JP7325296B2 - Object recognition method and object recognition system - Google Patents

Description

The present invention relates to an object recognition method and an object recognition system.

Patent Literature 1 discloses a parked vehicle detection device that identifies the parking position of a parked vehicle. The parked vehicle detection device detects other vehicles included in an image captured by an imaging device mounted on the own vehicle, calculates the position of the detected other vehicle, and detects the position of the other vehicle including the calculated position. The information is stored, and based on the information of other vehicles, it is determined whether the other vehicle is likely to be a parked vehicle.

JP 2015-76074 A

However, it is sometimes difficult to distinguish and recognize a stationary object from its background. For example, when recognizing a stationary object from a captured image, if the colors of the stationary object and the background are similar, the stationary object and the background are assimilated, making it difficult to distinguish between them.
Further, for example, when a stationary object is detected by a distance measuring device such as a laser radar, it may be difficult to distinguish between the stationary object and the background stationary object due to ranging errors.
SUMMARY OF THE INVENTION An object of the present invention is to improve the accuracy of estimating an area in which a stationary object exists on a road.

In an object recognition method according to an aspect of the present invention, a target extending along a road is detected from an observation point on the road, and a detection section, which is a section in which a portion of the target detected from the observation point exists, is detected. Based on the positions of the end points of a plurality of different detection sections detected at a plurality of observation points, a stationary object existence area, which is an area in which a stationary object exists on the road, is estimated.

ADVANTAGE OF THE INVENTION According to this invention, the estimation precision of the area|region where a stationary object exists on a road can be improved.

1 is a schematic configuration diagram of an example of an object recognition system according to a first embodiment; FIG. FIG. 4 is an explanatory diagram of an example of an object recognition method according to the first embodiment; FIG. 4 is an explanatory diagram of an example of an object recognition method according to the first embodiment; FIG. 4 is an explanatory diagram of an example of an object recognition method according to the first embodiment; FIG. 4 is an explanatory diagram of an example of an object recognition method according to the first embodiment; FIG. 4 is an explanatory diagram of an example of an object recognition method according to the first embodiment; It is a block diagram of an example of functional composition of an object recognition system of a 1st embodiment. 4 is a flow chart of an example of an object recognition method according to the first embodiment; It is explanatory drawing of an example of the object recognition method of 2nd Embodiment. It is explanatory drawing of an example of the object recognition method of 2nd Embodiment. It is explanatory drawing of an example of the object recognition method of 2nd Embodiment. It is explanatory drawing of an example of the object recognition method of 2nd Embodiment. It is explanatory drawing of an example of the object recognition method of 2nd Embodiment. 9 is a flow chart of an example of an object recognition method according to the second embodiment; FIG. 11 is an explanatory diagram of an example of a data exchange sequence in the object recognition system of the second embodiment; FIG. 11 is a schematic configuration diagram of an example of an object recognition system according to a third embodiment; FIG. 11 is a block diagram of an example of the functional configuration of an object recognition system according to a third embodiment; FIG. It is explanatory drawing of an example of the object recognition method of 3rd Embodiment. It is explanatory drawing of an example of the object recognition method of 3rd Embodiment. It is explanatory drawing of an example of the object recognition method of 3rd Embodiment. FIG. 10 is an explanatory diagram of a first example of correction processing for the estimation result of the stationary object existence area; FIG. 10 is an explanatory diagram of a second example of correction processing for the estimation result of the stationary object existence area; FIG. 12 is an explanatory diagram of a third example of correction processing for the estimation result of the stationary object existence area; FIG. 10 is an explanatory diagram of an example of correction processing for end points of a detection section; 11 is a flow chart of an example of an object recognition method according to the third embodiment; FIG. 11 is a schematic configuration diagram of an example of an object recognition system according to a fourth embodiment; FIG. 12 is a block diagram of an example of the functional configuration of an object recognition system according to the fourth embodiment; FIG. It is explanatory drawing of an example of the object recognition method of 4th Embodiment. It is explanatory drawing of an example of the object recognition method of 4th Embodiment. FIG. 12 is a flow chart of an example of an object recognition method according to the fourth embodiment; FIG.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the description of the drawings, the same or similar parts are denoted by the same or similar reference numerals, and overlapping descriptions are omitted. Each drawing is schematic and may differ from the actual one. The embodiments shown below exemplify devices and methods for embodying the technical idea of the present invention, and the technical idea of the present invention is specific to the devices and methods illustrated in the following embodiments. not something to do. Various modifications can be made to the technical idea of the present invention within the technical scope described in the claims.

(First embodiment)
(composition)
Please refer to FIG. The object recognition system 100 of the first embodiment includes an external sensor 10 mounted on the first vehicle 1, a positioning device 11, a communication device 12, and computers 15, 16 and 17. The first vehicle 1 is further provided with an in-vehicle device 13 and an actuator 14 .
The external sensor 10 comprises a plurality of different types of sensors for detecting objects around the first vehicle 1 . The external sensor 10 is an example of the "object detection means" described in the claims.

For example, the external sensor 10 comprises a camera mounted on the first vehicle 1 . The external sensor 10 includes, for example, one or more cameras that capture images of the surroundings of the first vehicle 1 . The external sensor 10 includes, for example, a 4K resolution camera with an angle of view of 90 degrees that captures the front of the first vehicle 1, a 4K resolution camera with an angle of view of 90 degrees that captures the rear of the first vehicle 1, A 4K resolution camera with an angle of view of 180 degrees for imaging the left side and a 4K resolution camera with an angle of view of 180 degrees for imaging the right side of the first vehicle 1 may be provided.

The external sensor 10 captures a 360-degree range around the first vehicle 1 by acquiring images captured by each camera while synchronizing these cameras.
Further, for example, the external sensor 10 includes a ranging sensor such as a laser radar, a millimeter wave radar, or LIDAR (Light Detection and Ranging, Laser Imaging Detection and Ranging), and detects an object or road in front of, behind, or to the side of the first vehicle 1. White lines, road surface position data or point cloud data may be obtained.

The positioning device 11 includes a global positioning system (GNSS) receiver, receives radio waves from a plurality of navigation satellites, and measures the current time and the current position and traveling direction of the first vehicle 1 . A GNSS receiver may be, for example, a global positioning system (GPS) receiver or the like. The positioning device 11 may be, for example, an inertial navigation device, and may measure the current position and traveling direction of the first vehicle 1 by odometry.

The communication device 12 distributes the reference time to the external sensor 10 based on the current time information obtained from the positioning device 11 . Further, the communication device 12 exchanges data with the section detection means 15b, the difference detection means 16b and the area estimation means 17b via the communication means 15a, 16a and 17a of the computers 15, 16 and 17, respectively.
The in-vehicle device 13 is an electronic control unit (ECU) that controls running of the first vehicle 1 . For example, the in-vehicle device 13 performs automatic travel control for automatically driving the first vehicle 1 without involving the driver, based on the current position of the first vehicle 1 and the surrounding environment.

Further, for example, the in-vehicle device 13 performs driving support control that controls only the steering or acceleration/deceleration of the first vehicle 1 based on the surrounding environment.
The in-vehicle device 13 includes a processor and peripheral components such as a storage device. The processor may be, for example, a CPU (Central Processing Unit) or an MPU (Micro-Processing Unit). The storage device may comprise a semiconductor storage device, a magnetic storage device, an optical storage device, or the like. The storage device may include memories such as registers, cache memories, ROMs (Read Only Memories) used as main storages, and RAMs (Random Access Memories).

The actuator 14 operates the steering wheel, the accelerator opening, and the braking device of the first vehicle 1 according to the control signal from the in-vehicle device 13 to cause the vehicle behavior of the first vehicle 1 . The actuator 14 includes a steering actuator, an accelerator opening actuator, and a brake control actuator. The steering actuator controls the steering direction and amount of steering of the first vehicle 1 . The accelerator opening actuator controls the accelerator opening of the first vehicle 1 . The brake control actuator controls the braking operation of the brake system of the first vehicle 1 .

The computers 15 to 17 are provided with communication means 15a to 17a, respectively, and operate as section detection means 15b, difference detection means 16b, and area estimation means 17b, which will be described later. Computers 15-17 may each be implemented as separate hardware or may be integrated into common hardware.
Also, for example, the computers 15 to 17 may be mounted in the first vehicle 1 . In this case, the communication device 12 and the communication means 15a to 17a may exchange data by wired or wireless communication means. For example, the communication device 12 and the communication means 15a to 17a may be provided with an in-vehicle communication network (in-vehicle LAN) such as CSMA/CA multiplex communication (CAN: Controller Area Network) or Flex Ray.

Computers 15 - 17 may be mounted at locations other than first vehicle 1 . In this case, for example, the communication device 12 and the communication means 15a to 17a exchange data with each other via wireless communication means such as vehicle-to-vehicle communication, public mobile phone network, road-to-vehicle communication, and satellite communication. The computers 15 to 17 may be installed, for example, in other vehicles, mobile edge computers on mobile phones/own networks, road traffic infrastructure servers installed on the roadside, or cloud servers on the Internet.

Next, an outline of an object recognition method by the object recognition system 100 of the first embodiment will be described.
As shown in FIG. 2A, it is assumed that the first vehicle 1 travels on a road 200 and there are targets 202 that extend along the road 200 or exist in a row along the road 200 . For example, the target 202 may be a road structure (guardrail, curb, etc.) extending along the road 200, road surface paint extending along the road 200 such as a road white line drawn on the road boundary, or road surface paint of the road 200. It can be a road surface. Also, the target 202 may be a building existing along the road (extending along the road 200).
A stationary object 201 exists on the road 200 and occupies part of the road 200 . Stationary object 201 may be, for example, a parked vehicle.

First, as shown in FIG. 2B, a target 202 is detected at point P1 (observation point). Reference signs Sd11 and Sd12 indicate sections in which portions of the target 202 that are detected from the point P1 are present (hereinafter referred to as "detection sections").
In the example of FIG. 2B, the stationary object 201 causes part of the target 202 to be a non-line-of-sight area from the point P1. Therefore, this portion cannot be detected by the external sensor 10, and the detection sections Sd11 and Sd12 are detected as discontinuous detection sections.
Of the pair of detection sections adjacent to each other across the discontinuous portion, the detection section Sd11 closer to the point P1 is referred to as a "first detection section". Also, the other detection section Sd12, which is farther from the point P1, is referred to as a "second detection section".

Further, the position Pe11 of one end closer to the second detection section Sd12 among both ends of the first detection section Sd11 is referred to as the "end point" of the first detection section Sd11.
The position Ps12 of one end closer to the first detection section Sd11 among both ends of the second detection section Sd12 (that is, the position Ps12 of one end closer to the point P1 among both ends of the second detection section Sd12) is set as the "start point" of the second detection section Sd12. ”.
Furthermore, the section between the end point Pe11 of the first detection section Sd11 and the start point Ps12 of the second detection section Sd12 is estimated as a non-detection section Su1, which is a section in which there is a portion of the target 202 that is not detected from the point P1.

See FIG. 2C. Next, the target 202 is detected from a point P2 (observation point) which is a predetermined distance away from the point P1. The detection sections Sd21 and Sd22 are a first detection section and a second detection section in which a portion of the target 202 detected from the point P2 exists. FIG. 2C shows the end point Pe21 of the first detection section Sd21 and the start point Ps22 of the second detection section Sd22. Also, a non-detection section Su2, which is a section in which a portion of the target 202 that is not detected from the point P2 exists, is estimated.

Next, the difference between the second detection section Sd12 detected at the point P1 and the second detection section Sd22 detected at the point P2 is detected. Specifically, the difference between the starting point Ps12 of the second detection section Sd12 and the starting point Ps22 of the second detection section Sd22 is detected.
When the first detection section Sd11 and the second detection section Sd12 are cut off by the stationary object 201 on the road, and the first detection section Sd21 and the second detection section Sd22 are cut off, FIGS. 2B and 2C show Thus, the starting point Ps22 moves from the starting point Ps12.

On the other hand, when the stationary object 201 does not exist and the target 202 itself is simply interrupted, the start point Ps12 and the start point Ps22 match from any point.
Therefore, when there is a difference between the starting point Ps12 and the starting point Ps22, it is determined that the stationary object 201 exists. If there is no difference between the start point Ps12 and the start point Ps22, it is determined that the stationary object 201 does not exist.

See FIG. 2D. When it is determined that the stationary object 201 exists, a section in which the target 202 is not detected from either of the points P1 and P2 is estimated as an "object existence possibility section". An object presence possibility interval is an interval in which a stationary object may exist.
For example, the section between the first detection section Sd21 and the second detection section Sd22 (the section of the road 200 from the end point Pe21 to the start point Ps22) is estimated as the object presence possibility section.

Note that when the object 201 is stationary, the end point Pe11 of the first detection section Sd11 and the end point Pe12 of the first detection section Sd21 coincide to form a common end point Pe. On the other hand, when the object 201 is stationary, the end point Pe11 and the end point Pe12 do not match.
Therefore, it may be determined that the stationary object 201 exists when there is a difference between the start point Ps12 and the start point Ps22 and the end point Pe11 and the end point Pe12 do not change. This can prevent the moving object from being erroneously detected as the stationary object 201 .

Finally, the stationary object existence area is estimated by restricting the object existence possibility section in the road width direction. See FIG. 2E.
For example, a line segment connecting the point P2 and the starting point Ps22 is calculated as the "line of sight boundary line 203". The line of sight boundary line is a line segment obtained by extending the boundary line between the area outside the line of sight range and the area where the line of sight is visible due to the stationary object 201 to point P2, which is an observation point.
For example, the intersection point Pv between the line of sight boundary line 203 and the perpendicular line 204 perpendicularly crossing the first detection section Sd21 at the common end point Pe is obtained. The width W of the stationary object existence area is calculated based on the length of the interval between the end point Pe and the intersection point Pv, and the area 205 having the width W from the end point Pe to the start point Ps22 is estimated as the stationary object existence area.

Alternatively, for example, an area surrounded by a line segment connecting the end point Pe and the intersection point Pv, a line segment connecting the intersection point Pv and the start point Ps22, and a line segment connecting the start point Ps22 and the end point Pe may be estimated as the stationary object presence area. good.
As described above, when a portion of the target 202 along the road is shielded from the observation point by the stationary object 201, the object recognition system 100 detects the endpoint of the portion of the target 202 detected from the observation point. changes as the observation point changes, the existence area of the stationary object 201 is estimated. As a result, it is possible to improve the accuracy of estimating an area in which a stationary object exists on the road.

Next, a functional configuration of the object recognition system 100 will be described. FIG. 3 is a block diagram of an example of the functional configuration of the object recognition system 100 of the first embodiment.
The object recognition system 100 includes an external sensor 10, a positioning device 11, a section detection means 15b, a difference detection means 16b, an area estimation means 17b, an actuator 14, a trajectory generation means 110, and a travel control means 111. The in-vehicle device 13 of the first vehicle 1 operates as a trajectory generation means 110 and a travel control means 111 .

The external sensor 10 detects objects around the first vehicle 1 . For example, the external sensor 10 captures a road white line of about 50 m with a vehicle-mounted camera within a horizontal line of sight of 360 degrees around the first vehicle, and detects it as a target 202 extending along the road 200 .
The positioning device 11 measures the current position of the first vehicle 1 in a common coordinate system (for example, a world coordinate system or a map coordinate system) based on predetermined coordinates.

The section detection means 15b detects the first detection section Sd11 and the second detection section Sd12 based on the detection result of the external sensor 10 at the point P1, and estimates the non-detection section Su1. Also, based on the detection result by the external sensor 10 at a point P2 which is a predetermined distance D from the point P1, the first detection section Sd21 and the second detection section Sd22 are detected, and the non-detection section Su2 is estimated.

The predetermined distance D between the points P1 and P2 may be determined according to the detection error range of the position of the target 202 by the external sensor 10 .
For example, if the target 202 is 50 m away from the first vehicle 1 and the external sensor 10 includes an error of about 10%, a +10% margin is given to satisfy that the points P1 and P2 are sufficiently far apart. Then, the predetermined distance D is determined as 50m×20%=10m.

The difference detection means 16b detects the second detection section Sd12 detected at the point P1 and the second detection section Sd12 detected at the point P2 based on the detection result of the section detection means 15b and the measurement result of the current position of the first vehicle 1 by the positioning device 11. The difference from the second detection section Sd22 is detected.
Specifically, the coordinate system of the detection section detected by the section detection means 15b based on the current position of the first vehicle 1 is converted into a common coordinate system, and the second detection section Sd12 detected at the point P1 and the second detection section Sd12 detected at the point P2 The detected second detection section Sd22 is compared on the common coordinate system.

For example, the difference between the starting point Ps12 of the second detection section Sd12 and the starting point Ps22 of the second detection section Sd22 is detected. The difference detection means 16b transmits the detection result to the area estimation means 17b.
The area estimation means 17b determines that the stationary object 201 exists when there is a difference between the second detection section Sd12 and the second detection section Sd22.
If there is no difference between the second detection section Sd12 and the second detection section Sd22, it is determined that the stationary object 201 does not exist.

In addition to the fact that there is a difference between the start point Ps12 and the start point Ps22, the area estimation means 17b also detects the stationary object 201 may be determined to exist.

When it is determined that the stationary object 201 exists, the region estimation means 17b selects a section in which the target 202 is not detected from either of the points P1 and P2 (for example, a section between the first detection section Sd21 and the second detection section Sd22). is estimated as the object existence possibility interval.
That is, the area estimating means 17b divides the non-detection section Su1 in which a portion of the target 202 that is not detected from the point P1 exists, and the non-detection section Su2 that is a section in which a portion of the target 202 that is not detected from the point P2 exists. Overlap regions are estimated as object presence probability intervals.

Further, the region estimation means 17b limits the object existence possibility section in the road width direction to estimate the stationary object existence region. The region estimating means 17b outputs the positional information of the estimated stationary object existence region to the in-vehicle device 13. FIG.
The position information of the stationary object existence area may include, for example, coordinates such as the latitude and longitude of the vertex of the stationary object existence area and link information connecting these coordinates. The position information of the stationary object existence area may express the section where the stationary object existence area exists by the lane identification information and the distance on the map database, and the stationary object existence area is expressed by the road link identification information and the coordinate information. An existing interval may be represented.

The trajectory generation means 110 of the in-vehicle device 13 generates a target travel trajectory and a speed profile on which the first vehicle 1 travels in automatic travel control or driving support control of the first vehicle 1 .
When the area estimating means 17b estimates the stationary object existence area ahead of the course of the first vehicle 1, the trajectory generating means 110 generates a target traveling trajectory and a speed profile so as to avoid the stationary object existing area.

The traveling control means 111 drives the actuator 14 so that the first vehicle 1 travels on the target traveling trajectory at a speed according to the speed profile generated by the trajectory generating means 110 . Accordingly, the actuator 14 controls the steering direction and steering amount of the steering of the first vehicle 1, for example, so that the first vehicle 1 avoids the stationary object existence area.
The actuator 14 also controls the braking device of the first vehicle 1 so that the first vehicle 1 decelerates or stops before the stationary object existence area.

As a result, in a situation where it is difficult to recognize the stationary object 201 and determine its size, the stationary object existence area can be estimated in advance based on the results of observing objects around the stationary object 201 . As a result, the vehicle behavior to avoid the stationary object 201 can be generated without waiting for the stationary object 201 to be recognized and its size determined.
Further, the in-vehicle device 13 may notify the presence of the stationary object presence area and its position information to the driver and passengers of the first vehicle 1 as audio information and visual information.

(motion)
Next, an example of the object recognition method of the first embodiment will be described with reference to FIG.
In step S1, the external sensor 10 detects the target object 202 within the line of sight from the point P1. The section detection means 15b detects the first detection section Sd11 and its end point Pe11, and the second detection section Sd12 and its start point Ps12.

In step S2, the positioning device 11 detects that the first vehicle 1 has reached a point P2 that is a predetermined distance D away from the point P1. The external sensor 10 detects a target 202 within the line of sight from the point P2. The section detection means 15b detects the first detection section Sd21 and its end point Pe21, and the second detection section Sd22 and its start point Ps22.
In step S3, the difference detection means 16b determines the difference between the second detection section Sd12 detected at the point P1 and the second detection section Sd22 detected at the point P2.

If there is a difference between the second detection section Sd12 and the second detection section Sd22 (step S3: Y), the process proceeds to step S4. If there is no difference between the second detection section Sd12 and the second detection section Sd22 (step S3: N), the processing ends without estimating the object presence possibility section.
In step S4, the region estimation means 17b estimates a section in which the target object 202 is not detected from either of the points P1 and P2 as an object presence possibility section.

(Effect of the first embodiment)
(1) The external sensor 10 detects targets 202 extending or existing along the road 200 from observation points P1 and P2 on the road 200 . The section detection means 15b detects detection sections Sd11, Sd12, Sd21 and Sd22, which are sections in which portions of the target 202 detected from the observation points P1 and P2 are present. The difference detecting means 16b and the area estimating means 17b detect a stationary object 201 on the road 200 based on the positions of the end points of the plurality of different detection sections Sd11, Sd12, Sd21 and Sd22 respectively detected at the plurality of observation points P1 and P2. Estimate the stationary object presence region, which is the region where

As a result, the accuracy of estimating the area where the stationary object 201 exists on the road 200 can be improved.
In addition, in a situation where it is difficult to recognize the stationary object 201 or determine its size, the stationary object existence area can be estimated in advance based on the results of observing targets around the stationary object 201 . As a result, the vehicle behavior to avoid the stationary object 201 can be generated without waiting for the stationary object 201 to be recognized and its size determined.

(2) The section detection means 15b estimates non-detection sections Su1 and Su2, which are sections in which there are portions not detected from the observation points P1 and P2 of the target 202, for the plurality of observation points P1 and P2. The region estimating means 17b estimates a stationary object existence region based on the overlapping portion of the non-detection sections Su1 and Su2.
As a result, the stationary object existence area is estimated based on both the section from the observation point P1 in which the target 202 is not detected and the section from the observation point P2 in which the target 202 is not detected, so that the stationary object existence area can be limited.

(3) When a plurality of discontinuous detection sections Sd11 and Sd12 are detected at the observation point P1 and a plurality of discontinuous detection sections Sd21 and Sd22 are detected at the observation point P2, the section detection means 15b Of the pair of detection intervals adjacent to each other across the part, the first detection intervals Sd11 and Sd21 relatively close to the observation points P1 and P2 and the second detection intervals Sd12 and Sd22 relatively far from the observation points P1 and P2 are selected. To detect. The region estimating means 17b detects the position of one of the endpoints Pe11 and Pe21 of the first detection section that is closer to the second detection section from a plurality of observation points P1 and P2. Determine that it exists.
This can prevent a moving object on the road 200 from being erroneously detected as a stationary object.

(Modification)
(1) Based on map information having known positional information of the target 202, the detected position (especially the position in the road width direction) of the target 202 detected by the external sensor 10 from the observation point may be corrected. For example, the detected position may be corrected to the position of the target 202 on the map that is closest to the detected position of the external sensor 10 . This makes it possible to detect the position of the detection section more accurately. The same applies to second to fourth embodiments described below.

(2) The section detection means 15b may detect targets outside the road boundary line of the road 200 as targets extending along the road 200 or existing continuously along the road 200 . For example, the section detection means 15b acquires the position information of the road boundary line from the map database, and detects the object detected by the external sensor 10 when the line segment connecting the observation point and the detection position of the external sensor 10 crosses the road boundary line. A landmark may be detected as a target that extends along the road 200 or exists in a row along the road 200 . As a result, a target that is blocked by the stationary object 201 on the road 200 can be selected. The same applies to second to fourth embodiments described below.

(3) The object recognition system 100 may include moving object detection means for detecting a moving object on the road 200 based on the detection signal from the external sensor 10 . The area estimating means 17b may estimate the stationary object existence area in an area where no moving object is detected on the road 200. FIG.
As mentioned above, moving objects are easier to detect than stationary objects. By estimating the stationary object existence area in an area where no moving object is detected, it is possible to prevent a moving object on the road 200 from being erroneously detected as a stationary object. The same applies to second to fourth embodiments described below.

(Second embodiment)
(composition)
Next, a second embodiment will be described. The object recognition system 100 of the second embodiment has the same functional configuration as the object recognition system 100 of the first embodiment, and the same constituent elements are given the same reference numerals.
The object recognition method of the second embodiment will be described with reference to FIGS. 5A to 5E.
First, as shown in FIG. 5A, the external sensor 10 detects objects around the first vehicle 1 from a point P1.

The section detection means 15b detects the first detection section Sd11 and the second detection section Sd12 based on the detection result by the external sensor 10. FIG.
The difference detection means 16b converts the coordinate system of the detection result of the section detection means 15b into the common coordinate system based on the measurement result of the current position of the first vehicle 1 by the positioning device 11, and determines the starting point Ps12 of the second detection section Sd12. and the point P1 is calculated as the line of sight boundary line 210 .

See FIG. 5B. The positioning device 11 detects that the first vehicle 1 has reached a point P2 that is a predetermined distance D away from the point P1.
The external sensor 10 detects objects around the first vehicle 1 from the point P2. The section detection means 15b detects the first detection section Sd21 and the second detection section Sd22 based on the detection result by the external sensor 10. FIG.

The difference detection means 16b converts the coordinate system of the detection result of the section detection means 15b into the common coordinate system based on the measurement result of the current position of the first vehicle 1 by the positioning device 11, and determines the starting point Ps22 of the second detection section Sd22. and the point P2 is calculated as the line of sight boundary line 211 .
The difference detection means 16b determines whether there is a difference between the second detection section Sd12 detected at the point P1 and the second detection section Sd22 detected at the point P2 based on the line of sight boundary line 210 and the line of sight boundary line 211. determine whether

When the first detection section Sd11 and the second detection section Sd12 are separated by the stationary object 201 on the road, and the first detection section Sd21 and the second detection section Sd22 are separated, the second detection section Sd12 and the second detection section Sd22 are separated. A difference occurs between the two detection intervals Sd22, so that the line of sight boundary 210 and the line of sight boundary 211 have only one point of intersection Pc, as shown in FIG. 5B. The position of the intersection point Pc coincides with the position of one of the vertices of the stationary object 201 .
On the other hand, when the stationary object 201 does not exist and the target 202 itself is simply interrupted, there is no difference between the second detection section Sd12 and the second detection section Sd22, and the line of sight boundary 210 and the line of sight boundary The intersection of lines 211 does not exist.

Therefore, the difference detection means 16b determines that there is a difference between the second detection section Sd12 and the second detection section Sd22 when the line of sight boundary line 210 and the line of sight boundary line 211 have only one intersection point Pc. Specifically, it is determined that there is a difference between the starting point Ps12 of the second detection section Sd12 and the starting point Ps22 of the second detection section Sd22. If there is no single intersection point Pc between the line of sight boundary line 210 and the line of sight boundary line 211, it is determined that there is no difference between the second detection section Sd12 and the second detection section Sd22.

See FIG. 5C. When there is a difference between the second detection section Sd12 and the second detection section Sd22, the region estimation means 17b detects a section in which the target 202 cannot be detected from either of the points P1 and P2 (that is, the first detection section Sd21 and the second detection interval Sd22) is estimated as an object presence possibility interval. If there is no difference between the second detection section Sd12 and the second detection section Sd22, it is determined that the stationary object 201 does not exist.
The region estimating means 17b determines that the stationary object 201 may be determined to exist.

The region estimating means 17b limits the object existence possibility section in the road width direction to estimate the stationary object existence region. See FIG. 5D. For example, the region estimating means 17b includes a line segment connecting the starting point Ps22 of the second detection section Sd22 and the common end point Pe of the first detection sections Sd11 and Sd21, a line segment connecting the common end point Pe and the intersection point Pc, A region surrounded by a line segment connecting the intersection point Pc and the start point Ps22 may be estimated as the stationary object existence region.

See FIG. 5E. For example, the region estimating means 17b obtains an intersection point Pv between the line of sight boundary line 211 and a perpendicular line 204 perpendicularly intersecting the first detection section Sd21 at the common end point Pe. The area estimating means 17b estimates an area surrounded by a line segment connecting the end point Pe and the intersection point Pv, a line segment connecting the intersection point Pv and the start point Ps22, and a line segment connecting the start point Ps22 and the end point Pe as a stationary object presence area. may

(motion)
Next, an example of the object recognition method of the second embodiment will be described with reference to FIG.
The processing of step S10 is the same as the processing of step S1 in FIG.
In step S11, the difference detection means 16b calculates a line of sight boundary line 210 connecting the point P1 and the start point Ps12 of the second detection section Sd12.

The process of step S12 is the same as the process of step S2 in FIG.
In step S13, the difference detection means 16b calculates the line of sight boundary line 211 connecting the point P2 and the start point Ps22 of the second detection section Sd22.
In step S14, the difference detection means 16b detects the second detection section Sd12 detected at the point P1 and the second detection section Sd12 detected at the point P2 depending on whether or not the only intersection point Pc of the boundary lines 210 and 211 can be calculated. It is determined whether or not there is a difference from Sd22.

If there is a difference between the second detection section Sd12 and the second detection section Sd22 (step S14: Y), the process proceeds to step S15. If there is no difference between the second detection section Sd12 and the second detection section Sd22 (step S14: N), the processing ends without estimating the object existence possibility section.
The processing of step S15 is the same as that of step S4 in FIG.
In step S16, the region estimating means 17b limits the object existence possibility section in the road width direction to estimate the stationary object existence region. Processing then ends.

Next, an example of a data exchange sequence in the object recognition system 100 will be described with reference to FIG.
At time t1, the external sensor 10 detects objects around the first vehicle 1 at the point P1. The positioning device 11 also measures the current position of the first vehicle 1 . The external sensor 10 and the positioning device 11 add data transmission time information to the sensor signal of the external sensor 10 (for example, data captured by an in-vehicle camera), the position data of the point P1, and the measurement time data, and send it to the section detection means 15b. Send. At time t2, the section detection means 15b receives these data.

The section detection means 15b detects the first detection section Sd11 and the second detection section Sd12 by time t3.
At time t3, the section detection means 15b detects the measurement time, the position data of the point P1, the data of the first detection section Sd11 and the second detection section Sd12 (in each section, two or more point groups including the start and end points are linked). combined data) is sent to the difference detection means 16b.

At time t4, the difference detection means 16b receives these data. The difference detection means 16b calculates a line of sight boundary line 210 connecting the point P1 and the starting point Ps12 of the second detection section Sd12.
At time t5, the positioning device 11 detects that the first vehicle 1 has reached the point P2.
At time t6, the external sensor 10 detects objects around the first vehicle 1 at the point P2. The external sensor 10 and the positioning device 11 add the data transmission time information to the sensor signal of the external sensor 10 (for example, data captured by an in-vehicle camera), the position data of the point P2, and the measurement time data, and send it to the section detection means 15b. Send.

At time t7, the section detection means 15b receives these data.
The section detection means 15b detects the first detection section Sd21 and the second detection section Sd22 by time t8.
At time t8, the section detection means 15b detects the measurement time, the position data of the point P2, the data of the first detection section Sd21 and the second detection section Sd22 (in each section, two or more point groups including the start and end points are linked). combined data) is sent to the difference detection means 16b.

At time t9, the difference detection means 16b receives these data. The difference detection means 16b calculates a line of sight boundary line 211 connecting the point P2 and the start point Ps22 of the second detection section Sd22.
At time t10, the difference detection means 16b detects the measurement time, the position data of the point P2, the data of the second detection section Sd22, and the data of the line-of-sight boundary lines 210 and 211 (two or more point groups including the start and end points are connected by links). data) and the position data of the common end point Pe are transmitted to the region estimation means 17b.

In order to further limit the stationary object existence area, the data of the first detection sections Sd11 and Sd21 (data obtained by linking two or more point groups including the start point and end point in each section) may be additionally transmitted. .
At time t11, the region estimating means 17b receives these data and estimates the stationary object existence region.

When the sensor signal of the external sensor 10 contains image data of a distant object (for example, one or more of the point groups for expressing the road boundary line is located at 30 m or more), uncompressed or lossless compression processing is performed. The processed data may be transmitted, and compressed images with demodulation identification information may be transmitted for other image areas. Also, the position coordinates may be expressed using latitude and longitude handled by GNSS, the distance from the nearest intersection, or a planar rectangular coordinate system (specified in Ministry of Land, Infrastructure, Transport and Tourism Notification No. 9, etc., 2002). Time information distributed from GNSS may be used as the time information.

(Effect of Second Embodiment)
(1) The difference detection means 16b detects line-of-sight boundaries 210, which are line segments connecting the observation points P1 and P2 with the endpoints Ps12 and Ps22, which are closer to the observation points P1 and P2 than the two ends of the second detection sections Sd12 and Sd22. and 211 are calculated for a plurality of observation points P1 and P2, respectively. The difference detecting means 16b and the area estimating means 17b estimate the stationary object existing area based on the line of sight boundary lines 210 and 211 calculated at the plurality of observation points P1 and P2.
In this way, by estimating the stationary object existence area based on changes in the line of sight boundaries 210 and 211 caused by detecting the second detection sections Sd12 and Sd22 from the plurality of observation points P1 and P2, the stationary object existence area can improve the estimation accuracy of

(2) The difference detection means 16b and the area estimation means 17b estimate the stationary object presence area based on the intersection point Pc of the line of sight boundary lines 210 and 211 calculated at the plurality of observation points P1 and P2. Since the position of one vertex of the stationary object 201 can be obtained from the intersection point Pc, the accuracy of estimating the stationary object existence area can be further improved.

(3) When a plurality of discontinuous detection sections are detected at each of the observation points P1 and P2, the section detection means 15b compares a pair of detection sections adjacent to each other across the discontinuity between the observation points. First detection sections Sd11 and Sd21 that are closer to the target, and second detection sections Sd12 and Sd22 that are relatively far from the observation point are detected. The region estimating means 17b determines the line of sight boundary line 211 and a perpendicular line 204 perpendicularly intersecting the first detection section Sd21 at one end point Pe near the second detection sections Sd12 and Sd22 of both ends of the first detection sections Sd11 and Sd21. Based on the intersection point Pv, the stationary object existence area is estimated.

As a result, when the observation point P2 is between the observation point P1 and the stationary object 201, the endpoints Pe11 and Pe21 of the first detection sections Sd11 and Sd21 become the common endpoint Pe. can be determined. Further, by finding the intersection point Pv between the perpendicular 204 and the line of sight boundary line 211, it is possible to determine the neighboring point of the other end point of the stationary object 201. FIG. As a result, the estimated presence area of the stationary object 201 can be further limited.

(Modification)
Similar to the modified example of the first embodiment described above, based on the map information having the known position information of the target 202, the detection position of the target 202 detected by the external sensor 10 from the observation point (especially the position in the road width direction) ) may be corrected. Thereby, the line-of-sight boundaries 210 and 211 can be corrected based on the map information.
This makes it possible to detect the position of the detection section more accurately. The same applies to third and fourth embodiments described below.

(Third embodiment)
(composition)
Next, a third embodiment will be described. In the object recognition system 100 of the third embodiment, sensors mounted on a plurality of vehicles detect the same object that extends along the road or that exists continuously along the road from a plurality of points. Each of the sensors mounted on a plurality of vehicles may detect the same object at the same time, or may detect the same object at different times.

Please refer to FIG. In addition to the configuration of the object recognition system 100 of the first embodiment shown in FIG. 1, the object recognition system 100 of the third embodiment includes an external sensor 20 mounted on the second vehicle 2, a positioning device 21, and a communication device 22.
The external sensors 20 comprise multiple different types of sensors for detecting objects around the second vehicle 2 . The positioning device 21 has a global positioning system (GNSS) receiver, receives radio waves from a plurality of navigation satellites, and measures the current time and the current position and traveling direction of the second vehicle 2 .

The external sensor 20 may have the same configuration as the external sensor 10 of the first vehicle 1, for example, and the positioning device 21 may have the same configuration as the positioning device 11 of the first vehicle 1, for example.
The communication device 22 distributes the reference time to the external sensor 20 based on the current time information obtained from the positioning device 21 . The communication device 22 also transmits and receives data to and from the section detection means 15b, the difference detection means 16b and the area estimation means 17b via the communication means 15a, 16a and 17a of the computers 15, 16 and 17, respectively.

Note that the computers 15 to 17 may be mounted on the second vehicle 2. In this case, the communication device 22 and the communication means 15a to 17a may exchange data by wired or wireless communication means. For example, the communication device 22 and the communication means 15a to 17a may be provided with an in-vehicle communication network (in-vehicle LAN) such as CSMA/CA multiplex communication (CAN: Controller Area Network) or Flex Ray.

Computers 15 - 17 may be mounted at locations other than second vehicle 2 . In this case, for example, the communication device 22 and the communication means 15a to 17a exchange data with each other via wireless communication means such as vehicle-to-vehicle communication, public mobile phone network, road-to-vehicle communication, satellite communication, and the like.
The computers 15 to 17 may be installed, for example, in other vehicles such as the first vehicle 1, mobile edge computers on a mobile phone/private network, road traffic infrastructure servers installed on the roadside, or cloud servers on the Internet.

When observing simultaneously from a plurality of points by the first vehicle 1 and the second vehicle 2, the communication device 12 of the first vehicle 1 and the communication device 22 of the second vehicle 2 can be directly connected by low-delay wireless communication. good. The wireless system may establish a communication connection between the first vehicle 1 and the second vehicle 2 using, for example, the DSRC system defined by IEEE 802.11p or the cellular V2X system defined by 3GPP Release 14 or later.

See FIG. The object recognition system 100 of the third embodiment has the same functional configuration as the object recognition systems 100 of the first and second embodiments, and the same constituent elements are given the same reference numerals.
The object recognition system 100 mounts section detection means 15b1 and 15b2 having the same function as the section detection section 15b of the first and second embodiments in the first vehicle 1 and the second vehicle 2, respectively. The object recognition system 100 also includes a map database 18a and road structure correction means 18b. In FIG. 9, the map database is denoted as "map DB". The map database 18a and the road structure correction means 18b may be mounted on the first vehicle 1 or the second vehicle 2, and other vehicles, mobile phones/mobile edge computers on private network, road traffic installed on the roadside. It may be implemented in either an infrastructure server or a cloud server on the Internet.

External sensors 10 and 20 detect objects around the first vehicle 1 and the second vehicle 2, respectively. The positioning devices 11 and 21 measure the current positions of the first vehicle 1 and the second vehicle 2, respectively.
See FIG. 10A. A first vehicle 1 is located at a point P1 and a second vehicle 2 is located at a point P2. The second vehicle 2 may be, for example, an oncoming vehicle of the first vehicle 1 or a crossing vehicle. The “intersecting vehicle” of the first vehicle 1 means a vehicle traveling on a route that intersects with the route of the first vehicle 1 .
The object recognition system 100 of the third embodiment estimates a stationary object existence area existing between points P1 and P2 where the first vehicle 1 and the second vehicle 2 are located.

The section detection means 15b1 of the first vehicle 1 detects a first detection section Sd11, a second detection section Sd12, an end point Pe11 of the first detection section Sd11, and a second detection section Sd12 based on the detection result of the external sensor 10 at the point P1. is detected, and the non-detection section Su1 is estimated. The communication device 12 of the first vehicle 1 transmits these data to the difference detection means 16b.

The section detection means 15b2 of the second vehicle 2 detects a first detection section Sd21, a second detection section Sd22, an end point Pe21 of the first detection section Sd21, and a second detection section Sd22 based on the detection result of the external sensor 20 at the point P2. is detected, and the non-detection section Su2 is estimated. The communication device 12 of the second vehicle 2 transmits these data to the difference detection means 16b.

The difference detection means 16b detects the difference between the second detection section Sd12 detected at the point P1 and the first detection section Sd21 detected at the point P2. For example, the difference between the start point Ps12 of the second detection section Sd12 and the end point Pe21 of the first detection section Sd21 is detected.
Alternatively, the difference between the first detection section Sd11 detected at the point P1 and the second detection section Sd22 detected at the point P2 is detected. For example, the difference between the end point Pe11 of the first detection section Sd11 and the start point Ps22 of the second detection section Sd22 is detected.

Further, the difference detection means 16b calculates a line of sight boundary line 210 connecting the point P1 and the starting point Ps12 of the second detection section Sd12. Also, a line of sight boundary line 211 connecting the point P2 and the start point Ps22 of the second detection section Sd22 is calculated.
The difference detection means 16b transmits the detection result of the difference in the detection section and the calculated line-of-sight boundary lines 210 and 211 to the area estimation means 17b.

The area estimation means 17b determines that the stationary object 201 exists between the points P1 and P2 when there is a difference between the second detection section Sd12 and the first detection section Sd21. Alternatively, when there is a difference between the first detection section Sd11 and the second detection section Sd22, it is determined that the stationary object 201 exists between the points P1 and P2.

See FIG. 10B. When it is determined that the stationary object 201 exists, the area estimation means 17b estimates a section in which the target 202 is not detected from either of the points P1 and P2 as an object existence possibility section. For example, the section between the end points Pe11 and Pe21 of the first detection sections Sd11 and Sd21 (that is, the overlapping area of the non-detection section Su1 and the non-detection section Su2) is estimated as the object presence possibility section.
See FIG. 10C. The area estimating means 17b estimates the stationary object existence area 218 by restricting the object existence possibility section in the road width direction.

For example, the region estimating means 17b obtains an intersection point Pv1 between the line of sight boundary line 211 and a perpendicular line 216 perpendicularly crossing the first detection section Sd11 at the end point Pe11. Also, an intersection point Pv2 between the line of sight boundary line 210 and the perpendicular line 217 perpendicularly crossing the first detection section Sd21 at the end point Pe21 is obtained. A region surrounded by a line segment connecting the end point Pe11 and the intersection point Pv1, a line segment connecting the intersection point Pv1 and the intersection point Pv2, a line segment connecting the intersection point Pv2 and the end point Pe21, and a line segment connecting the end point Pe21 and the end point Pe11 is defined as a stationary object existence region. estimated as

See FIG. After the stationary object existence area is estimated, the road structure correction means 18b reads the position information of the target object 202 in the section where the stationary object existence area exists from the map database 18a.
The map database 18a stores map information such as the shape of roads, the positions of targets such as features and landmarks, and section information where targets exist. The targets stored in the map database 18a include targets 202 that extend along the road 200 or exist in a row along the road 200. FIG. As the map database 18a, for example, high-precision map data suitable as a map for autonomous driving may be stored.

Based on the position information of the target 202 read from the map database 18a, the road structure correction means 18b determines the undetectable range, which is the range in which the target 202 cannot be detected from the point P1 or P2 due to the road shape and road structure. Calculate
In such a range, the area estimator 17b may erroneously estimate the stationary object existence area. Therefore, the road structure correction means 18b corrects the estimation result of the stationary object existence area so as not to estimate the stationary object existence area in the undetectable range. Specifically, when the stationary object existence area includes the undetectable range, the estimation of the stationary object existence area is cancelled.

See FIG. 11A. For example, it is assumed that a guardrail or a curbstone is detected as the target object 220 extending along the road, and that there is a curved section ahead of the route of the first vehicle 1 located at the point P1. In such a case, part of the target 220 in the curve section is in the blind spot when viewed from the point P1. A line segment 221 indicates the boundary line between the line-of-sight range and the non-line-of-sight range in the curve section viewed from the point P1.
The road structure correction means 18b calculates the range of the target object 220 that is not detected by the external sensor 10 in the curve section as the undetectable range. If the stationary object existence area includes an undetectable range, the estimation of the stationary object existence area is cancelled.

See FIG. 11B. For example, it is assumed that a road wall is detected as a target 230 extending along the road, and an intersection is ahead of the route of the first vehicle 1 located at the point P1. The target 230 does not exist within the intersection, and part of the target 230 along the cross road is in the blind spot when viewed from the point P1. A line segment 231 indicates the boundary line between the line-of-sight range and the non-line-of-sight range at the intersection viewed from the point P1.
The road structure correction means 18b calculates the range of the target object 220 that is not detected by the external sensor 10 at the intersection as the undetectable range. If the stationary object existence area includes an undetectable range, the estimation of the stationary object existence area is canceled. For example, if the position of the stationary object existence area is equal to the position of the intersection on the map, and the length of the stationary object existence area is close to the width of the intersecting road, the estimation of the stationary object existence area is canceled.

See FIG. 11C. For example, it is assumed that the road surface is detected as a target 240 extending along the road, and that there is an elevation difference (concave portion) in front of the route of the first vehicle 1 located at the point P1. In such a case, part of the concave portion of the road surface is in the blind spot when viewed from the point P1.
The road structure correction means 18b calculates the range of the road surface 240 that is not detected by the external sensor 10 due to the difference in road height as the undetectable range. If the stationary object existence area includes an undetectable range, the estimation of the stationary object existence area is cancelled.
See FIG. The trajectory generation means 110 generates a target travel trajectory and a speed profile so as to avoid the stationary object existence areas that have not been canceled by the road structure correction means 18b among the stationary object existence areas estimated by the area estimation means 17b.

Note that when the section detection means 15b1 and 15b2 detect a road white line as a target extending along the road, the end points of the broken road white line or the road white line having periodic missing portions are detected as the first detection section. Alternatively, the detection result of the end point of the road white line may be corrected so as not to be detected as the end point of the second detection section.
See FIG. 11D. When the interval between the end points of the road white line 250 is a constant interval T, the section detection means 15b1 and 15b2 detect the end points of the road white line existing at the constant interval T as the end points of the first detection section or the second detection section. prohibited. The section detection means 15b1 and 15b2 detect the end points 251 and 252 as the end points of the first detection section and the second detection section when the difference between the interval L between the end points 251 and 252 of the road white line and the constant interval T is larger than a predetermined threshold. Detect as

(motion)
Next, an example of the object recognition method of the third embodiment will be described with reference to FIG.
The processing of steps S30 to S33 is the same as the processing of steps S10 to S13 in FIG.
In step S34, the difference detection means 16b determines the difference between the detection section detected at the point P1 and the detection section detected at the point P2.

If there is a difference between the detected section detected at point P1 and the detected section detected at point P2 (step S34: Y), the process proceeds to step S35. If there is no difference (step S34: N), the process ends without estimating the object existence possibility interval.
In step S35, the region estimating means 17b estimates a section in which the target 202 is not detected from either of the points P1 and P2 as an object presence possibility section.

In step S36, the region estimating means 17b limits the object existence possibility section in the road width direction to estimate the stationary object existence region.
In step S37, the road structure correction means 18b determines whether there is an undetectable range in which the target object 202 cannot be detected due to the road shape or road structure within the stationary object presence area estimated by the area estimation means 17b. determine whether or not

If there is an undetectable range within the stationary object presence area (step S37: Y), the process proceeds to step S39. If there is no undetectable range within the stationary object presence area (step S37: N), the process proceeds to step S38.
In step S38, the road structure correction means 18b determines that a stationary object exists within the stationary object existence area. The trajectory generation means 110 determines whether or not the first vehicle 1 traveling on the target travel trajectory interferes with the stationary object existence area, and performs travel control of the first vehicle 1 so as to avoid the stationary object existence area. Determine if it is necessary.

When the stationary object existence area is to be avoided, the trajectory generation means 110 generates a target travel trajectory and a speed profile for avoiding the stationary object existence area. Processing then ends.
On the other hand, in step S39, the road structure correction means 18b determines that a stationary object exists within the stationary object existence area. The stationary object presence area estimated by the area estimation means 17b is cancelled, and the process is terminated.

(Effect of the third embodiment)
(1) The road structure correction means 18b calculates an undetectable range, which is a range in which targets are not detected from the observation points P1 and P2, based on map information having positional information of targets along the road. The road structure correction means 18b prohibits the estimation of the stationary object existence area in the undetectable range.
As a result, it is possible to prevent erroneous detection of a section along the road where a target object cannot be detected as a stationary object existing area regardless of the presence or absence of a stationary object.

(2) External sensors 10 and 20 detect road markings 250 as targets extending along the road. The section detection means 15b1 and 15b2 do not detect the end points of the road division lines 250 existing at regular intervals as the end points of the detection section.
As a result, it is possible to prevent erroneous detection of a stationary object existing area in a section where a broken road white line or a road white line having periodic missing portions is drawn.

(3) The external sensor 10 of the first vehicle 1 detects the target 202 from the point P1, which is one of the plurality of observation points, and detects the target 202 outside the second vehicle 2, which is an oncoming vehicle or an intersecting vehicle of the first vehicle 1. The sensor 20 detects a target from point P2, which is another one of the plurality of observation points.
As a result, blind spots can be eliminated and the stationary object existence area can be detected with higher accuracy.

(4) The external sensor 10 of the first vehicle 1 detects the target object 202 from the point P1, which is one of the plurality of observation points, and detects the target object 202 outside the second vehicle 2, which is an oncoming vehicle or an intersecting vehicle of the first vehicle 1. The sensor 20 detects a target from point P2, which is another one of the plurality of observation points.
When a plurality of discontinuous detection sections are detected at each of the points P1 and P2, the section detection means 15b1 and 15b2 compare points P1 and P2 among a pair of detection sections adjacent to each other across the discontinuous portion. First detection sections Sd11 and Sd21 close to the point, second detection sections Sd12 and Sd22 relatively far from the points P1 and P2, and one of both ends of the first detection sections Sd11 and Sd21 closer to the second detection sections Sd12 and Sd22. First detection section end points Pe11 and Pe21, which are end points, are detected.
The region estimating means 17b estimates a stationary object existing region based on the first detected section end points Pe11 and Pe21 detected by the first vehicle 1 and the second vehicle 2, respectively.
As a result, the positions of both ends of the stationary object 201 can be specified, so that the stationary object existence area can be detected with higher accuracy.

(Modification)
The section detection means 15b, the difference detection means 16b, and the area estimation means 17b use the same method as the object recognition method of the first and second embodiments to perform the first detection detected by the second vehicle 2 from a plurality of different points. A stationary object existence area may be estimated based on the section and the second detection section. The area estimation means 17b may transmit the estimated stationary object existence area to the first vehicle 1 . The first vehicle 1 may be a vehicle located farther from the stationary object existence area than the second vehicle 2, for example.
As a result, based on the information of the stationary object existence area estimated by the second vehicle 2 , it is possible to carry out travel control of the subsequent first vehicle 1 .

(Fourth embodiment)
(composition)
Next, a fourth embodiment will be described. The object recognition system 100 of the fourth embodiment estimates stationary object existence areas in the same area at different times, and compares the stationary object existence areas at different times. If the stationary object existence area does not change with time, it is determined that the stationary object exists in the stationary object existence area. When the stationary object existence area changes with time, it is determined that the train temporarily stopped exists in the stationary object existence area.

Please refer to FIG. In addition to the configuration of the object recognition system 100 of the third embodiment shown in FIG. 8, the object recognition system 100 of the fourth embodiment includes an external sensor 30 mounted on the third vehicle 3, a positioning device 31 and a communication device 32.
The external sensors 30 comprise a plurality of different types of sensors for detecting objects around the third vehicle 3 . The positioning device 31 includes a global positioning system (GNSS) receiver, receives radio waves from a plurality of navigation satellites, and measures the current time and the current position and traveling direction of the third vehicle 3 .

The external sensor 30 may have the same configuration as the external sensor 20 of the second vehicle 2, for example, and the positioning device 31 may have the same configuration as the positioning device 21 of the second vehicle 2, for example.
The communication device 32 distributes the reference time to the external sensor 30 based on the current time information obtained from the positioning device 31 . The communication device 32 also transmits and receives data to and from the section detection means 15b, the difference detection means 16b and the area estimation means 17b via the communication means 15a, 16a and 17a of the computers 15, 16 and 17, respectively.

Note that the computers 15 to 17 may be installed in the third vehicle 3. In this case, the communication device 32 and the communication means 15a to 17a may exchange data by wired or wireless communication means. For example, the communication device 32 and the communication means 15a to 17a may be provided with an in-vehicle communication network (in-vehicle LAN) such as CSMA/CA multiplex communication (CAN: Controller Area Network) or Flex Ray.

Computers 15 - 17 may be mounted at locations other than third vehicle 3 . In this case, for example, the communication device 32 and the communication means 15a to 17a exchange data with each other via wireless communication means such as vehicle-to-vehicle communication, public mobile phone network, road-to-vehicle communication, and satellite communication.
The computers 15 to 17 are implemented in, for example, other vehicles such as the first vehicle 1 and the second vehicle 2, a mobile edge computer on a mobile phone/private network, a road traffic infrastructure server installed on the roadside, or a cloud server on the Internet. may be

Please refer to FIG. The object recognition system 100 of the fourth embodiment has functional configurations similar to those of the object recognition systems 100 of the first and second embodiments, and the same components are denoted by the same reference numerals.
The object recognition system 100 mounts section detection means 15b2 and 15b3 having the same function as the section detection section 15b of the first and second embodiments in the second vehicle 2 and the third vehicle 3, respectively.

The object recognition system 100 also includes a stop train determination means 19 . The stop train determination means 19 may be mounted on the first vehicle 1 to the third vehicle 3, other vehicles, mobile phones/mobile edge computers on private networks, road traffic infrastructure servers installed on the roadside, or It may be implemented on any cloud server on the Internet.

External sensors 20 and 30 detect objects around the second vehicle 2 and the third vehicle 3, respectively. The positioning devices 21 and 31 measure the current positions of the second vehicle 2 and the third vehicle 3, respectively.
The section detecting means 15b2, the difference detecting means 16b, and the area estimating means 17b estimate the stationary object existing area based on the detection result of the external sensor 20 of the second vehicle 2 and the measurement result of the positioning device 21. FIG.
The section detecting means 15b3, the difference detecting means 16b, and the area estimating means 17b estimate the stationary object existence area based on the detection result of the external sensor 30 of the third vehicle 3 and the measurement result of the positioning device 31. FIG.

15A and 15B respectively show stationary object existence regions estimated based on measurement results obtained from the second vehicle 2 and the third vehicle 3 that passed through the same region at different times. The third vehicle 3 passes through the same area as the second vehicle 2 with a delay of a predetermined time t (for example, 30 seconds).
Outlined rectangles represent targets along the road. Hatched regions r1 to r4 represent stationary object existence regions.
As the external sensors 20 and 30, side sensors for detecting targets on the sides of the second vehicle 2 and the third vehicle 3 are provided. A region in which does not detect a target along the road may be estimated as a stationary object existence region.

The stationary train determination means 19 compares the stationary object existence areas estimated in the same area at different times, and determines whether the stationary object existence area changes with time. If the stationary object existence area does not change with time, it is determined that the stationary object exists in the stationary object existence area. When the stationary object existence area changes with time, it is determined that the train temporarily stopped exists in the stationary object existence area.

In the examples shown in FIGS. 15A and 15B, the position of the stationary object existence region r1 does not change even after the predetermined time t has passed. Also, the section s1 between the stationary object existence areas r1 and r2 is extended, and the position of the stationary object existence area r2 is changed. Similarly, the changes in the positions of r3 and r4 are changed, and the section s2 between the stationary object existence areas r2 and r3 and the section s3 between the stationary object existence areas r3 and r4 are also extended.

The stopped train determination means 19 detects the stationary object existence areas r1 to r4 detected in the specific area and one or more of the sections s1 to s3 therebetween. R4 is determined to be a stopped train, and the stationary object presence region estimated by the region estimation means 17b is cancelled.
When none of the stationary object existence areas r1 to r4 and the sections s1 to s3 change with time, it is determined that the stationary object exists in the stationary object existence areas r1 to r4, and the position information of the stationary object existence areas r1 to r4 is set to the first It is transmitted to the in-vehicle device 13 of the vehicle 1 .

The trajectory generation means 110 of the in-vehicle device 13 generates a target travel trajectory and a speed profile so as to avoid the stationary object existence areas r1 to r4 transmitted from the stop train determination means 19. FIG.
The first vehicle 1 may be a vehicle located farther from the stationary object existence regions r1 to r4 than the second vehicle 2 and the third vehicle 3, for example. As a result, based on the information of the stationary object existing area estimated by the second vehicle 2 and the third vehicle 3, the traveling control of the following first vehicle 1 can be performed.

(motion)
Next, an example of the object recognition method of the fourth embodiment will be described with reference to FIG. The processes of steps S40 to S46 are the same as steps S10 to S16 of the second embodiment shown in FIG. 6, and are executed for each of a plurality of vehicles (eg, second vehicle 2 and third vehicle 3).
In step S47, the stopped train determination means 19 compares the stationary object existence areas estimated in the same area by a plurality of vehicles at different times, and determines whether or not the stationary object existence area changes with time.

If the stationary object existence area changes with time (step S47: Y), the process proceeds to step S49. If the stationary object presence area does not change over time (step S47: N), the process proceeds to step S48.
In step S48, the stopped train determination means 19 determines that a stationary object exists within the stationary object existence area. The trajectory generation means 110 determines whether or not the first vehicle 1 traveling on the target travel trajectory interferes with the stationary object existence area, and performs travel control of the first vehicle 1 so as to avoid the stationary object existence area. Determine if it is necessary.

When the stationary object existence area is to be avoided, the trajectory generation means 110 generates a target travel trajectory and a speed profile for avoiding the stationary object existence area. Processing then ends.
In step S49, the stopped vehicle train determination means 19 determines that the stopped vehicle train exists in the stationary object existence area, and cancels the stationary object existence area estimated by the area estimation means 17b. Processing then ends.

(Effect of the fourth embodiment)
The stopped train determination means 19 estimates stationary object existence areas at different times, and determines that a stationary object exists in the stationary object existence areas when there is no change over time in the stationary object existence areas. As a result, it is possible to prevent erroneous detection of a temporarily stopped object, such as a stopped vehicle, as a stationary object.

DESCRIPTION OF SYMBOLS 1... 1st vehicle, 2... 2nd vehicle, 3... 3rd vehicle, 10... External sensor, 11... Positioning apparatus, 12... Communication apparatus, 13... In-vehicle apparatus, 14... Actuator, 15... Computer, 15a... Communication means , 15b, 15b1 to 15b3 section detection means 16 computer 16b difference detection means 17 computer 17a communication means 17b area estimation means 18a map database 18b road structure correction means 19 Stop train determination means 20 External sensor 21 Positioning device 22 Communication device 30 External sensor 31 Positioning device 32 Communication device 100 Object recognition system 110 Trajectory generation means 111 travel control means

Claims (14)

Detecting a target extending along the road from an observation point on the road with a sensor ,
the computer
detecting a detection section, which is a section in which a portion of the target detected from the observation point exists;
A line segment connecting one of the two ends of the detection section closer to the observation point to the observation point based on the positions of the end points of the plurality of different detection sections respectively detected at the plurality of observation points. calculating a boundary line for each of the plurality of observation points;
estimating a stationary object existence area, which is an area in which a stationary object exists on the road, based on the line of sight boundary calculated at the plurality of observation points;
An object recognition method characterized by: 2. The object recognition method according to claim 1, wherein the computer estimates the stationary object existence area based on intersections of the lines of sight calculated at the plurality of observation points. The computer is
estimating, for each of the plurality of observation points, a non-detection section, which is a section in which there is a portion of the target that is not detected from the observation point;
estimating the stationary object existence area based on the overlapping portion of the non-detection interval;
3. The object recognition method according to claim 1 or 2, characterized in that: The computer is
when a plurality of discontinuous detection intervals are detected at each of the observation points, a first detection interval relatively close to the observation point among a pair of the detection intervals adjacent across a discontinuous portion; detecting a second detection section relatively far from the observation point;
a perpendicular line perpendicularly crossing the first detection section at one end point near the second detection section among both ends of the first detection section; and one end point near the observation point among both ends of the second detection section. estimating the stationary object existence area based on the intersection with the line of sight boundary connecting the observation point;
3. The object recognition method according to claim 1 or 2, characterized in that: 3. The computer corrects the line of sight boundary by correcting the detected position of the target detected from the observation point based on map information having position information of the target. 5. The object recognition method according to 1 or 4. The computer is
calculating an undetectable range, which is a range in which the target is not detected from the observation point, based on map information having position information of the target;
not estimating the stationary object existence area in the undetectable range;
The object recognition method according to any one of claims 1 to 4, characterized in that: 6. The target is a road marking line extending along the road, and end points of the road marking line existing at regular intervals are not detected as end points of the detection section. The object recognition method according to any one of 1. 8. The object recognition method according to any one of claims 1 to 7, wherein a target outside a road boundary line of the road is detected by the sensor as the target. Detecting a target extending along the road from an observation point on the road with a sensor ,
the computer
detecting a detection section, which is a section in which a portion of the target detected from the observation point exists;
estimating a stationary object presence area, which is an area in which a stationary object exists on the road, based on the positions of the end points of a plurality of different detection sections detected at the plurality of observation points;
The target is detected from a first vehicle at one of the plurality of observation points, and the target is detected from a second vehicle, which is an oncoming vehicle or an intersecting vehicle of the first vehicle, at another one of the plurality of observation points. An object recognition method characterized by detecting Detecting a target extending along the road from an observation point on the road with a sensor ,
the computer
detecting a detection section, which is a section in which a portion of the target detected from the observation point exists;
estimating a stationary object presence area, which is an area in which a stationary object exists on the road, based on the positions of the end points of a plurality of different detection sections detected at the plurality of observation points;
The target is detected from a first vehicle at one of the plurality of observation points, and the target is detected from a second vehicle that is an oncoming vehicle of the first vehicle at another one of the plurality of observation points. ,
when a plurality of discontinuous detection intervals are detected at each of the observation points, a first detection interval relatively close to the observation point among a pair of the detection intervals adjacent across a discontinuous portion; detecting a second detection interval relatively far from the observation point and a first detection interval end point that is one end point closer to the second detection interval out of both ends of the first detection interval;
estimating the stationary object existence area based on the first detection section end points detected by the first vehicle and the second vehicle;
An object recognition method characterized by: The computer is
estimating each of the stationary object existence regions at different times;
determining that a stationary object exists in the stationary object existence area when the stationary object existence area does not change over time;
The object recognition method according to any one of claims 1 to 9, characterized in that: Detecting a target extending along the road from an observation point on the road with a sensor ,
the computer
detecting a detection section, which is a section in which a portion of the target detected from the observation point exists;
estimating a stationary object presence area, which is an area in which a stationary object exists on the road, based on the positions of the end points of a plurality of different detection sections detected at the plurality of observation points;
when a plurality of discontinuous detection intervals are detected at each of the observation points, a first detection interval relatively close to the observation point among a pair of the detection intervals adjacent across a discontinuous portion; detecting a second detection section relatively far from the observation point;
It is determined that a stationary object exists when the position of one of the ends of the first detection section, which is closer to the second detection section, does not change even when detected from the plurality of observation points. Object recognition method. 13. The object recognition method according to any one of claims 1 to 12, wherein the computer estimates the stationary object existence area in an area where no moving object is detected on the road. a sensor that detects a target extending along the road or present along the road from an observation point on the road;
Detecting a detection section, which is a section in which a portion of the target detected from the observation point exists, and determining the detection section based on the positions of the end points of a plurality of different detection sections detected at the plurality of observation points. A line of sight boundary line, which is a line segment connecting one of the two ends of the observation point near the observation point and the observation point, is calculated for each of the plurality of observation points, and the line of sight boundary line calculated at the plurality of observation points a computer for estimating a stationary object existence area, which is an area where a stationary object exists on the road, based on
An object recognition system comprising:

Images