JP5471195B2 - Object detection device - Google Patents

Description

  The present invention relates to an object detection apparatus that detects an object based on information acquired by a radar and a camera.

In recent years, safety systems such as PCS (Pre-crash safety system) for collision avoidance or reduction of damage caused by collision have been developed. In order to suitably realize such a safety system, it is necessary to accurately grasp the positions and sizes of vehicles other than the own vehicle, obstacles, pedestrians, and the distance between them and the own vehicle. As a technique for grasping these, there is known an object detection device that detects an object by fusing radar target information acquired by a radar and image target information acquired from an image of a camera.

  According to the radar target information acquired by the radar, the distance between the host vehicle and the target can be calculated. However, it is difficult to calculate the horizontal position and width of the target with high accuracy from the radar target information. On the other hand, according to the image target information acquired from the camera image, the horizontal position and width of the target can be calculated with higher accuracy than the radar target information. Therefore, by fusing the radar target information and the image target information, the position and size of the object and the distance from the host vehicle can be grasped more accurately.

Patent Document 1 discloses a vehicle object recognition device that makes it easier to intervene in a safety system for collision avoidance control as the lateral width of an external target in front of the host vehicle recognized by sensor fusion between a ranging radar and an image sensor becomes longer. Has been. In the object recognition apparatus according to Patent Document 1, a radar width which is a lateral width of an external target obtained using a radar and an image width which is a lateral width of an external target obtained using an image sensor are short. Is selected as the sensor fusion recognition width.

  Patent Documents 2 and 3 also disclose a technique for recognizing or detecting an object by fusing radar target information acquired by a radar and image target information acquired from an image of a camera.

JP 2006-240454 A JP 2005-329779 A JP 2006-266927 A

  In the object detection apparatus, a stereo camera or a monocular camera can be used as a camera for acquiring image target information. When a monocular camera is used, costs can be reduced compared to when a stereo camera is used. However, it is difficult to obtain accurate target information in the depth direction from an image of a monocular camera. Therefore, in the image target information acquired from the image of the monocular camera, an object or pattern existing behind the target when viewed from the own vehicle may be recognized as the same object as the target. . In this case, the width of the target is detected to be larger than the actual width.

  If it is determined by such erroneous detection that the detected object protrudes from the own lane, there is a risk of causing a malfunction of a safety system such as PCS.

  The present invention has been made in view of the above problems, and detects an object by fusing radar target information acquired by a radar and image target information acquired by a camera. An object of the present invention is to provide a technique capable of further improving the detection accuracy of an object in an apparatus.

  The present invention uses only image target information within a predetermined horizontal range set in accordance with the intensity of the received wave of the radar from the position of the target recognized by the radar, and uses the radar target information and the image. The target information is fused.

More specifically, the object detection device according to the present invention is:
An object detection device that detects an object by fusing radar target information acquired by a radar and image target information acquired from an image of a monocular camera,
The lateral distance between the position of the target recognized by the radar and the horizontal end of the target recognized by the monocular camera has a predetermined horizontal width set according to the intensity of the received wave of the radar. If it exceeds, radar target information and image target information are fused without using image target information in the meantime.

  The intensity of the received wave of the radar changes according to the target. Here, the predetermined horizontal width is set as the maximum value of the horizontal width of the target corresponding to the intensity of the received wave of the radar or a value obtained by adding a certain margin to the maximum value.

  According to the present invention, the image target information excluding the portion with low reliability is fused with the radar target information. This suppresses erroneous detection of the horizontal width of the target. Therefore, the object detection accuracy can be further improved.

  The radar may recognize the target as an electromagnetic wave reflection point. The monocular camera may recognize a target as an image.

  In the present invention, the predetermined lateral width may be set for each threshold level to which the intensity of the received wave of the radar belongs. That is, when the intensity of the received wave of the radar is greater than or equal to the first threshold, the predetermined horizontal width is set to the first predetermined horizontal width, the intensity of the received wave of the radar is lower than the first threshold, and the second threshold In the above case, the predetermined lateral width may be set to a second predetermined lateral width that is smaller than the first predetermined lateral width. According to this, the predetermined lateral width can be set as a value corresponding to the target recognized by the radar.

  In the above case, the first threshold value may be a value determined on the assumption that the target is a large object such as a vehicle. Further, the second threshold value may be a value determined on the assumption that the target is a small object of the order of a human body.

  According to the present invention, the object detection accuracy can be further improved. As a result, malfunction of a safety system such as PCS can be suppressed.

It is a block diagram which shows the one part schematic structure of PCS which concerns on an Example. In an Example, it is a figure for demonstrating how to be recognized as a target of a wall when a vehicle drive | works the narrow road with a wall along a road. It is a figure for demonstrating the method of fusion of the radar target information and image target information which concern on an Example. It is a flowchart which shows the flow of fusion of the radar target information and image target information which concern on an Example.

  Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. The dimensions, materials, shapes, relative arrangements, and the like of the components described in the present embodiment are not intended to limit the technical scope of the invention to those unless otherwise specified.

(Schematic configuration of the system)
Here, an embodiment in the case where the object detection apparatus according to the present invention is applied to a PCS will be described. FIG. 1 is a block diagram illustrating a schematic configuration of a part of the PCS according to the present embodiment. In this embodiment, the PCS 1 is mounted on the vehicle 100. The PCS 1 includes a millimeter wave radar 2, a monocular camera 3, a camera ECU 4, and a PCS ECU 5. The PCS ECU 5 includes a fusion calculation unit 6, a physical value calculation unit 7, and a collision determination unit 8.

  The PCS 1 detects a vehicle other than the own vehicle, an obstacle, a pedestrian, and the like based on information acquired by the millimeter wave radar 2 and the monocular camera 3. When it is determined that there is a high possibility of collision with these objects (including pedestrians), a warning is sent to the driver and collision damage reduction control is performed. Here, examples of the collision damage reduction control include reduction of the collision speed by winding the seat belt or pre-crash brake.

  The millimeter wave radar 2 is attached to the front center portion of the vehicle 100. The millimeter wave radar 2 scans the front and oblique front of the vehicle 100 in the horizontal direction with millimeter wave electromagnetic waves, and receives the electromagnetic waves reflected on the surface of an object outside the vehicle. Thus, the millimeter wave radar 2 recognizes the target as an electromagnetic wave reflection point. Further, the millimeter wave radar 2 acquires target information (radar target information) from millimeter wave transmission / reception data. The radar target information here is the lateral position of the target, the distance between the host vehicle 100 and the target, and the relative speed between the host vehicle 100 and the target. The radar target information is input to the fusion calculation unit 6 of the camera ECU 4 and the PCS ECU 5.

  The monocular camera 3 is a CCD camera and is attached to the front center portion of the vehicle 100. The monocular camera 3 captures images in front of and obliquely forward of the vehicle 100. Thereby, the monocular camera 3 recognizes the target as an image. An image captured by the monocular camera 3 is input to the camera ECU 4 as an image signal. The camera ECU 4 acquires target information (image target information) from the input image signal. The image target information here is the horizontal position of the target, the horizontal width and the height of the target. The image target information is input to the fusion calculation unit 6 of the PCS ECU 5.

  At this time, the image target information is input to the fusion calculation unit 6 in association with the radar target information for the same target. Whether the target recognized by the millimeter wave radar 2 and the target recognized by the monocular camera 3 are the same target or not is determined by the lateral position of the target in the radar target information and the image target information. Judgment based on. For example, when the target recognized by the millimeter wave radar 2 and the target recognized by the monocular camera 3 overlap, it is determined that both targets are the same.

  Note that the mounting positions of the millimeter wave radar 2 and the monocular camera 3 are not limited to the front center of the vehicle. For example, these may be attached to the rear side of the vehicle.

The fusion calculation unit 6 fuses the radar target information input from the millimeter wave radar 2 and the image target information input from the camera ECU 4. The fusion method will be described later. The calculation result by the fusion calculation unit 6 is input to the physical value calculation unit 7.

  Based on the calculation result of the fusion calculation unit 6, the physical value calculation unit 7 determines the horizontal position of the target, the horizontal width and height of the target, the distance between the host vehicle 100 and the target, and between the host vehicle 100 and the target. Calculate the relative speed. The calculation result by the physical value calculation unit 7 is input to the collision determination unit 8.

  The collision determination unit 8 determines whether or not the host vehicle 100 and the target are highly likely to collide based on the calculation result of the physical value calculation unit 7. When the collision determination unit 8 determines that there is a high possibility that the host vehicle 100 and the target will collide, the PCS ECU 5 transmits a control signal to a seat belt retracting actuator, a pre-crash brake actuator, and the like (not shown). To implement collision damage reduction control. Further, in this case, the PCS ECU 5 transmits a warning ON signal to a warning device (not shown) for the driver.

  In this embodiment, the millimeter wave radar 2, the monocular camera 3, the camera ECU 4, the fusion calculation unit 6, and the physical value calculation unit 7 constitute an object detection apparatus according to the present invention.

(Fusion method)
Hereinafter, a fusion method between the radar target information input from the millimeter wave radar 2 and the image target information input from the camera ECU 4 in the fusion calculation unit 6 will be described with reference to FIGS. The image target information according to the present embodiment is acquired from the image of the monocular camera 3. However, it is difficult to obtain accurate target information in the depth direction from the image of the monocular camera 3. Therefore, in the image target information acquired from the image of the monocular camera 3, an object or a pattern that is actually present behind the target as viewed from the own vehicle 100 is recognized as the same object as the target. There is. In this case, the width of the target is detected to be larger than the actual width.

  Here, a target recognized as a reflection point of electromagnetic waves by the millimeter wave radar 2 is referred to as a radar target, and a target recognized as an image by the monocular camera 3 is referred to as an image target. For example, as shown in FIG. 2, when the vehicle 100 travels on a narrow road 200 with a wall 201 along the road, the millimeter wave radar 2 recognizes the wall at the point A as a radar target. At this time, the monocular camera 3 also recognizes the wall at the point A as an image target. However, according to the monocular camera 3, the wall behind the point A may be recognized as being at the point A. As a result, in the image target information, the width of the wall at the point A is a value larger than the actual width.

  As described above, the image target information erroneously detected as the width of the target being larger than the actual width is fused with the radar target information, and when the object is detected based on the result, the detected object Is determined to protrude from the own lane (in the case of FIG. 2, it is determined that the wall at the point A protrudes from the own lane). In this case, the PCS 1 may malfunction.

  Therefore, in this embodiment, a predetermined lateral width is set according to the intensity of the received wave of the millimeter wave radar 2. The intensity of the received wave of the millimeter wave radar 2 changes according to the target. The predetermined horizontal width is set as a value obtained by adding a certain margin to the maximum value of the horizontal width of the target corresponding to the intensity of the received wave of the millimeter wave radar 2. If the lateral distance between the position of the radar target recognized by the millimeter wave radar 2 and the horizontal end of the image target recognized by the monocular camera 3 exceeds a predetermined lateral width, the image between them The radar target information and the image target information are fused without using the target information.

  FIG. 3A shows a radar target and an image target actually recognized by the millimeter wave radar 2 and the monocular camera 3. FIG. 3B shows the radar target and the image target that are fused in the fusion calculation unit 6.

  In FIG. 3A, the lateral distance Li between the position of the radar target recognized by the millimeter wave radar 2 and the right lateral end of the image target recognized by the monocular camera 3 is a predetermined lateral width α. Is over. In this case, there is a high possibility of erroneous detection on the right side of the position of the radar target in the image target. For this reason, as shown in FIG. 3B, the radar target information and the image target information are fused without using the image target information on the right side of the position of the radar target.

  According to the fusion method as described above, the image target information excluding the portion with low reliability is fused with the radar target information. This suppresses erroneous detection of the horizontal width of the target. As a result, malfunction of PCS1 can be suppressed.

(Fusion flow)
A fusion flow of the radar target information and the image target information in the fusion calculation unit according to the present embodiment will be described based on the flowchart shown in FIG. This flow is a flow that is repeatedly executed in the fusion calculation unit 6.

  In this flow, first, in step S101, radar target information and image target information are read.

  Next, in step S102, it is determined whether there is image target information that can be fused with radar target information. That is, it is determined whether there is image target information for the same target as the target recognized by the millimeter wave radar 2. If an affirmative determination is made in step S102, the process of step S103 is executed next. If a negative determination is made, the execution of this flow is temporarily terminated.

  In step S103, it is determined whether or not the intensity Ir of the received wave of the millimeter wave radar 2 is equal to or higher than a predetermined high threshold value Ir1. Here, the high threshold value Ir1 is a value determined on the assumption that the target is a large object such as a vehicle. For example, the high threshold value Ir1 may be set as the minimum value of the intensity of the reflected wave reflected by the vehicle. If an affirmative determination is made in step S103, the process of step S104 is executed next. If a negative determination is made, the process of step S107 is executed next.

  In step S104, the lateral distance Li between the position of the radar target recognized by the millimeter wave radar 2 and the horizontal end of the image target recognized by the monocular camera 3 has a first predetermined lateral width α1. It is determined whether or not it exceeds. This determination is performed on both left and right ends of the image target.

  Here, the first predetermined lateral width α1 is a value set by adding a certain margin to the maximum value of the lateral width of a target corresponding to the high threshold value Ir1, that is, a large object of the size of a vehicle. When the high threshold value Ir1 is set as the minimum value of the intensity of the reflected wave reflected by the vehicle, the first predetermined lateral width α1 is set as a value obtained by adding a margin to the lateral width of the vehicle (for example, 2 m). If an affirmative determination is made in step S104, the process of step S105 is executed next, and if a negative determination is made, the process of step S106 is executed next.

  In step S105, the image target information between the lateral end of the one determined that the distance Li from the radar target position exceeds the first predetermined lateral width α1 and the radar target position is used. First, the radar target information and the image target information are fused.

  In step S106, the radar target information and the image target information are fused using the image target information between the left and right ends (that is, all the image target information for the image target).

  On the other hand, in step S107, it is determined whether or not the intensity Ir of the received wave of the millimeter wave radar 2 is greater than or equal to a predetermined low threshold value Ir2. Here, the low threshold value Ir2 is a value determined on the assumption that the target is a small object of the order of a human body. For example, the low threshold value Ir2 may be set as the minimum value of the intensity of the reflected wave reflected by the human body (pedestrian). If an affirmative determination is made in step S107, then the process of step S107 is executed next. If a negative determination is made, the execution of this flow is temporarily terminated.

  In step S108, whether or not the lateral distance Li between the position of the radar target and the horizontal end of the image target exceeds a second predetermined lateral width α2 that is smaller than the first predetermined lateral width α1. Is determined. This determination is performed on both left and right ends of the image target.

  Here, the second predetermined lateral width α2 is a value set by adding a certain margin to the maximum value of the lateral width of a target corresponding to the low threshold value Ir2, that is, a small object of the human body. When the low threshold value Ir2 is set as the minimum value of the intensity of the reflected wave reflected by the human body, the second predetermined lateral width α2 is set as a value obtained by adding a margin to the lateral width of the human body (for example, 1 m). If an affirmative determination is made in step S108, the process of step S109 is executed next. If a negative determination is made, the process of step S110 is executed next.

  In step S109, the image target information from the lateral end to the position of the radar target on which the distance Li from the position of the radar target is determined to exceed the second predetermined lateral width α2 is used. First, the radar target information and the image target information are fused.

  In step S110, the radar target information and the image target information are fused using the image target information between the left and right ends (that is, all image target information for the image target).

  As described above, in this embodiment, the case where the target is a large object such as a vehicle is distinguished from the case where the target is a small object such as a human body, which is a criterion for determining the reliability of the image target information. A predetermined lateral width is set. Therefore, it is possible to determine the reliability of the image target information corresponding to the target.

  In the present embodiment, the first and second predetermined lateral widths α1 and α2 are set for each threshold level (high threshold Ir1 and low threshold Ir2) to which the intensity of the received wave of the millimeter wave radar 2 belongs. Three or more predetermined lateral widths may be set according to the intensity of the received wave.

1 ... PCS (Pre-crash safety system)
2 ... Millimeter wave radar 3 ... Monocular camera 4 ... Camera ECU
5 ... PCS ECU
6 ... Fusion calculation unit 7 ... Physical ground calculation unit 8 ... Collision determination unit 100 ... Vehicle 200 ... Road 201 ... Wall

Claims (4)

The object detection device includes a fusion calculation unit that fuses radar target information acquired by a radar and image target information acquired from an image of a monocular camera, and detects an object based on a calculation result of the fusion calculation unit. And
The fusion calculation unit sets a lateral distance between the position of the target recognized by the radar and the horizontal end of the target recognized by the monocular camera according to the intensity of the received wave of the radar. An object detection apparatus that fuses radar target information and image target information without using image target information in the case where the predetermined horizontal width is exceeded. The radar recognizes target as the reflection point of the electromagnetic wave, the monocular camera object detecting apparatus mounting serial to claim 1, wherein recognizing the target as an image. If the intensity of the received wave of the radar is greater than or equal to a first threshold, the predetermined lateral width is set to a first predetermined lateral width,
When the intensity of the received wave of the radar is lower than the first threshold and greater than or equal to the second threshold, the predetermined lateral width is set to a second predetermined lateral width that is smaller than the first predetermined lateral width. object detection device of the mounting serial to claim 1 or 2, characterized in.   The first threshold is a value determined on the assumption that the target is a large object such as a vehicle, and the second threshold is that the target is a small object such as a human body. The object detection apparatus according to claim 3, wherein the object detection apparatus has a value determined on the assumption.

Images