JP2018084987A - Object recognition device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object recognition device capable of inhibiting a target of roadside object from being misdetected as a target in a lane such as a preceding vehicle.SOLUTION: An object recognition device is configured to: obtain target information on objects existing in the vicinity of an own vehicle, lane boundary line information on the position and shape of the lane boundary line of the road on which the own vehicle travels, and road shoulder information on the width of the road shoulder area in which the own vehicle travels; set an effective lane area inside the road shoulder side lane boundary line separating an adjacent lane adjacent to the road shoulder area from the road shoulder area on the basis of the lane boundary line information and the road shoulder information; determine whether or not an object specified in a target frame is an object located within the adjacent lane on the basis of the overlapping degree of the target frame included in the target information and the effective lane area; and set the effective lane area such that a margin from the road shoulder side lane boundary line to the effective lane area is narrowed as the width of the road shoulder area increases, and that the margin from the road shoulder side lane boundary line to the effective lane area is widen as the width of the road shoulder area decreases.SELECTED DRAWING: Figure 8

Description

  The present invention relates to an object recognition apparatus suitable for use in an automatic driving system.

  In Patent Document 1, when the traveling path of the host vehicle is estimated from the vehicle speed and the yaw rate, and a preceding vehicle exists on the estimated traveling path, it is determined that the preceding vehicle is traveling in the same lane as the host vehicle. A method is disclosed.

JP 2008-176400 A

  On roads with a narrow shoulder width, there is a case where a target when a roadside object outside the road is converted into the lane. In such a case, in the technique described in the above publication, there is a possibility that a roadside target is erroneously detected as a lane target such as a preceding vehicle.

  The present invention has been made in view of the above-described problems, and provides an object recognition device capable of reducing erroneous detection of a roadside target as a lane target such as a preceding vehicle. With the goal.

In order to achieve the above object, an object recognition apparatus according to the present invention includes:
Target information acquisition means for acquiring target information related to an object existing around the host vehicle;
Lane boundary information acquisition means for acquiring lane boundary information regarding the position and shape of the lane boundary of the road on which the host vehicle travels;
Shoulder information acquisition means for acquiring shoulder information related to the width of the shoulder region of the road on which the host vehicle travels;
Based on the lane boundary information and the road shoulder information, effective lane area setting means for setting an effective lane area inside a roadside lane boundary line that divides the adjacent lane adjacent to the road shoulder area and the road shoulder area. When,
Based on the degree of overlap between the target frame included in the target information and the effective lane area, the object specified by the target frame is an object located in the adjacent lane or located outside the adjacent lane. A determination means for determining whether the object is an object,
The effective lane area setting means reduces the margin from the roadside lane boundary line to the effective lane area as the width of the roadside area increases, and decreases from the roadside lane boundary line as the width of the roadside area decreases. The effective lane region is set so as to increase a margin to the effective lane region.

  According to the object recognition device of the present invention, the effective lane region and the object that are set inside the roadside lane boundary line that divides the roadside region and the adjacent lane adjacent to the roadside region and whose range is changed according to the roadside width. Since it is judged whether the object specified in the target frame is an object in the adjacent lane from the overlap with the target frame, the target frame when the roadside object outside the road is targeted on a road with a narrow shoulder width Even if it protrudes into the lane, it can be prevented from erroneously detecting it as being in the lane.

1 is a block diagram showing a configuration of an automatic driving system according to an embodiment. It is a figure explaining the function of a lane boundary line information acquisition part. It is a figure explaining the function of a road shoulder information acquisition part. It is a figure explaining the function of an effective lane area setting part. It is a figure explaining the function of an effective lane area setting part. It is a figure explaining the function of a lane inside / outside determination part. It is a figure explaining the function of a lane inside / outside determination part. It is a flowchart which shows the flow of the object recognition process which concerns on embodiment. It is a figure which shows the recognition result by the conventional object recognition process. It is a figure which shows the recognition result by the object recognition process which concerns on embodiment.

  Embodiments of the present invention will be described below with reference to the drawings. However, in the embodiment shown below, when referring to the number of each element, quantity, quantity, range, etc., unless otherwise specified or clearly specified in principle, the reference However, the present invention is not limited to these numbers. Further, the structures described in the embodiments described below are not necessarily essential to the present invention unless otherwise specified or clearly specified in principle.

  FIG. 1 is a diagram illustrating a configuration of an automatic driving system according to an embodiment. The automatic driving system performs automatic driving of a vehicle on which the automatic driving system is mounted. The automatic driving system 1 includes a control device 10 that functions as an object recognition device, a lidar (LIDAR: Laser Imaging Detection and Ranging) 2 that is an autonomous recognition sensor, a radar 3, and a stereo camera 4. These autonomous recognition sensors are connected to the control device 10 directly or via a communication network such as a CAN (Controller Area Network) built in the vehicle. The control device 10 can grasp the situation of substantially the entire periphery of the vehicle from the object information obtained by the autonomous recognition sensor.

  The automatic driving system 1 further includes a GPS (Global Positioning System) receiver 5, a map database 6, a navigation system 7, an HMI (Human Machine Interface) 8, and an actuator 9. The GPS receiver 5 is means for acquiring position information indicating the position of the host vehicle based on a signal transmitted from a GPS satellite. The map database 6 is formed in a storage such as an HDD or an SSD mounted on the vehicle, for example. The map information included in the map database 6 includes road position information, road shape information, intersection and branch point position information, road lane information, and the like. The navigation system 7 calculates the route traveled by the host vehicle based on the position information of the host vehicle measured by the GPS receiver 5 and the map information in the map database 6, and the calculated route information is transmitted via the HMI 8. While transmitting to a driver | operator, it outputs to the control apparatus 10. The HMI 8 is an interface for outputting and inputting information between the occupant and the control device 10. The actuator 9 is a device that operates according to an operation signal from the control device 10 and changes the traveling state of the vehicle by the operation. The actuator 9 is provided in each of a drive system, a braking system, and a steering system, for example. In addition to these, the automatic driving system 1 is provided with an internal sensor for obtaining information on the traveling state of the host vehicle, such as a vehicle speed sensor and an acceleration sensor.

  The control device 10 is an ECU (Electronic Control Unit) having at least one CPU, at least one ROM, and at least one RAM. The ROM stores various types of data including various programs and maps for automatic operation. Various functions are realized in the control device 10 by loading a program stored in the ROM into the RAM and executing it by the CPU. In addition, the control apparatus 10 may be comprised from several ECU.

  In FIG. 1, among the functions for automatic driving that the control device 10 has, in particular, functions related to object recognition are represented by blocks. The illustration of other functions for automatic operation of the control device 10 is omitted.

  The control device 10 has a function of recognizing an object existing in the vicinity of the host vehicle and discriminating between an object located in the lane and an object located outside the lane. This function is realized by the target information acquisition unit 11, the lane boundary information acquisition unit 12, the road shoulder information acquisition unit 13, the effective lane area setting unit 14, and the lane inside / outside determination unit 15 included in the control device 10. However, these units 11, 12, 13, 14, and 15 do not exist as hardware in the control device 10, but are realized by software when a program stored in the ROM is executed by the CPU. .

  The target information acquisition unit 11 acquires information on the position and range of each three-dimensional object existing around the host vehicle. Specifically, the target information acquisition unit 11 first acquires distance measurement data of a three-dimensional object around the host vehicle from an autonomous recognition sensor. The distance measurement data includes information regarding the distance to the distance measurement point, information regarding the height of the distance measurement point, and information regarding the direction of the distance measurement point. The target information acquisition unit 11 uses information obtained by the rider 2, uses information obtained by the radar 3, uses information obtained by the stereo camera 4, and information on a plurality of types of autonomous recognition sensors by sensor fusion. The distance measurement data can be acquired by at least one of the above methods.

  The target information acquisition unit 11 specifies the position of each ranging point in the traveling plane based on the acquired ranging data. As a coordinate system for expressing the distance measurement data, any orthogonal coordinate system can be used as long as it is an orthogonal coordinate system in which the origin of the coordinate system is placed on the host vehicle. For example, the center of the rear wheel axis of the vehicle at each time is the origin of the coordinate system, the X axis (vertical axis) and forward in the rear direction of the host vehicle, the Y axis (horizontal axis) and the right side in the left and right direction A positive coordinate system can be used. Alternatively, the left and right center of the front end of the host vehicle can be used as the coordinate system origin. Hereinafter, an orthogonal coordinate system in which ranging data is expressed is referred to as a reference coordinate system.

  The target information acquisition unit 11 then clusters the acquired distance measurement data. Clustering is performed based on the position and height of each ranging point. Further, it is performed so as to maintain continuity with the result of clustering in the previous frame. The specific method of clustering is not particularly limited, and a known clustering method can be used. The clustered distance measuring point group is surrounded by a rectangular target frame for each target in the reference coordinate system. The position of the target frame in the reference coordinate system indicates the position of the target relative to the host vehicle, and the range of the target frame indicates the range of the target on the plane where the host vehicle is located.

  The lane boundary information acquisition unit 12 acquires lane boundary information regarding the position and shape of the lane boundary of the road on which the host vehicle is traveling. For example, map data around the host vehicle is retrieved from the map database 6 based on the position and posture of the host vehicle estimated from the output of the GPS receiver 5. In the map data, the position of the point sequence constituting the lane boundary line is registered in the GPS coordinate system. The lane boundary information acquisition unit 12 reads the position data of the point sequence that constitutes the lane boundary from the map data, and performs coordinate conversion of the read position data from the GPS coordinate system to the reference coordinate system. As shown in FIG. 2, the position of each lane boundary line with respect to the host vehicle and the shape of each lane boundary line are specified by expressing the point sequence constituting the lane boundary line in the reference coordinate system.

  The shoulder information acquisition unit 13 acquires the shoulder information related to the width of the shoulder area of the road on which the host vehicle travels. The road shoulder region means a region between a road boundary line that divides the inside and outside of a road region where the vehicle can travel, and a road shoulder side lane boundary line that is closest to the road boundary line. There are two road boundaries on the left and right. In the example shown in FIG. 3, four lines are drawn, but the leftmost line is a road boundary line, and the right three lines are lane boundary lines. Since the lane is divided by each lane boundary line, the lane boundary line is indicated as a lane line in FIG. A belt-like area defined by two adjacent lane markings is one lane. The shoulder information acquisition unit 13 includes information on the position and shape of the road boundary line, the lane line that is paired with the road boundary line, that is, the position of the section line closest to the road boundary line (hereinafter referred to as the road shoulder side lane line) and Information on the shape is acquired from the map database 6. The shoulder information acquisition unit 13 calculates the difference in lateral position between the road boundary line and the roadside dividing line at each vertical position in the reference coordinate system. This difference is the width of the shoulder region in the vertical position, that is, the shoulder width. Although only the shoulder area on the left side of the road is illustrated in FIG. 3, the acquisition of the shoulder information by the shoulder information acquisition unit 13 is performed for the left and right shoulder areas.

  The lane boundary information acquired by the lane boundary information acquisition unit 12 and the road shoulder information acquired by the road shoulder information acquisition unit 13 are input to the effective lane area setting unit 14. The effective lane area setting unit 14 sets the effective lane area inside the roadside lane marking based on the lane boundary information and the roadside information. The effective lane region is a region where it can be determined that an object corresponding to the target frame is located in the lane if the target frame is included therein. The specific contents of the effective lane area and the setting method by the effective lane area setting unit 14 can be described with reference to FIG.

  In FIG. 4, two lane markings, that is, a left lane marking and a right lane marking that constitute a lane, are drawn. The positions and shapes of the left and right lane markings are included in the lane boundary information. Outside the left lane line, roadside objects are installed along the road. The roadside object is specifically a wall or a guardrail, and its inner surface serves as a road boundary line. Therefore, the area from the left lane marking to the roadside that is the road boundary is the road shoulder area. The distance from the left lane line, which is the roadside line, to the roadside object, that is, the road width is included in the roadside information.

  The effective lane area setting unit 14 calculates the position and shape of the lane center line based on the position and shape of the left lane line and the right lane line. Then, the left effective lane width is set from the lane center line toward the left lane line, and the right effective lane width is set from the lane center line toward the right lane line. The left effective lane width and the right effective lane width are set for each vertical position of the reference coordinate system. The left effective lane width and the right effective lane width are individually set. The region that falls within the left effective lane width and the right effective lane width is the above-described effective lane region.

  The effective lane area setting unit 14 obtains the shoulder width of the shoulder area for each vertical position of the reference coordinate system from the road shoulder information when the road area touches the left lane line, and the left effective lane according to the shoulder width for each vertical position. Determine the width setting. Specifically, when the road shoulder region is in contact with the left lane, the left effective lane width setting value is increased at a position where the shoulder width is wide, and the left effective lane width setting value is decreased at a position where the road shoulder width is narrow. The effective lane region setting unit 14 obtains the shoulder width of the shoulder region for each vertical position of the reference coordinate system from the road shoulder information when the road shoulder region is in contact with the right lane marking, and the right lane region according to the shoulder width for each vertical position. Determine the effective lane width setting. Specifically, when the road shoulder region is in contact with the right lane, the right effective lane width setting value is increased at a position where the shoulder width is wide, and the right effective lane width setting value is decreased at a position where the road shoulder width is narrow. The maximum value of the left effective lane width and the right effective lane width is the distance from the left lane line to the right lane line, that is, half the lane width.

  In the example shown in FIG. 4, the road area is in contact with the left lane line, but the road area is not in contact with the right lane line (that is, the lane divided by the left lane line and the right lane line is on the left side) (In this case, it is not an adjacent lane adjacent to the right shoulder area (not shown).) In such a case, the right effective lane width is fixed to half of the maximum lane width. On the other hand, only the left side effective lane width is changed to a width according to the shoulder width. That is, the allowance from the left lane line to the effective lane region is reduced as the road width is wide, and the allowance from the left lane line to the effective lane region is increased as the road width is narrow. This margin is provided on the assumption that when a roadside object is converted into a target, the target frame corresponding to the roadside object protrudes into the lane. By providing a margin between the lane line (shoulder-side lane boundary line) in contact with the shoulder area and the effective lane area, the target frame corresponding to the roadside object has entered the lane beyond the lane line. However, it is possible to prevent the target frame from being erroneously determined as that of an object located in the lane.

  The amount of intrusion that the target frame corresponding to the roadside object enters the lane beyond the lane line increases as the distance between the roadside object and the lane, that is, the shoulder width becomes narrower. Conversely, if the roadside width is wide and the roadside object is far away from the lane, the target frame corresponding to the roadside object is unlikely to enter the lane beyond the lane line. Therefore, the effective lane area setting unit 14 reduces the margin by increasing the setting value of the effective lane width as the shoulder width increases, and reduces the margin by decreasing the setting value of the effective lane width as the shoulder width decreases. Has increased. Note that the minimum value of the margin is zero, and the maximum value is the width of one vehicle to one vehicle. The reason that the maximum value of the margin is limited in this way is to prevent the vehicle from being erroneously determined as an object located outside the lane when the vehicle is traveling near the shoulder.

  There is no limitation in the specific calculation method of the setting value of the effective lane width by the effective lane area setting unit 14. For example, even if the setting value of the effective lane width is calculated by a linear expression from the shoulder width obtained from the shoulder information so that the setting value of the effective lane width also decreases linearly as the shoulder width of the adjacent shoulder area decreases. Good. In addition, even if the shoulder width of the adjacent shoulder region is the same, the target frame of the roadside object tends to protrude into the lane as the road curvature increases (that is, the road radius decreases). For example, in the example shown in FIG. 5, the target frame 1 of the roadside object 1 in the straight portion does not protrude into the lane, but the target frame 2 of the roadside object 2 in the curved portion protrudes into the lane. Therefore, the set value of the effective lane width may be decreased as the road shoulder width is smaller and the curvature is larger.

  The target information acquired by the target information acquisition unit 11 and the information related to the effective lane region set by the effective lane region setting unit 14 are input to the lane inside / outside determination unit 15. The lane inside / outside determination unit 15 is based on the degree of overlap between the target frame included in the target information and the effective lane area, and the object specified by the target frame is an object located in the lane or located outside the lane. Determine if it is an object. The specific determination method can be described with reference to FIGS.

In FIG. 6, the left effective lane width and the right effective lane width indicating the effective lane area, and the target frame of the preceding other vehicle are drawn in the reference coordinate system. The four black circles drawn in FIG. 6 are points indicating the four corners of the target frame. Moreover, the four white circles drawn in FIG. 6 have shown a part of point sequence which comprises a lane centerline. The lane inside / outside determination unit 15 determines whether the object corresponding to the target frame is located in the lane or outside the lane according to the following procedure.
(1) For each of the four corner points of the target frame, a point pair on the lane center line sandwiching them is specified based on the X coordinate of the corner point.
(2) The left and right effective lane widths corresponding to the vertical positions of the point pairs on the specified lane center line are read.
(3) The vertical distance between each of the four corner points of the target frame and the line segment connecting the pair of points on the corresponding lane center line is calculated. Specifically, the vertical distance is the length of the perpendicular illustrated in FIG.
(4) For each of the four corner points of the target frame, it is determined whether or not the condition that the vertical distance is smaller than the effective lane width (left effective lane width in the example shown in FIG. 7) is satisfied. If this condition is satisfied, it is determined that the corner point is located in the lane.
(5) When it is determined that at least one of the four corner points of the target frame is located in the lane, it is determined that the object corresponding to the target frame is an object located in the lane.

  When there are a plurality of target frames detected in the vicinity of the host vehicle, the lane inside / outside determination unit 15 performs the determination process on each target frame according to the above procedure. In addition, you may change said procedure as follows. For example, as a first modified example, when it is determined that, among the four corner points of the target frame, a number of corner points greater than or equal to a predetermined value, or a corner point greater than or equal to a predetermined ratio is located in the lane The object corresponding to the target frame may be determined to be an object located in the lane. As a second modification, in addition to the four corner points, a plurality of points on the target frame are taken, and it is determined whether or not each point is located in the lane, and a number of points greater than or equal to a predetermined value, or When it is determined that a point greater than or equal to a predetermined ratio is located in the lane, the object corresponding to the target frame may be determined to be an object located in the lane.

  The control device 10 functions as an object recognition device due to the functions of the above-described units 11, 12, 13, 14, and 15. FIG. 8 is a flowchart showing the flow of object recognition processing according to the present embodiment. The control device 10 as the object recognition device repeatedly performs the processing shown in this flowchart at a predetermined cycle.

  First, in step S1, the control device 10 estimates the position and orientation of the host vehicle from the GPS signal, and the lane boundary information in which the position and shape of the lane boundary around the host vehicle are expressed in the reference coordinate system. Is obtained from the map database 6.

  In step S <b> 2, the control device 10 acquires information about a road shoulder region around the host vehicle from the map database 6, and acquires a road shoulder width at each vertical position in the reference coordinate system by calculation.

  In step S3, the control device 10 calculates the position and shape of the lane center line of each lane in the reference coordinate system based on the lane boundary information acquired in step S1. Further, the control device 10 determines the effective lane width on the left side with respect to the lane center line based on the road shoulder width in each vertical position of the reference coordinate system acquired in step S2 and the positional relationship between the road shoulder region and each lane. The effective lane width on the right side is calculated for each vertical position of the reference coordinate system.

  In step S4, the control device 10 acquires distance measurement data of each object around the host vehicle from the autonomous recognition sensor, and obtains a target corresponding to each object by clustering the distance measurement data.

  In step S5, the control device 10 determines the target frame based on the overlap between the target frame of the target acquired in step S4 and the effective lane area defined by the left and right effective lane widths calculated in step S3. It is determined whether the object specified in is an object located in the lane or an object located outside the lane. The specific determination method is as described above.

  According to the object recognition processing performed in the above procedure, the object specified by the target frame is in the lane based on the overlap between the effective lane area whose range is changed according to the shoulder width and the target frame. Therefore, even if the roadside object outside the road on the road with a narrow shoulder width is outside the road, the target frame may be misdetected as a target in the lane. Is suppressed. This effect will be described with reference to the drawings.

  FIG. 9 is a diagram illustrating a recognition result obtained by a conventional object recognition process. Here, the case where the roadside object 1 is located at a position away from the lane and the roadside object 2 is located near the lane will be described as an example. As shown in FIG. 9, when the roadside objects 1 and 2 are converted into targets based on the distance measurement data obtained by the autonomous recognition sensor, May tilt. At this time, the target frame 1 indicating the roadside object 1 does not reach the lane because the shoulder width at the position is wide. However, the target frame 2 indicating the roadside object 2 protrudes into the lane due to inclination because the shoulder width at that position is narrow. According to the conventional object recognition processing, when a lane and a target frame overlap, an object specified by the target frame is detected as an object in the lane. For this reason, if the target frame 2 protrudes into the lane, it will be recognized as if an object exists there. As a result, unexpected sudden braking, lane change, and the like may be performed by the automatic driving system so as to avoid collision with the object.

  FIG. 10 is a diagram illustrating a recognition result by the object recognition process performed in the above-described procedure. Since the shoulder width at the position of the roadside object 2 is narrow, a large allowance from the section line (roadside lane boundary line) to the effective lane region is taken at that position. Therefore, even if the target frame 2 indicating the roadside object 2 is inclined and the target frame 2 protrudes in the lane, there is a margin, so that it does not enter the effective lane region. Therefore, according to the object recognition process performed in the above-described procedure, it is possible to prevent the object specified by the target frame 2, that is, the roadside object 2 from being erroneously detected as an object in the lane. If it is reduced that roadside objects are erroneously detected as objects in the lane, unexpected control by the automatic driving system is also reduced.

  Although the embodiment of the present invention has been described above, as described above, the present invention is not limited to this embodiment. For example, in FIG. 2, the lane boundary line is represented by a point sequence, but the lane boundary line registered in the map data may be a set of line segments or a set of curves.

  Means for estimating the position and orientation of the host vehicle are not limited to GPS signals. For example, the position and orientation of the host vehicle may be estimated by associating feature points obtained from the output of the rider 2 or the stereo camera 4 with feature points such as signs and white lines registered on the map.

  Acquisition of lane boundary information is not limited to acquisition from the map database 6. For example, the front of the host vehicle may be photographed by the stereo camera 4 and the lane boundary information may be acquired by white line recognition processing for the obtained photographed image.

1 Autonomous Driving System 2 Rider 3 Radar 4 Stereo Camera 5 GPS Receiver 6 Map Database 7 Navigation System 8 HMI
9 Actuator 10 Control device (object recognition device)
11 Target information acquisition unit (target information acquisition means)
12 Lane boundary information acquisition unit (lane boundary information acquisition means)
13 Road shoulder information acquisition unit (road shoulder information acquisition means)
14 Effective lane area setting part (Effective lane area setting means)
15 Lane inside / outside judgment part (judgment means)

Claims (1)

Target information acquisition means for acquiring target information related to an object existing around the host vehicle;
Lane boundary information acquisition means for acquiring lane boundary information regarding the position and shape of the lane boundary of the road on which the host vehicle travels;
Shoulder information acquisition means for acquiring shoulder information related to the width of the shoulder region of the road on which the host vehicle travels;
Based on the lane boundary information and the road shoulder information, effective lane area setting means for setting an effective lane area inside a roadside lane boundary line that divides the adjacent lane adjacent to the road shoulder area and the road shoulder area. When,
Based on the degree of overlap between the target frame included in the target information and the effective lane area, the object specified by the target frame is an object located in the adjacent lane or located outside the adjacent lane. A determination means for determining whether the object is an object,
The effective lane area setting means reduces the margin from the roadside lane boundary line to the effective lane area as the width of the roadside area increases, and decreases from the roadside lane boundary line as the width of the roadside area decreases. The object recognition apparatus, wherein the effective lane region is set so as to increase a margin to the effective lane region.

Images