point correction unit (S19 to S22) for correcting the low-precision reflection point (41), which is the reflection point belonging to the steep history shape, at a position corresponding to the arrival direction at the normal reflection point;
   An object detection device (1) comprising:
2. The shape determination unit is a range extraction unit that extracts, as the steep history shape, a low accuracy range in which the detection accuracy of the reflection point is estimated to be low from the history of the reflection point. The object detection apparatus described.
3. The range extraction unit includes a first extraction unit (S101 to S116) that extracts, as the low accuracy range, a range in which the history direction of the reflection point is opposite to the moving direction of the distance detection unit. The object detection apparatus according to claim 2.
4. The range extraction unit includes:
   An angle calculation unit (S172, S175, S184) for calculating the surface angle of the object from the history of the reflection points;
   The object detection apparatus according to claim 2, further comprising: a second extraction unit (S171 to S189) that extracts a range of the reflection point that constitutes the surface angle equal to or greater than a predetermined value as the low accuracy range. .
5. The reflection point correction unit sets the arrival direction in the low accuracy range as the end point arrival direction at the end point (421) that is the normal reflection point adjacent to the end of the low accuracy range, and the sensor at the low accuracy reflection point. The first correction unit (S121) for correcting the accuracy low reflection point at a point (431) separated from the position (22) in the arrival direction of the end point by the detection distance is provided. The object detection device according to claim 1.
6. The accuracy low reflection point to be corrected is the correction target reflection point, the sensor position at the correction target reflection point is the target sensor position (243), and the sensor position adjacent to the target sensor position is the adjacent sensor position (242). The normal reflection point at the adjacent sensor position as the adjacent reflection point (423),
   The reflection point correction unit is
   A normal line (151) of the adjacent reflection surface (732), which is a reflection surface composed of the adjacent reflection point and the normal reflection point (422) adjacent thereto, is a straight line connecting the adjacent reflection point and the adjacent sensor position. A reflection surface calculation unit (S191 to S194) for calculating a reflection surface (733) connected to the adjacent reflection point with a line (153) moved symmetrically with respect to (152) as a normal line;
   A second correction unit (S195) that corrects the correction target reflection point at an intersection (433) of the reflection surface calculated by the reflection surface calculation unit and an arc (154) having a radius of the detection distance at the target sensor position. The object detection device according to any one of claims 2 to 4, further comprising:
7. The reflection point correction unit corrects the low accuracy reflection point to a point (434) that is symmetrical with respect to the low accuracy reflection point with respect to the end point (421) that is the normal reflection point adjacent to the end of the low accuracy range. The object detection apparatus according to any one of claims 2 to 4, further comprising a third correction unit (S201).
8. The reflection point correction unit reduces the accuracy to a point (435) offset from the end point (421) which is the normal reflection point adjacent to the end of the low accuracy range by a predetermined value in the moving direction of the vehicle or the opposite direction. The object detection device according to any one of claims 2 to 4, further comprising a fourth correction unit (S211) for correcting the reflection point.
9. Of the corrected reflection point (43) corrected by the reflection point correction unit with the lateral direction of the vehicle as the depth direction of the object, the depth from the reflection point (441) whose position in the depth direction is the smallest The object detection according to any one of claims 1 to 8, further comprising a reflection point invalid portion (S131 to S147) that invalidates the corrected reflection point (432) that is separated by a first threshold or more in the direction. apparatus.
10. The reflection point (441) having the smallest position in the depth direction among the reflection points after the correction by the reflection point correction unit with the side direction of the vehicle as the depth direction of the object. A difference determination unit (S153) for determining whether or not a difference in the depth direction of the maximum reflection point (442) is equal to or less than a second threshold;
    A curve calculation unit (S154 to S157) for calculating an approximate curve (300, 301, 302) for the history of the detection distance;
    An estimated distance calculation unit (S159) that calculates a point on the approximate curve in a range where the detection distance is not detected as an estimated detection distance (96);
    When the difference determination unit determines that the difference is equal to or less than the second threshold, the history of the estimated detection distance and the estimated detection distance until the difference is equal to or greater than the second threshold. The additional point estimation unit (S160 to S163) for estimating an additional reflection point (45) using the sensor position history, comprising: an additional reflection point (45). Object detection device.
11. The object detection device according to claim 10, wherein the approximate curve is a parabola (301).
12. 11. The object detection apparatus according to claim 10, wherein the approximate curve is a spline curve (302).
13. A first parabola calculation unit (S311) that calculates a parabola (501) that approximates the history shape of the detection distance near the corner as a first parabola;
    A vertex reflection point calculation unit (S312 and S313) for calculating the reflection point as a vertex reflection point (47) using the vertex (502) of the first parabola and a point (504) on the first parabola in the vicinity thereof; ,
    A second parabola calculation unit (S314) that calculates, as a second parabola, a parabola (510) passing through the end point (421) that is the normal reflection point adjacent to the end of the low accuracy range, with the apex reflection point as a vertex;
    A corner determining unit that determines the side direction of the vehicle as the depth direction of the object, and determines the point (48) on the second parabola that has a threshold value as a difference in the depth direction from the vertex reflection point as the corner. S315),
    The object detection apparatus according to claim 2, further comprising:
14. When the vehicle (6) is moving, the exploration waves are sequentially transmitted to the objects (7, 8) existing on the side of the vehicle, and the reflected waves reflected by the exploration waves hitting the object are sequentially received, A distance detector (2) for sequentially detecting the distance to the object based on the received reflected wave;
    A position calculation unit (12, 13, S33) for calculating a sensor position which is a position of the distance detection unit when detecting the distance of the object;
    A point at which the reflected wave arrival direction is estimated using the detection distance history detected by the distance detection unit and the sensor position history, and the detection position is separated from the sensor position by the detection distance. A reflection point estimation unit (S38) for estimating a reflection point of
    A corner determination unit (S39) for determining a corner of the object based on the history of the reflection points;
    A shape determination unit (S36) for determining a history shape near the corner (7a, 8a) of the object,
    The corner determination unit (S40, S45, S46) determines the position of the corner based on the history of the detection distance instead of the history of the reflection point when the history shape is a straight line. An object detection device (1).
15. The object detection device according to claim 14, wherein the reflection point estimation unit stops the estimation of the reflection point when the history shape is a straight line.
16. When the history shape is a straight line, the corner determination unit (S40) is configured to move the sensor from the sensor position when the distance measurement sensor detects the detection distance corresponding to the corner to the side of the vehicle. The object detection device according to claim 14 or 15, wherein a point (972) separated by a detection distance is set as the position of the corner (8a).
17. An average value calculation unit (S46) for calculating an average value of the history of the detection distance;
    When the history shape is a straight line, the corner determination unit (S47) may move the sensor from the sensor position when the distance measurement sensor detects the detection distance corresponding to the corner to the side of the vehicle. The object detection device according to claim 14 or 15, wherein a point separated by an average value is set as the position of the corner (8a).
18. A straight line calculation unit (S43) for calculating an approximate straight line (601) for the history of the detection distance;
    An inclination determination unit (S44, S45) for determining whether or not the approximate straight line is inclined with respect to the moving direction of the vehicle,
    The corner determination unit (S48) is a point that is separated from the sensor position to the side of the vehicle by the detection distance when the history shape is linear and the approximate straight line is inclined. The intersection (602) of the line (155) extending in the direction perpendicular to the moving direction of the vehicle through the end point (972) of the distance measuring point history and the approximate straight line is set as the position of the corner (8a). The object detection apparatus according to claim 14 or 15, characterized in that:

Images