JP2019095956A - Gradient change detection device, method and program, and vehicle - Google Patents

Abstract

To provide a gradient change detection device capable of appropriately detecting the presence/absence of a gradient change in a road surface where a vehicle travels.SOLUTION: A gradient change detection device comprises: an image acquisition unit (30) which acquires a picked-up image obtained by capturing an image of a view ahead of a traveling vehicle; a farthest point detection unit (40) which detects the farthest point in a road surface area corresponding to a road surface where the vehicle travels, in the picked-up image; and a gradient change determination unit (60) which determines the presence/absence of a gradient change in the road surface on the basis of a vertical direction position of the farthest point in the picked-up image.SELECTED DRAWING: Figure 6

Description

  The present invention relates to a slope change detection device, method and program for detecting a slope change of a road surface on which a host vehicle is traveling.

  Conventionally, in order to use for automatic vehicle control, various gradient detection devices are used which capture an image of the traveling direction of the own vehicle and detect the gradient of the road surface on which the own vehicle travels from the captured image (for example, patent See references 1 to 6).

JP, 2016-205887, A Japanese Patent Application Laid-Open No. 3-277916 JP 2012-255703 A JP-A-9-325026 JP, 2010-2334, A JP 2008-33781 A

  In order to perform automatic vehicle control properly, it is necessary to appropriately detect the presence or absence of a gradient change on the road surface on which the host vehicle is traveling.

  The present invention has been made in view of the above problems, and an object thereof is to provide a gradient change detection device, method and program capable of appropriately detecting the presence or absence of a gradient change on the road surface on which the vehicle is traveling. It is.

  The present invention adopts the following technical means in order to solve the above problems. The claims and the reference numerals in the parentheses described in this section are an example showing the correspondence with the specific means described in the embodiment described later as one embodiment, and the technical scope of the present invention is limited. It is not something to do.

  In the first embodiment of the present invention, an image acquisition unit (30) for acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and a road surface area (U) corresponding to a road surface on which the host vehicle travels in the captured image. A farthest point detection unit (40) that detects the farthest point (K) of the road, and a slope change determination unit that determines the presence or absence of a slope change on the road based on the vertical position of the 60) and a gradient change detection device.

  In the second embodiment of the present invention, an image acquisition step of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and detecting the farthest point of a road surface area corresponding to the road surface where the host vehicle travels in the captured image. The gradient change detection method includes: a farthest point detection step; and a gradient change determination step of determining presence or absence of a gradient change on the road surface based on the vertical position of the farthest point in the captured image.

  According to a third aspect of the present invention, there is provided an image acquisition function for acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and a farthest surface area of a road surface area equivalent to the road surface on which the host vehicle travels Gradient change detection for realizing a gradient change determination function of determining the presence or absence of a gradient change on the road surface based on the farthest point detection function of detecting a point and the vertical position of the farthest point in the captured image It is a program.

  The fourth embodiment of the present invention is an image acquisition section (30) for acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and a road surface area corresponding to the road surface on which the host vehicle travels in the captured image. A gradient change detection device comprising: a feature acquisition unit (70) for acquiring a feature; and a gradient change determination unit (90) that determines presence or absence of a gradient change on the road surface based on features of the shape of the road surface area. .

  The fifth embodiment of the present invention acquires an image acquisition step of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and acquiring features of a shape of a road surface area corresponding to a road surface on which the host vehicle travels in the captured image. And a gradient change determination step of determining presence or absence of a gradient change on the road surface based on the feature of the shape of the road surface area.

  According to a sixth aspect of the present invention, there is provided an image acquisition function of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle, and a shape of a road surface area corresponding to a road surface where the host vehicle travels in the captured image. It is a gradient change detection program for realizing a feature acquisition function of acquiring a feature and a gradient change determination function of determining presence or absence of a gradient change on the road surface based on the feature of the shape of the road surface area.

  A seventh embodiment of the present invention is a vehicle that alleviates or stops the vehicle line control when there is a gradient change on the road surface on which the host vehicle is traveling.

  According to the present invention, it is possible to appropriately detect the presence or absence of a gradient change on the road surface on which the host vehicle is traveling.

FIG. 7 is a schematic view showing a problem caused by a change in gradient in distance measurement using image recognition in the prior art. FIG. 10 is a schematic view showing a problem caused by a gradient change in white line recognition using image recognition according to the prior art. FIG. 2 is a schematic view showing the principle of gradient change detection according to the first embodiment of the present invention. FIG. 2 is a schematic view showing gradient change detection according to the first embodiment of the present invention. FIG. 5 is a graph showing linear approximation in gradient change detection according to the first embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS The block diagram which shows the gradient change detection system of 1st Embodiment of this invention. FIG. 2 is a flow chart showing a gradient change detection method according to the first embodiment of the present invention. FIG. 7 is a flow chart showing the farthest point detection step of the first embodiment of the present invention. FIG. 5 is a flow chart showing the gradient change determination step of the first embodiment of the present invention. The schematic diagram which shows the principle of the gradient change detection of 2nd Embodiment of this invention. The schematic diagram which shows the gradient change detection of 2nd Embodiment of this invention. The graph which shows the straight line approximation in gradient change detection of a 2nd embodiment of the present invention. The block diagram which shows the gradient change detection system of 2nd Embodiment of this invention. FIG. 7 is a flow chart showing a gradient change detection method according to a second embodiment of the present invention. The flowchart which shows the gradient change determination step of 2nd Embodiment of this invention.

  With reference to FIGS. 1 and 2, the problem in ranging or white line recognition will be described as an example of the problem caused by the presence of a gradient change in prior art image recognition.

The problems that occur in ranging will be described with reference to FIG.
As shown in FIG. 1A, in distance measurement using image recognition in the prior art, the distance from the vehicle to the distance measurement object is estimated based on the lower end position of the distance measurement object. In the estimation, it is assumed that the host vehicle and the object to be measured are on the same plane. For this reason, as shown in FIG. 1 (b), even if there is a gradient on the road surface on which the host vehicle is traveling, if the ranging objects are also on the same plane having the gradient, the distance estimation will be special. Problem does not occur. However, as shown in FIG. 1 (c) and FIG. 1 (d), when the subject vehicle and the object for distance measurement do not exist in the same plane, that is, gradient change P, between the subject vehicle and object for distance measurement. If Q exists, the distance estimation target can not be accurately estimated because the assumption of the distance estimation is not satisfied. As a result, there is a possibility that appropriate control can not be performed in control based on the distance between the host vehicle and the distance measurement target, such as automatic braking control and inter-vehicle distance control.

A problem that occurs in white line recognition will be described with reference to FIG.
When a bird's-eye view image is generated based on image recognition of a captured image, for example, if the lane in which the host vehicle is traveling is a straight lane, the straight lane is reproduced in the bird's-eye view image as shown in FIG. The white line R defining the is linear. However, when a slope change such as a concave or convex portion exists on the road surface on which the vehicle travels, as shown in FIG. 2 (b) to FIG. A white line R defining a lane, which is not reproduced, has a shape or arrangement of curvature different from the real one. As a result, there is a possibility that appropriate control can not be performed in automobile line control based on the result of white line recognition, such as automobile line maintenance control and automobile line deviation prevention control.

  In each of the following embodiments, in order to cope with such a problem or the like, a change in slope on the road surface on which the host vehicle is traveling is detected.

A first embodiment of the present invention will be described with reference to FIGS. 3 to 9.
In the gradient detection apparatus and method of the present embodiment, the presence or absence of the gradient change is determined based on the vertical position of the farthest point of the road surface area in the captured image. Specifically, the gradient change is based on the vertical position of the farthest point of the road surface area with respect to the vertical position of the vanishing point in the captured image, and the time change of the vertical position of the farthest point of the road surface area in the captured image. Determine the presence or absence of

The gradient change detection of the present embodiment will be described with reference to FIGS. 3 to 5.
The principle of the gradient change detection according to this embodiment will be described with reference to FIG.
A focus of expansion (FOE) as an infinite point exists in a captured image captured by the imaging device. In the present embodiment, the vanishing point of the captured image captured in the horizontal plane is particularly defined as a vanishing point, and calibration is performed so that the vanishing point coincides with the image center of the captured image.

  As shown in FIG. 3, when there is a concave portion P where the slope changes on the road surface on which the host vehicle is traveling, the road surface farther than the concave portion P is at a higher position than the plane on which the host vehicle exists. Since the image is taken in, the farthest point K of the road surface area U is located at a position higher than the vanishing point F in the captured image. On the other hand, in the case where the convex portion Q where the slope changes is present on the road surface on which the host vehicle is traveling, the road surface farther than the convex portion Q is at a lower position than the plane on which the host vehicle is present and enters the blind spot. Since the image is not captured, the farthest point K of the road surface area U is located at a position lower than the vanishing point F in the captured image.

  Thus, the vertical position of the farthest point K of the road surface area U in the captured image changes in accordance with the presence or absence of a gradient change on the road surface on which the vehicle is traveling. It is possible to determine the presence or absence of a gradient change based on the vertical position.

  The gradient change detection of the present embodiment will be described in detail with reference to FIGS. 4 and 5.

  As shown in FIG. 4, when the host vehicle travels the road surface, a graph is assumed in which the horizontal axis is time t and the vertical axis is the vertical position h of the farthest point K of the road surface area U in the captured image. When there is a concave portion on the road surface in the traveling direction of the host vehicle, the farthest point K of the road surface area U moves upward from the vanishing point F in accordance with the progression of the host vehicle. It takes a constant vertical position h1 higher than the directional position H. On the other hand, when a convex portion is present on the road surface in the traveling direction of the host vehicle, the farthest point K of the road surface area U moves downward from the vanishing point F in accordance with the progression of the host vehicle It takes a constant vertical position h2 which is lower than the vertical position H of F.

  As described above, when there is a gradient change on the road surface in the traveling direction of the own vehicle, the vertical position H of the vanishing point and the vertical direction position h of the farthest point K of the road surface area U are constant at a predetermined time. There will be a difference of For this reason, when the difference amount indicating the difference between the vertical position h of the farthest point K with the vertical position H of the vanishing point F at a predetermined time exceeds a predetermined threshold, the gradient changes to the road surface on which the vehicle travels. It can be determined that exists.

  Further, when there is a gradient change on the road surface in the traveling direction of the vehicle, a constant change occurs in the vertical position h of the farthest point K of the road surface area U at a predetermined time interval. For this reason, when the amount of change indicating the change in the vertical position h of the farthest point K at a predetermined time interval exceeds a predetermined threshold, it can be determined that there is a slope change on the road surface on which the vehicle travels.

  In this embodiment, with respect to the difference amount Δ of the vertical direction position h of the farthest point with respect to the vertical direction position H of the vanishing point, as shown in FIG. 5A, time-series data of the vertical direction position h of the farthest point The linear approximation h = at + b is performed using the least squares method, and the absolute value | H−b | of the difference H−b between the vertical direction position H of the vanishing point and the intercept b of the approximate straight line is the difference amount Δ. In the present embodiment, the least squares method is used as the method of linear approximation, but various other methods can be used. In a modification of the present embodiment, robust estimation such as Random Sample Consensus (RANSAC), Least Median of Squares (LMedS), or M estimation may be used as a method of linear approximation. The same applies to the linear approximation described below.

  When the difference amount Δ of the vertical direction position h of the farthest point with respect to the vertical direction position H of the vanishing point is used as the determination reference of the presence or absence of the gradient change, time series data is not necessary in principle. Robustness to noise can be obtained by using series data. In the modification of this embodiment, the difference between the vertical position h of the farthest point with respect to the vertical position H of the vanishing point is the vertical direction position h of the farthest point with respect to the vertical position H of the vanishing point at a predetermined time. The absolute value | H−h | of the difference H−h may be used.

  Further, as shown in FIG. 5B, with respect to the change amount δ of the vertical position h of the farthest point at a predetermined time interval, the time series data of the vertical position h of the farthest point is subjected to the least squares method The linear approximation h = at + b is performed, and the absolute value | a | of the inclination a of the approximate straight line is set as the variation δ.

  Here, when the weather is good, etc., when the visibility of the traveling direction of the vehicle is relatively good, when the gradient change of the road surface is relatively far from the vehicle, the farthest point is constant in a relatively short time In order to move to the vertical position, it is appropriate to use the difference amount of the vertical position of the farthest point with respect to the vertical position of the vanishing point as a criterion of the presence or absence of the gradient change. On the other hand, in the case where the view of the traveling direction of the host vehicle is relatively poor, such as when the weather is bad, the farthest point takes a relatively long time and is constant when the host vehicle relatively approaches the gradient change of the road surface. In order to move to the vertical position, it is appropriate to use the amount of change in the vertical position of the farthest point as the determination criterion of the presence or absence of the gradient change.

The gradient change detection system of this embodiment will be described with reference to FIG.
The gradient change detection system according to the present embodiment is a system for detecting a gradient change on the road surface on which the vehicle travels, and includes an imaging device 10 and a gradient change detection device 20 as shown in FIG. The imaging device 10 captures an image of the traveling direction of the host vehicle. In the present embodiment, a monocular camera is used as the imaging device 10. In the gradient detection device 20, the image acquisition unit 30 has a function of acquiring a captured image captured by the imaging device 10. The farthest point detection unit 40 has a function of detecting the farthest point of the road surface area corresponding to the road surface on which the host vehicle travels in the captured image acquired by the image acquisition unit 30. The farthest point position storage unit 50 has a function of storing, as time-series data, the vertical position of the farthest point detected by the farthest point detection unit 40 in the captured image. The gradient change determination unit 60 has a function of determining the presence or absence of the gradient change on the road surface on which the vehicle is traveling based on the time series data of the vertical position of the farthest point stored in the farthest point position storage unit 50. Have.

  The gradient change detection method of this embodiment will be described with reference to FIGS. 7 to 9.

  As shown in FIG. 7, the gradient change detection method of the present embodiment is a method of detecting a gradient change on the road surface on which the host vehicle is traveling, and includes the following steps. The farthest point detection step and the gradient change determination step will be described in more detail below.

Image acquisition step S30
In the image acquisition step S30, an image of the traveling direction of the host vehicle is captured to acquire a captured image.

Farthest point detection step S40
In the farthest point detection step S40, the farthest point of the road surface area corresponding to the road surface on which the host vehicle travels is detected in the captured image acquired in the image acquisition step S30.

Most distant point position storage step S50
In the farthest point position storing step S50, for the farthest point of the road surface area detected in the farthest point detecting step S40, the vertical position in the captured image of the farthest point is stored as time series data.

Gradient change determination step S60
In the gradient change determination step S60, based on the time-series data of the vertical position of the farthest point stored in the farthest point position storage step S50, it is determined whether or not there is a gradient change on the road surface on which the vehicle is traveling.

The farthest point detection step S40 of the present embodiment will be described with reference to FIG.
The farthest point detection step S40 is a step of detecting the farthest point of the road surface area corresponding to the road surface on which the vehicle travels in the captured image obtained by capturing an image of the traveling direction of the vehicle as described above. As shown in FIG. 8, it has the following steps.

Image division step S41
In the image division step S41, the captured image acquired in the image acquisition step S30 is divided into each area. In the present embodiment, segmentation based on deep learning is used to divide a captured image.

Road surface area extraction step S42
In the road surface area extraction step S42, a road surface area corresponding to the road surface on which the vehicle travels is extracted from the plurality of areas of the captured image divided in the image division step S41.

Farthest point extraction step S43
In the farthest point extraction step S43, the farthest point of the road surface area extracted in the road surface area extraction step S42 is detected.

  In addition, it is possible to use various methods as a detection method of the farthest point of a road surface area | region. In the modification of this embodiment, as a method of detecting the farthest point of the road surface area, a method of detecting a horizontal line from a horizontal edge distribution, a method of using an extrusion point of an optical probe, or the like may be used.

The gradient change determination step S60 of this embodiment will be described with reference to FIG.
As described above, the gradient change determination step S60 is a step of determining the presence or absence of the gradient change in the road surface on which the vehicle is traveling based on the time-series data of the vertical position of the farthest point of the road surface region in the captured image. , As shown in FIG. 9, the following steps are included.

Straight line approximation step S61
In the linear approximation step S61, the time series of the vertical position of the farthest point stored in the farthest position storage step S50 in the coordinate system in which the horizontal axis is time t and the vertical axis is the vertical position h of the farthest point. The data is subjected to linear approximation h = at + b using the least squares method, and the inclination a and the intercept b of the approximate straight line are calculated.

Change amount acquisition step S62
In the change amount acquisition step S62, the absolute value | a | of the inclination of the approximate straight line calculated in the straight line approximation step S61 is acquired as the change amount δ of the vertical position of the farthest point.

Difference amount acquisition step S63
In the difference amount acquisition step S63, the difference between the vertical position of the vanishing point and the intercept of the approximate straight line calculated in the straight line approximation step S61 as the difference amount Δ of the vertical position of the farthest point with respect to the vertical position of the vanishing point. Obtain the absolute value | H−b |.

Amount of change judgment step S64
In the change amount determination step S64, it is determined whether the change amount δ = | a | of the vertical direction position of the farthest point acquired in the change amount acquisition step S62 is equal to or greater than a predetermined threshold value Thδ. If it is determined that the change amount δ is equal to or larger than the predetermined threshold value Thδ, the process proceeds to the gradient change existence recognition step S66. On the other hand, when it is determined that the change amount δ is less than the predetermined threshold value Thδ, the process proceeds to the difference amount determination step S65.

Difference amount determination step S65
In the difference amount determination step S65, whether the difference amount Δ = | H−b | of the vertical direction position of the farthest point with respect to the vertical direction position of the vanishing point acquired in the difference amount acquisition step S63 becomes a predetermined threshold ThΔ or more Judge whether or not. If it is determined that the difference amount Δ is equal to or larger than the predetermined threshold value ThΔ, the process proceeds to the gradient change existence determining step S66. On the other hand, when it is determined that the difference amount Δ = | H−b | is less than the predetermined threshold value ThΔ, the process proceeds to the gradient change nonexistence determination step S67.

Gradient change existence recognition step S66
In the gradient change existence determining step S66, it is determined that there is a gradient change on the road surface on which the vehicle travels, when the change amount δ or the difference amount Δ is determined to be equal to or larger than predetermined threshold values Thδ and ThΔ.

Gradient change nonexistence recognition step S67
In the gradient change nonexistence determining step S67, it is determined that there is no gradient change on the road surface on which the vehicle travels, in the case where the change amount δ and the difference amount Δ are determined to be less than predetermined threshold values Thδ and ThΔ.

  Hereinafter, vehicle control using the result of gradient change detection will be described.

  As described with reference to FIG. 1, in distance measurement using image recognition according to the prior art, when there is a gradient change between the vehicle and the distance measurement object, from the vehicle to the distance measurement object The distance can not be accurately estimated, and there is a possibility that appropriate control can not be performed in automatic vehicle control based on the distance between the host vehicle and the distance measurement target, such as automatic braking control and inter-vehicle distance control. On the other hand, when the gradient change is present in the distance measurement using the image recognition using the result of the gradient change detection of this embodiment, the distance measured is corrected. It is possible to take measures such as reducing the reliability and not performing the ranging itself. Moreover, in the automatic vehicle control based on the distance between the own vehicle and the distance measurement object, when there is a gradient change, it is possible to take measures such as correcting or stopping the control.

  As described with reference to FIG. 2, in white line recognition using image recognition according to the prior art, when there is a gradient change on the road surface on which the vehicle is traveling, the white line defining the lane has a curvature different from reality. There is a possibility that appropriate control can not be performed in vehicle line control based on the result of white line recognition, such as vehicle line maintenance control and vehicle line deviation prevention control. On the other hand, in the white line recognition using image recognition using the result of the gradient change detection of the present embodiment, if there is a gradient change, the shape or arrangement of the recognized white line is corrected. It is possible to take measures such as reducing the reliability of recognition and not performing white line recognition itself. Further, in the vehicle line control based on the result of the white line recognition, if there is a gradient change, it is possible to take measures such as relaxing or stopping the control.

  In addition, when there is a gradient change, in the automatic vehicle control, control such as adjusting the amount of torque to achieve desired acceleration performance, changing the setting of the suspension to achieve desired ride comfort, etc. It is possible.

  Also, in the case of determining the position of the vehicle on a map in automatic driving of the vehicle, it is possible to determine the position of the vehicle using a gradient change as a landmark in an area such as a desert where there is no appropriate landmark. It is.

  The gradient change detection device and method of the present embodiment have the following effects.

  In the gradient change detection device and method of the present embodiment, the farthest point of the road surface area corresponding to the road surface on which the host vehicle travels is detected in the captured image obtained by capturing an image of the traveling direction of the host vehicle Based on the vertical position of the farthest point, it is determined whether or not there is a gradient change on the road surface on which the vehicle is traveling. For this reason, it is possible to appropriately detect a change in slope on the road surface on which the host vehicle is traveling.

  In particular, it is possible to detect a gradient change only from a captured image captured by a single-eye camera, and a stereo camera or other sensors such as a radar, a rider, and an acceleration sensor are not required to detect the gradient change. It is possible to realize gradient change detection in

  In addition, it is possible to detect the gradient change only from the vertical position of the farthest point of the road surface area in the captured image, and if there is only a road surface that is a gradient change detection target, white lines, stationary objects, etc. It is possible to perform gradient change detection in a wide range of environments, since it does not require any other objects.

A second embodiment of the present invention will be described with reference to FIGS. 10 to 15.
In the gradient change detection apparatus and method of the present embodiment, the area of the actual road surface area with respect to the area of the reference road surface area assuming that there is no gradient change on the road surface as the feature of the shape of the road surface area in the captured image The ratio is used to determine the presence or absence of a gradient change based on the area ratio. Specifically, the area ratio when there is no gradient change on the road surface is used as a reference value of the area ratio, and the presence or absence of the gradient change is determined based on the area ratio to the reference value. Determine the presence or absence of gradient change.

The gradient change detection of this embodiment will be described with reference to FIGS. 10 to 12.
The principle of the gradient change detection of this embodiment will be described with reference to FIG.
As shown in FIG. 10, when there is no gradient change on the road surface on which the host vehicle is traveling, the road surface area U in the captured image is bisected with the lower side of the captured image as the base and the farthest point K of the road surface area as the vertex. It has a triangular shape. On the other hand, when the concave portion P where the gradient changes is present on the road surface on which the vehicle travels, the road surface area U in the captured image is a bisected side with the lower side of the captured image as the base and the farthest point K of the road surface area as the vertex. In the triangle, it has a shape in which both sides are concave inward. In addition, when the convex portion Q where the slope changes is present on the road surface on which the host vehicle is traveling, the road surface area U in the captured image has two sides with the lower side of the captured image as the base and the farthest point K of the road surface area as the vertex. In the equilateral triangle, both sides are outwardly convex.

  As described above, since the feature of the shape of the road surface area U in the captured image changes in accordance with the presence or absence of the gradient change on the road surface on which the vehicle travels, based on the feature of the shape of the road surface area U in the captured image It is possible to determine the presence or absence of a gradient change.

  The gradient change detection of this embodiment will be described in detail with reference to FIGS. 11 and 12.

  As shown in FIG. 11, in the present embodiment, a road surface area in the case where it is assumed that there is no gradient change in the road surface in the captured image is taken as a reference road surface area W. The reference road surface area W is an isosceles triangle in which the lower side of the captured image is a base and the farthest point of the road surface area is a vertex. Then, as a feature of the shape of the road surface area U, the area ratio of the area of the road surface area U to the area of the reference road surface area W is used. Further, as the reference value of the area ratio, the area ratio 1 in the case where there is no change in slope on the road surface on which the host vehicle is traveling is used.

  When the host vehicle travels on the road surface, a graph is assumed in which the horizontal axis is time t and the vertical axis is area ratio s. When a concave portion exists on the road surface in the traveling direction of the own vehicle, both sides of the shape of the road surface area U with respect to the isosceles triangle which is the shape of the reference road surface region W according to the progression of the own vehicle. The area ratio s decreases from the reference value 1 and then takes a constant value s 1 smaller than the reference value 1 because the area ratio s decreases from the reference value 1 because the area is recessed inwardly and then has a constant shape. On the other hand, when a convex portion is present on the road surface in the traveling direction of the host vehicle, the shape of the road surface area U with respect to the isosceles triangle which is the shape of the reference road surface area W Since both side portions project outward and then become a constant shape, the area ratio s increases from the reference value 1 and then takes a constant value s2 larger than the reference value 1.

  That is, when there is a gradient change on the road surface in the traveling direction of the host vehicle, a fixed difference exists between the area ratio s and the reference value 1 at a predetermined time. Therefore, when the difference between the area ratio s at the predetermined time and the reference value 1 exceeds the predetermined threshold, it can be determined that there is a gradient change on the road surface on which the vehicle travels.

  In addition, when there is a gradient change on the road surface in the traveling direction of the host vehicle, a constant change occurs in the area ratio s at a predetermined time interval. For this reason, when the amount of change in the area ratio s at a predetermined time interval exceeds a predetermined threshold, it can be determined that there is a change in slope on the road surface on which the vehicle travels.

  In this embodiment, as shown in FIG. 12A, the time series data of the area ratio is subjected to linear approximation s = αt + β using the least square method, and the reference value 1 The absolute value | 1−β | of the difference 1−β between the and the segment β of the approximate straight line is taken as the difference amount Λ. As in the first embodiment, when using the difference amount 判定 as the determination criterion of the presence or absence of the gradient change, in principle, time series data is not necessary, but using time series data is robust against noise. Sex is obtained. In the modification of the present embodiment, the absolute value | 1−s | of the difference 1−s between the reference value 1 and the area ratio s at a predetermined time may be used as the difference amount に 対 す る from the reference value of the area ratio.

  Further, as shown in FIG. 12B, with respect to the change amount λ of the area ratio at a predetermined time interval, the time series data of the area ratio is subjected to linear approximation s = αt + β using the least squares method, and the inclination of the approximation straight line The absolute value | α | of α is the change amount λ.

The gradient change detection system of the present embodiment will be described with reference to FIG. The description of the same configuration as that of the first embodiment will be omitted.
As shown in FIG. 13, in the gradient change detection device 22 of the gradient change detection system, the area ratio calculation unit 70 as the feature acquisition unit is configured to set the road surface on which the vehicle travels from the captured image acquired by the image acquisition unit 30. It has a function of extracting the corresponding road surface area and calculating the area ratio of the area of the road surface area to the area of the reference road surface area, which is the road surface area assuming that no gradient exists on the road surface, as a feature of the shape of the road surface area. . The area ratio storage unit 80 has a function of storing the area ratio calculated by the area ratio calculation unit 70 as time series data. The gradient change determination unit 90 has a function of determining the presence or absence of a gradient change on the road surface on which the vehicle travels based on the time series data of the area ratio stored in the area ratio storage unit 70.

  The gradient change detection method of the present embodiment will be described with reference to FIGS. 14 and 15. The description of the same steps as in the first embodiment will be omitted.

  As shown in FIG. 14, the gradient change detection method of the present embodiment has the following steps. The gradient change determination step will be described in more detail below.

Image acquisition step S30
It is the same as that of the first embodiment.

Area ratio calculation step S70
In the area ratio calculation step S70, the captured image acquired in the image acquisition step S30 is divided into each area, and a road surface area corresponding to the road surface on which the vehicle travels is extracted from a plurality of areas of the divided captured image. Furthermore, the area ratio of the area of the road surface area to the area of the reference road surface area when assuming that there is no gradient on the road surface is calculated as the feature quantity of the shape of the road surface area. In the present embodiment, segmentation based on deep learning is used to divide a captured image.

Area ratio storage step S80
In the area ratio storage step S80, the area ratios calculated in the area ratio calculation step S70 are stored as time series data.

Gradient change determination step S90
In the gradient change determination step S90, based on the time series data of the area ratio stored in the area ratio storage step S80, it is determined whether or not there is a gradient change on the road surface on which the vehicle travels.

The gradient change determination step of this embodiment will be described with reference to FIG.
The gradient change determination step is a step of determining the presence or absence of the gradient change on the road surface on which the vehicle travels based on the time series data of the area ratio of the area of the road surface area to the area of the reference road surface area as described above. , As shown in FIG. 15, the following steps are included.

Straight line approximation step S91
In the linear approximation step S91, in the coordinate system where the horizontal axis is time t and the vertical axis is area ratio s, the time series data of the area ratio stored in the area ratio storage step S80 is linearly approximated using the least squares method. Then, the inclination α and the intercept β of the straight line are calculated.

Change amount acquisition step S92
In the change amount acquisition step S92, the absolute value | α | of the inclination calculated in the linear approximation step S91 is acquired as the change amount λ of the area ratio at a predetermined time interval.

Difference amount acquisition step S93
In the difference amount acquisition step S93, the absolute value | 1-β | of the difference between the reference value of the area ratio and the intercept calculated in the linear approximation step S92 is acquired as the difference amount に 対 す る with respect to the reference value of the area ratio.

Amount of change judgment step S94
In the change amount determination step S94, it is determined whether or not the change amount λ of the area ratio acquired in the change amount acquisition step S92 is equal to or more than a predetermined threshold value Thλ. If it is determined that the change amount λ is equal to or greater than the predetermined threshold value Thλ, the process proceeds to the gradient change existence recognition step S96. On the other hand, if it is determined that the change amount λ is less than the predetermined threshold value Thλ, the process proceeds to the difference amount determination step S95.

Difference amount determination step S95
In the difference amount determination step S95, it is determined whether the difference amount Λ of the area ratio with respect to the reference value acquired in the difference amount acquisition step S93 is equal to or greater than a predetermined threshold Th 否. If it is determined that the difference amount Λ is greater than or equal to the predetermined threshold value ThΛ, the process proceeds to the gradient change existence recognition step S96. On the other hand, when it is determined that the difference amount Λ is less than the predetermined threshold value ThΛ, the process proceeds to the gradient change nonexistence determination step S97.

Gradient change existence recognition step S96
In the gradient change existence determining step S96, it is determined that there is a gradient change on the road surface on which the vehicle travels when the change amount λ or the difference amount 判断 is determined to be equal to or larger than predetermined threshold values Thλ, ThΛ.

Gradient change absence recognition step S97
In the gradient change nonexistence determining step S97, it is determined that there is no gradient change on the road surface on which the host vehicle is traveling, when it is determined that the variation amount λ or the difference amount Λ is less than predetermined threshold values Thλ and Th ,.

  The gradient change detection device and method of the present embodiment exhibit the same effects as those of the first embodiment.

  In this embodiment, the area ratio of the area of the road surface area to the area of the reference road surface area when assuming that there is no gradient change on the road surface is used as the feature of the shape of the road surface area used to determine the presence or absence of the gradient change. However, it is possible to use the shape features of various other road surface areas.

  As a modified example of the present embodiment, it is possible to use a kurtosis indicating the sharpness of the road surface area as a feature of the shape of the road surface area used to determine the presence or absence of a gradient change. That is, as described with reference to FIG. 10, when there is a concave portion with a changing slope on the road surface on which the vehicle travels, the kurtosis of the road surface region in the captured image becomes relatively large. On the other hand, when there is a convex portion whose slope changes on the road surface on which the vehicle travels, the kurtosis of the road surface area in the captured image becomes relatively small. Therefore, based on the kurtosis of the road surface area in the captured image, it is possible to determine the presence or absence of the gradient change on the road surface.

  Moreover, it is possible to use the shape itself of a road surface area | region as a characteristic of the shape of the road surface area | region used for determination of the presence or absence of a gradient change. That is, it is possible to perform machine learning of the relationship between the presence or absence of the gradient change on the road surface and the shape of the road surface region in the captured image, and to determine the presence or absence of the gradient change on the road surface from the shape of the road surface region itself by pattern matching.

  Although the gradient change detection apparatus and method have been described in each of the above-described embodiments and the modification thereof, a program for causing a computer to realize various functions of the gradient change detection apparatus of each embodiment and the modification thereof and the computer A program for executing various procedures of the gradient change detection method of the embodiment and the variation thereof is also included in the scope of the present invention.

DESCRIPTION OF SYMBOLS 10 ... Imaging device 20, 22 ... Gradient change detection apparatus 30 ... Image acquisition part 40 ... Farthest point detection part 50 ... Farthest point position memory | storage part 60, 90 ... Gradient change determination part 70 ... Area ratio calculation part 80 ... Area ratio Memory area R: White line P: Concave portion Q: Convex part F: Disappearing point U: Road surface area K: Farthest point W: Reference road surface area

Claims (13)

An image acquisition unit (30) for acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A farthest point detection unit (40) for detecting the farthest point (K) of the road surface area (U) corresponding to the road surface on which the vehicle travels in the captured image;
A gradient change determination unit (60) that determines the presence or absence of a gradient change on the road surface based on the vertical position of the farthest point in the captured image;
A gradient change detection device comprising: The gradient change determination unit makes a determination based on the position in the vertical direction with respect to the vanishing point (F) of the farthest point in the captured image.
The gradient change detection device according to claim 1. The gradient change determination unit makes a determination based on a temporal change in the vertical position of the farthest point in the captured image.
The gradient change detection device according to claim 1. The farthest point detection unit divides the captured image into a plurality of areas using segmentation based on deep learning, extracts the road surface area from the plurality of areas, and detects the farthest point from the road surface area.
The gradient change detection device according to claim 1. An image acquisition step of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A farthest point detection step of detecting the farthest point of a road surface area corresponding to the road surface on which the vehicle travels in the captured image.
A gradient change determination step of determining presence or absence of a gradient change on the road surface based on the vertical position of the farthest point in the captured image;
A gradient change detection method comprising: On the computer
An image acquisition function of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A farthest point detection function of detecting the farthest point of a road surface area corresponding to the road surface on which the vehicle travels in the captured image;
A gradient change determination function of determining presence or absence of a gradient change on the road surface based on the vertical position of the farthest point in the captured image;
Gradient change detection program to realize. An image acquisition unit (30) for acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A feature acquiring unit (70) for acquiring features of a shape of a road surface area (U) corresponding to a road surface on which the host vehicle travels in the captured image;
A gradient change determination unit (90) that determines presence or absence of a gradient change on the road surface based on features of the shape of the road surface area;
A gradient change detection device comprising: The gradient change determination unit makes a determination based on the feature of the shape of the road surface region with respect to the feature of the shape of the road surface region when there is no gradient change in the road surface.
The gradient change detection device according to claim 7. The gradient change determination unit makes a determination based on a time change of the feature of the shape of the road surface area.
The gradient change detection device according to claim 7. The feature acquisition unit divides the captured image into a plurality of regions using segmentation based on deep learning, extracts the road surface region from the plurality of regions, and acquires a shape feature from the road surface region.
The gradient change detection device according to claim 7. An image acquisition step of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A feature acquisition step of acquiring features of a shape of a road surface area corresponding to a road surface on which the vehicle travels in the captured image;
A gradient change determination step of determining presence or absence of a gradient change on the road surface based on features of the shape of the road surface area;
A gradient change detection method comprising: On the computer
An image acquisition function of acquiring a captured image obtained by capturing an image of the traveling direction of the host vehicle;
A feature acquiring function of acquiring features of a shape of a road surface area corresponding to a road surface on which the host vehicle travels in the captured image;
A gradient change determination function of determining presence or absence of a gradient change on the road surface based on features of the shape of the road surface area;
Gradient change detection program to realize. A vehicle that relaxes or stops the vehicle line control when there is a gradient change on the road surface on which the host vehicle is traveling.

Images