JP2021051560A - Intersection detection system - Google Patents

Abstract

To provide an intersection detection system capable of detecting an intersection on the basis of a factor other than disconnection of a white line.SOLUTION: An intersection detection system 1 includes an imaging device 2 that is installed in an automobile 10 and images a road surface on which the automobile 10 travels, and a detection device 3 that detects an intersection X using an image taken by the imaging device 2. The detection device 3 includes an extraction program P1 that extracts a curve S that bends toward outside in a vehicle width direction at an end in the vehicle width direction of the road surface, and a determination program P2 that determines the intersection X on the basis of the curve S.SELECTED DRAWING: Figure 1

Description

The present invention relates to an intersection detection device.

Conventionally, it has been required to detect an intersection existing in a vehicle's traveling path from the viewpoint of driving support and automatic driving. In detecting an intersection, it is also considered to refer to map data such as a GPS (Global Positioning System) system. However, the cost of using map data is high. In addition, the GPS system has problems such as being prone to misalignment.

Therefore, a method of detecting an intersection existing in a traveling path from a camera mounted on a vehicle or the like is being studied. More specifically, for example, there are an imaging means for imaging the road surface in front of the vehicle, a detecting means for detecting an intersection based on the captured image, a white line detecting means for detecting a white line on the road surface, and an intersection. A white line detection device including a setting means for setting at least one of a plurality of threshold values used for searching a linear approximation region for white line detection has been proposed (see, for example, Patent Document 1).

The above white line detection device detects an edge point on the road surface from an image taken in front of the vehicle, calculates the position of the intersection based on the presence or absence of a break in the edge point, and further, a distant side on an extension of the broken white line. The length of the intersection can be calculated by detecting the white line of.

Japanese Unexamined Patent Publication No. 2014-149636

On the other hand, the white line on the road surface may be interrupted at places other than intersections. Further, when the white line is blurred, it is difficult to detect the break of the white line. Therefore, an intersection detection device capable of detecting an intersection based on an element other than the break of the white line on the road surface is required.

In particular, at a large intersection with a traffic light, the road surface is relatively often provided with a white line that breaks at the intersection, so it is easy to detect the intersection based on the break of the white line. On the other hand, in the case of four Tsuji (for example, a living road) where there is no traffic light, there are relatively many cases where the road surface is not provided with a white line that is interrupted at the intersection, and the above method may not be able to detect the intersection.

The present invention is an intersection detection device capable of detecting an intersection based on an element other than a break in a white line.

The present invention [1] includes a photographing means mounted on a vehicle and photographing a road surface on which the vehicle travels, and a detecting means for detecting an intersection from an image taken by the photographing means, and the detecting means is a vehicle width of the road surface. An intersection detection device including an extraction means for extracting a curved line that curves outward in the vehicle width direction at a direction end portion and a determination means for determining an intersection from the curved line is included.

In such an intersection detection device, when a curved line extending in the vehicle width direction of the vehicle is extracted at the end portion of the road surface in the vehicle width direction, the intersection is determined from the curved line. Therefore, according to the above-mentioned intersection detection device, an intersection can be detected based on an element other than the break of the white line.

In the present invention [2], the intersection detection according to the above [1], wherein the determination means calculates the size of an intersection based on the curved lines when the extracting means extracts a plurality of curved lines. Includes equipment.

In such an intersection detection device, when a plurality of curved lines are extracted at the end portion of the road surface in the vehicle width direction, the size of the intersection is calculated based on the curved lines.

More specifically, when curved lines are detected at both ends of the road surface in the vehicle width direction and are arranged so as to face each other in the vehicle width direction, the intersection is based on the distance between the curved lines. The size can be calculated.

Further, when a curved line is detected at one end of the road surface in the vehicle width direction and another curved line is detected in the front side far away in the front-rear direction of the vehicle with respect to the curved line, those curved lines are detected. The size of the intersection can be calculated based on the distance between them.

Further, a curved line is detected at one end of the road surface in the vehicle width direction, and another curved line is detected at a diagonal distance intersecting the curved line in the front-rear direction and the vehicle width direction of the vehicle. If so, the size of the intersection can be calculated based on the distance between those curved lines.

According to the intersection detection device of the present invention, an intersection can be detected based on an element other than the break of the white line.

FIG. 1 is a schematic view showing an embodiment of an automobile and an intersection detection device. FIG. 2 is a schematic view showing an intersection detected by the intersection detection device shown in FIG. FIG. 3 is a schematic view showing a mode in which only the curved line on the rear side in the front-rear direction of the automobile is photographed at the intersection. FIG. 4 is a schematic view showing a mode in which only a curved line on one side in the vehicle width direction of an automobile is photographed at an intersection. FIG. 5 is a schematic view showing a mode in which only curved lines located diagonally at an intersection are photographed.

1. 1. Overall Configuration of Intersection Detection Device In FIG. 1, the automobile 10 is a vehicle traveling on a road. FIG. 1 shows a car 10 entering the intersection X as a bird's-eye view seen from above.

The automobile 10 is provided with an intersection detection device 1 in order to detect the intersection X from the viewpoints of safe driving and automatic driving. FIG. 1 shows an automobile 10 and an intersection detection device 1 as an enlarged view.

The intersection detection device 1 is mounted on the automobile 10, and includes a photographing device 2 as a photographing means for photographing the road surface on which the vehicle 10 travels, and a detection device 3 as a detecting means for detecting the intersection X from the imaging by the photographing device 2. It has.

The photographing device 2 is an imaging device for photographing the road surface in front of the automobile 10 as an image (still image or moving image).

Examples of the photographing device 2 include a still image camera, a moving image camera, and the like, and a moving image camera is preferable.

Further, the photographing device 2 is arranged in the vehicle interior of the automobile 10 toward the front of the automobile 10. As a result, the photographing device 2 can photograph the front outer view in the course direction of the automobile 10, and can acquire the image as data (hereinafter referred to as imaged data).

The detection device 3 is an information processing device for detecting an intersection X in front of the automobile 10.

More specifically, the detection device 3 has an extraction program P1 as an extraction means for extracting a curved line S that curves outward in the vehicle width direction at an end portion of the road surface in the vehicle width direction, and an intersection from the curved line S. It is provided with a determination program P2 as a determination means for determining X.

Further, the detection device 3 is electrically connected to the photographing device 2 (see the broken line in FIG. 1). As a result, the imaging data obtained by the photographing device 2 can be input to the extraction program P1.

The extraction program P1 is, for example, a program capable of recognizing a diagram drawn on the road surface in front of the automobile 10 from the imaging data obtained by the photographing device 2 and detecting the curved line S.

The curved line S is a line that curves toward the outside in the vehicle width direction at the end (one end or both ends) of the road surface of the intersection X in the vehicle width direction.

The curved line S includes, for example, a painted line drawn on the road surface, for example, a boundary line caused by a difference in material (for example, a boundary line between asphalt (running road) and concrete (road shoulder)), for example, a road shoulder and a sidewalk. Examples include the contour line of the curb that is placed at the boundary of. These can be used alone or in combination of two or more. The curved line S preferably includes a painted line drawn on the road surface. Hereinafter, the case where the curved line S is a painted line will be described in detail.

The curved line S is drawn at a corner at the intersection X, and is, for example, an outside line of the roadway. The outside line of the roadway is, for example, a lane marking line that separates the roadway and the roadside zone when there is no sidewalk at the intersection X, and separates the roadway and the shoulder when there is a sidewalk at the intersection X. It is a lane marking. The curved line S is preferably a lane marking line that separates the roadway and the roadside zone.

The curved line S may be present in any one of the corners (four in FIG. 1) at the intersection X, or may be present in all the corners.

In FIG. 1, curved lines S exist in all four corners of the intersection X.

In the following, when distinguishing each curved line S, the corner on the road side where the automobile 10 travels (the rear side in the front-rear direction (traveling direction A2) of the vehicle body of the automobile 10) and one corner in the vehicle width direction of the automobile 10 ( The curved line S (on the lower left side of the paper surface) is defined as the curved line S1.

Further, the curved line S on the road side on which the automobile 10 travels (the rear side in the front-rear direction of the vehicle body of the automobile 10) and the other corner (lower right side of the paper surface) in the vehicle width direction of the automobile 10 is referred to as the curved line S2. To do.

Further, the curved line S on the road side facing the automobile 10 (front side in the front-rear direction of the vehicle body of the automobile 10) and one corner (upper left side of the paper surface) of the automobile 10 in the vehicle width direction is defined as the curved line S3.

Further, the curved line S on the road side facing the automobile 10 (front side in the front-rear direction of the vehicle body of the automobile 10) and the other corner (upper right side of the paper surface) of the automobile 10 in the vehicle width direction is defined as the curved line S4.

The form of these curved lines S differs depending on the intersection X, but may be connected to the lateral line L.

The side line L is, for example, a painted line drawn on the road surface, for example, a boundary line caused by a difference in material (for example, a boundary line between asphalt (running road) and concrete (road shoulder)), for example, a road shoulder and a sidewalk. Examples include the outline of the curb that is placed at the boundary. These can be used alone or in combination of two or more. As the lateral line L, a painted line drawn on the road surface is preferable. Hereinafter, the case where the lateral line L is a painted line will be described in detail.

The lateral line L is a line extending along the front-rear direction (roadway) of the vehicle body at the vehicle width direction end portion (one end portion or both end portions) of the road surface of the traveling path of the automobile 10. The lateral line L is drawn, for example, at the end (one side or both sides) of the road connected to the intersection X in the vehicle width direction, and is, for example, an outside line of the road. The siding L is preferably a siding that separates the roadway from the roadside zone.

Then, the lateral line L is preferably connected to the curved line S at the corner of the intersection.

In other words, preferably, the curved line S is a line that is continuous from the lateral line L extending along the front-rear direction of the vehicle body and curves outward in the vehicle width direction.

The colors of the curved line S and the side line L are not particularly limited, but are usually white lines.

Then, the extraction program P1 can extract the curved line S imaged by the photographing device 2 as data (hereinafter, curved line data).

The details of the extraction of the curved line S by the extraction program P1 will be described later.

The determination program P2 is a program capable of determining the intersection X from the curved line data extracted by the extraction program P1.

The determination program P2 is continuously provided from the extraction program P1. As a result, the determination program P2 can input the curve line data output by the extraction program P1.

Further, the determination program P2 is configured to be able to determine, for example, the presence / absence of an intersection X, the position of the intersection X, the angle of the intersection X, the size of the intersection X, and the like based on the curved line data.

That is, the determination program P2 can determine the intersection X based on the curved line data.

The details of the determination of the intersection X by the determination program P2 will be described later.

Further, the intersection detection device 1 can be provided with a notification device 6 for notifying the driver of the automobile 10 of the intersection X (presence / absence, position, angle, size, etc.) determined above, if necessary. ..

The notification device 6 is not particularly limited as long as it can notify the driver of the intersection X (presence / absence, position, angle, size, etc.), but for example, a voice guidance device for notifying the driver of the intersection X by voice, for example, a virtual image ( An example is a visual guidance device that notifies the driver of an intersection X by displaying a virtual image for augmented reality. A visual guidance device is preferably used.

In such a case, the notification device 6 uses, for example, a generation device 4 that generates a virtual image icon for notifying the driver of the intersection X (presence / absence, position, angle, size, etc.) and the virtual image icon of the automobile 10. It is provided with a display device 5 for superimposing and displaying on the front outside view.

The generation device 4 is an information processing device for generating a virtual image icon indicating the intersection X.

The virtual image icon indicating the intersection X is a virtual image for notifying the driver of the automobile 10 of the intersection X (presence / absence, position, angle, size, etc.), and is superimposed on the outside view of the automobile 10 in the display device 5 described later. Is displayed.

The generation device 4 has a generation program P3 for generating a virtual image icon.

Further, the generation device 4 is electrically connected to the detection device 3 (see the broken line in FIG. 1). As a result, the intersection data obtained by the detection device 3 can be input to the generation program P3.

Then, the generation program P3 is configured to be able to generate a virtual image icon that can be duplicated in the outside view from the intersection data related to the intersection X based on a known object generation algorithm or the like.

The generation program P3 can acquire the virtual image icon indicating the intersection X as data (hereinafter, virtual image data) in the generation device 4.

The display device 5 is a display device that superimposes and displays the virtual image icon generated by the generation device 4 on the front outside view (including the road surface) of the automobile 10.

More specifically, the display device 5 includes, for example, a head-up display, an organic EL display, a semitransparent liquid crystal display, and the like, and transmits the front outside view in the visual field range (for example, windshield) of the driver of the automobile 10. Arranged as possible.

Further, the display device 5 is electrically connected to the generation device 4 (see the broken line in FIG. 1).

As a result, the display device 5 can display the virtual image icon generated by the generation device 4. More specifically, the display device 5 is capable of transmitting the front outer view of the automobile 10 and displaying a virtual image icon so as to overlap the front outer view.

That is, the display device 5 can display the actual front outside view and the virtual image icon indicating the intersection X in an overlapping manner as augmented reality (AR) technology.

In the intersection detection device 1, the photographing device 2, the detection device 3, the generation device 4, and the display device 5 may be independent of each other, or a plurality of devices may be integrally configured.

2. Intersection Detection The method of detecting an intersection X by the intersection detection device 1 will be described in detail below with reference to FIGS. 1 and 2.

More specifically, in the intersection detection device 1, first, while the automobile 10 is running or stopped, the road surface in front of the automobile 10 by the photographing device 2 is shown by the arrow A1 and the fan region Z1 in FIG. Is photographed, and the photographed data is input to the extraction program P1 of the detection device 3. Then, the extraction program P1 performs arithmetic processing on the image data according to a predetermined algorithm.

For example, in the extraction program P1, while the automobile 10 is traveling, the lateral line L drawn on the vehicle width direction end portion (one end portion or both end portions) of the road surface is extracted from the imaging data.

Further, the extraction program P1 extracts the lateral line in the curved portion as the curved line S when the curvature of the lateral line L toward the outside in the vehicle width direction is confirmed.

Further, the extraction program P1 extracts one or a plurality of curved lines S as curved line data according to the shooting range (shooting distance, shooting angle) of the shooting device 2.

The method of extracting the curved line S by the extraction program P1 is not particularly limited, but for example, edge extraction, pattern matching, a machine learning algorithm, or the like is adopted.

More specifically, the curved line S is distinguished from, for example, the lateral line L in the curved portion of the road by analyzing elements such as the edge, shape, color, and amount of change in brightness by a machine learning algorithm. And extracted as a curved line S.

Further, the number of curved lines S extracted by the extraction program P1 is not particularly limited. For example, when the four corners of the intersection X are within the shooting range of the shooting device 2, as shown as the fan region Z1 in FIG. 1, the extraction program P1 draws four curved lines S drawn at each corner. Can be extracted.

As a result, the detection device 3 detects the curved line S as data (curved line data). The curved line data includes the position, size, radius of curvature, and the like of the curved line S.

Next, in this method, the curved line data obtained above is input to the determination program P2, and the determination program P2 processes the curve line data according to a predetermined algorithm.

More specifically, in the determination program P2, for example, when the curved line S exists, it is determined that the intersection X exists. Further, the determination program P2 can determine that the intersection X does not exist when the curved line S does not exist.

Further, the determination program P2 can calculate, for example, the curvature φ and the radius of curvature ρ of the curved line S, and can calculate the intersection angle of the intersection from the curvature φ and the radius of curvature ρ.

Further, when the extraction program P1 extracts a plurality of curved lines S, the determination program P2 can calculate the distances D of the plurality of curved lines S, and can calculate the size of the intersection from the distances.

That is, the width of the road on which the automobile 10 travels is calculated from the distance D between the curved line S1 and the curved line S2. Further, the width of the road facing the automobile 10 is calculated from the distance D between the curved line S3 and the curved line S4. Further, the width of the roadway intersecting the roadway on which the automobile 10 travels is calculated from the distance D (and / or the distance D between the curved line S2 and the curved line S4) between the curved line S1 and the curved line S3. Further, the diagonal distance of the intersection X is calculated from the distance D between the curved line S1 and the curved line S3 (and / or the distance D between the curved line S2 and the curved line S4).

Then, the size of the intersection X is calculated from the width and diagonal distance of these roadways.

By such a determination program P2, the intersection X can be acquired as data (hereinafter, intersection data). The intersection data includes, for example, the presence / absence of the intersection X, the position, the angle, the size, and the like.

Next, in this method, the intersection data obtained above is input to the generation program P3 of the generation device 4, and the generation program P3 processes the intersection data according to a predetermined algorithm.

As a result, in the generation device 4, a virtual image icon for notifying the driver of the intersection X (presence / absence, position, angle, size, etc.) is generated as virtual image data. The method for generating virtual image data is not particularly limited, and a known calculation method is adopted.

After that, in this method, the virtual image data generated by the generation device 4 is output to the display device 5. That is, the display device 5 displays a virtual image icon for notifying the driver of the intersection X (presence / absence, position, angle, size, etc.).

Thereby, the intersection X can be detected based on the curved line S, and the driver can be notified about the intersection X.

Although not shown in detail, the intersection detection device 1 may further include an indoor camera (eye tracking camera) for observing the concentrated position of the line of sight of the driver of the automobile 10. Then, the display position of the intersection X can be adjusted to the concentrated position of the driver's line of sight. As a result, the driver can be notified of the intersection X efficiently.

3. 3. Action / Effect As described above, when the curved line S extending in the vehicle width direction of the automobile 10 is extracted at the end of the road surface in the vehicle width direction, the intersection detection device 1 determines the intersection from the curved line S. .. Therefore, according to the above-mentioned intersection detection device 1, the intersection X can be detected based on an element other than the break of the white line.

That is, in detecting an intersection X, for example, it is considered to refer to a white line for dividing a lane and determine that the break of the white line is an intersection.

However, the above white line may be interrupted at places other than intersections. Further, when the white line is blurred, it is difficult to detect the break of the white line.

In particular, at an intersection with a traffic light, a white line that breaks at the intersection is relatively often provided on the road surface, so that it is easy to detect the intersection based on the break of the white line. On the other hand, in four Tsuji (for example, a living road) without a traffic light, there are relatively many cases where the road surface is not provided with a white line that is interrupted at an intersection.

On the other hand, the intersection detection device 1 detects an intersection based on a curved line at the end (road shoulder) in the vehicle width direction, instead of a white line for dividing the lane.

Therefore, according to the intersection detection device 1 described above, the intersection X can be detected based on an element other than the break of the white line.

Further, in the above-mentioned intersection detection device 1, preferably, when a plurality of curved lines S are extracted at the end portion of the road surface in the vehicle width direction, the size of the intersection is calculated based on the curved lines S.

More specifically, when curved lines S are detected at both ends of the road surface in the vehicle width direction and are arranged so as to face each other in the vehicle width direction (for example, the curved lines S1 and the curved lines S2 are When it is detected (for example, when the curved line S3 and the curved line S4 are detected), the size of the intersection can be calculated based on the distance D of the curved lines S.

Further, when a curved line S is detected at one end of the road surface in the vehicle width direction, and another curved line S is detected in the front-rear direction of the vehicle with respect to the curved line (for example). , When the curved line S1 and the curved line S3 are detected, for example, when the curved line S2 and the curved line S4 are detected), the size of the intersection is calculated based on the distance D between the curved lines S. it can.

Further, a curved line S is detected at one end of the road surface in the vehicle width direction, and another curved line S is diagonally distant from the curved line and intersects the vehicle in the front-rear direction and the vehicle width direction. (For example, when the curved line S1 and the curved line S4 are detected, for example, when the curved line S2 and the curved line S3 are detected), based on the distance between the curved lines S, The size of the intersection can be calculated.

4. Modification example In the above-mentioned intersection detection device 1, as shown as a fan region Z1 in FIG. 1, the photographing range of the photographing device 2 is relatively wide, and the curved lines of all the corners (four) of the intersection X are photographed by the photographing device 2. However, the shooting range of the shooting device 2 may be relatively narrow.

For example, as shown in FIG. 3, at the intersection X, the photographing device 2 displays only the curved lines S1 and the curved lines S2 at the corners on the road side (rear side in the front-rear direction of the vehicle body of the automobile 10) on which the automobile 10 travels. It may be photographed (see fan area Z2).

Even in such a case, if the curve line S is imaged, the extraction program P1 can extract at least one curve line S, and the determination program P2 can determine the position and angle of the intersection X. ..

More specifically, the intersection detection device 1 can determine the position and angle of the intersection X by extracting at least one curved line S. Further, the size of the intersection X can be determined by extracting two or more curved lines S.

Further, for example, if the curved lines S at the corners on both sides in the vehicle width direction on the road side (rear side in the front-rear direction of the vehicle body of the vehicle 10) on which the vehicle 10 travels are photographed, the extraction program P1 can execute the two bays. The curve S can be extracted, and the determination program P2 can calculate the distance D between the two curved lines S and determine the size of the intersection X.

Further, depending on the intersection X, as shown in FIG. 4, a curved line S may be drawn only at one corner of the automobile 10 in the vehicle width direction, and the photographing device 2 may draw the curved line S at the intersection X of the automobile 10. Only the curved line S1 and the curved line S3 on one side in the vehicle width direction may be photographed (see fan region Z2). Further, although not shown, the photographing device 2 may photograph only the curved line S2 and the curved line S4 on the other side in the vehicle width direction of the automobile 10 at the intersection X.

Even in such a case, if the curve line S is imaged, the extraction program P1 can extract at least one curve line S, and the determination program P2 can determine the position and angle of the intersection X. ..

Further, for example, if the curved lines S at the two corners on one side in the vehicle width direction of the automobile 10 are photographed, the extraction program P1 can extract the two curved lines S, and the determination program P2 can extract the curved lines S. The distance D of S can be calculated and the size of the intersection X can be determined.

Further, depending on the intersection X, as shown in FIG. 5, the curved line S may be drawn only at the two diagonal corners, and the photographing device 2 is positioned diagonally at the intersection X. Only the curved line S1 and the curved line S4 (or the curved line S2 and the curved line S3) may be photographed (see the fan region Z4).

Even in such a case, if the curve line S is imaged, the extraction program P1 can extract at least one curve line S, and the determination program P2 can determine the position and angle of the intersection X. ..

Further, for example, if the curved lines S of the two diagonally located corners are both photographed, the extraction program P1 can extract the two curved lines S, and the determination program P2 can extract the curved lines S of the curved lines S. The distance D can be calculated and the size of the intersection X can be determined.

Although not shown, the photographing device can also photograph only one curved line S at the intersection X. Even in such a case, the extraction program P1 can extract at least one curved line S, and the determination program P2 can determine the position and angle of the intersection X based on the one curved line S.

1 Intersection detection device 2 Imaging device 3 Detection device S Curved line X Intersection

Claims (2)

A shooting means that is mounted on a vehicle and photographs the road surface on which the vehicle travels.
A detection means for detecting an intersection from an image taken by the photographing means is provided.
The detection means
An extraction means for extracting a curved line that curves outward in the vehicle width direction at the end of the road surface in the vehicle width direction,
An intersection detection device comprising a determination means for determining an intersection from the curved line. The determination means
The intersection detection device according to claim 1, wherein when the extraction means extracts a plurality of curved lines, the size of the intersection is calculated based on the curved lines.

Images