WO2023188685A1

WO2023188685A1 - Computation processing device

Info

Publication number: WO2023188685A1
Application number: PCT/JP2023/000607
Authority: WO
Inventors: 真荒木
Original assignee: 本田技研工業株式会社
Priority date: 2022-03-28
Filing date: 2023-01-12
Publication date: 2023-10-05

Abstract

The present invention can improve the speed at which a surrounding object is detected and reduce the cost required for detecting the object.　A computation processing device according to the present invention recognizes the surrounding object on the basis of detection information from a first sensor (12) and detection information from a second sensor (11), the first sensor (12) detecting vehicle information of a vehicle and the second sensor (11) detecting conditions surrounding the vehicle. The computation processing device comprises: a detection range determination unit 41 that determines the range of detection of the surrounding conditions on the basis of the detection result from the first sensor (12); an image generation unit 42 that generates, on the basis of the detection information from the second sensor (11), a recognition image comprising an image of the detection range; and an image recognition unit 43 that carries out image recognition of an object to be detected in the recognition image. The amount of data of the recognition image is less than the amount of data of an image of the surrounding conditions obtained from the detection information of the second sensor (11).

Description

arithmetic processing unit

The present invention relates to an arithmetic processing device.

A computing processing device is known that is equipped with a sensor that detects the surrounding situation of a vehicle and recognizes surrounding objects based on the detected information. This type of device is equipped with a vehicle speed sensor that detects the vehicle speed of the target vehicle, and a camera and distance measurement sensor that detect the surrounding conditions of the target vehicle. There is a device that displays an object detected by a distance sensor by switching the display (for example, Patent Document 1).

International Publication No. 2017/068692

With the increase in the number of pixels in image sensors, the amount of image data is increasing.
In conventional configurations, image recognition processing is performed using a large amount of image data, which may result in a situation where it takes time to detect an object. If hardware resources such as high-performance arithmetic processing devices and large-capacity memories are adopted in order to shorten the time, this will be disadvantageous in terms of cost reduction.
The present invention has been made in view of the above-mentioned circumstances, and aims to improve the detection speed of surrounding objects and reduce the cost required for detecting objects.

This specification includes all contents of Japanese patent application/Japanese Patent Application No. 2022-052644 filed on March 28, 2022.
In order to achieve the above object, a first sensor that detects vehicle information of a vehicle, a second sensor that detects the surrounding situation of the vehicle, and a calculation process that recognizes surrounding objects based on the detection information of the second sensor. The apparatus includes a detection range determination unit that determines a detection range of the surrounding situation based on a detection result of the first sensor, and a recognition image consisting of an image of the detection range based on detection information of the second sensor. an image recognition unit that recognizes an object to be detected based on the recognition image, and the recognition image is an image of the surrounding situation obtained from the detection information of the second sensor. It is characterized by a smaller amount of data.

It becomes possible to improve the detection speed of surrounding objects and reduce the cost required to detect objects.

FIG. 1 is a diagram showing a driving support device according to an embodiment of the arithmetic processing device of the present invention. FIG. 2 is a diagram showing an example of an image taken by a camera. FIG. 3 is a diagram for explaining the basic operation of the driving support device. FIG. 4 is a flowchart showing the detection range determination process. FIG. 5 is a diagram schematically showing a road, a photographed image, and a detection range when the vehicle is turning to the right. FIG. 6 is a diagram schematically showing a road, a detection range, and a center position when the vehicle is traveling straight. FIG. 7 is a diagram showing an example of the angle of view of a photographed image and the angle of view of a detection range.

Hereinafter, one embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing a driving support device according to an embodiment of the arithmetic processing device of the present invention. This driving support device 10 is a device mounted on the vehicle 1, and has a function of recognizing objects existing around the vehicle 1 and notifying the recognition results. In this embodiment, the vehicle 1 is a motorcycle. The vehicle 1 may be a vehicle other than a motorcycle, for example, a saddle type vehicle including all saddle type vehicles such as a motorcycle, or any vehicle other than a saddle type vehicle.

The driving support device 10 includes a surrounding detection sensor 11, a vehicle sensor 12, a calculation section 13, a storage section 14, and an output section 15. The surroundings detection sensor 11 is a sensor that detects the surroundings of the vehicle 1, and includes a camera 21 that photographs at least the front of the vehicle 1.
FIG. 2 is a diagram showing an example of an image GC taken by the camera 21.
From the perspective of photographing objects existing around the vehicle 1, the camera 21 captures images of the road (road surface) G1 on which the vehicle 1 is traveling, as well as other vehicles G2, facilities G3, and sidewalks located in front of the vehicle 1, to the left and right of the front. Specifications such as the angle of view (horizontal angle of view, vertical angle of view) and resolution, as well as the shooting range, etc., have been established to enable shooting of G4 and the like.

The vehicle sensor 12 is a sensor that detects vehicle information DM. More specifically, the vehicle sensor 12 includes a vehicle speed sensor 31 that detects the speed of the vehicle 1 (hereinafter referred to as vehicle speed), and a vehicle speed sensor 31 that detects the speed of the vehicle 1 (hereinafter referred to as the vehicle speed), and the steering angle of the vehicle 1 (also referred to as the steering angle or the steering angle of the front wheels). It includes a steering angle sensor 32 that detects the steering angle, and a tilt angle sensor 33 that detects the tilt angle of the vehicle 1, and outputs vehicle information DM consisting of the vehicle speed, steering angle, and tilt angle of the vehicle 1. The tilt angle detected by the tilt angle sensor 33 is a tilt angle of the vehicle body to the left and right, and is, for example, a roll angle.

The arithmetic unit 13 includes an arithmetic processing device capable of performing various arithmetic processes. The calculation unit 13 functions as a detection range determination unit 41, an image generation unit 42, and an image recognition unit 43 by executing the control program DP stored in the storage unit 14 using a calculation processing device. The detection range determination unit 41 performs a surrounding situation determination process to determine a detection range R of the surrounding situation. The image generation unit 42 performs image generation processing to generate a recognition image GT consisting of an image of the detection range R from the captured image GC of the camera 21.

The image recognition unit 43 performs image recognition processing on the recognition image GT to recognize the object to be detected. Objects to be detected are objects that may affect the running of the vehicle 1 or objects for which it is desirable to notify the occupants, such as other vehicles, obstacles, and pedestrians.

The storage unit 14 is a storage device that stores various data used by the driving support device 10. The storage unit 14 stores detection range determination data DR in addition to the control program DP. The detection range determination data DR is information for determining the detection range R based on the vehicle speed, steering angle, and left and right tilt angle (hereinafter referred to as "left and right tilt angle") of the vehicle 1.
For example, map data for determining the detection range R from three parameters including the vehicle speed, steering angle, and left/right tilt angle of the vehicle 1 is applied to the detection range determination data DR.

The detection range determining data DR satisfies at least the following three conditions.
<First condition>
The detection range determination data DR is such that the larger the vehicle speed is than a predetermined value, the smaller the angle of view (horizontal angle of view, vertical angle of view) corresponding to the size of the detection range R. The angle of view is also formed to increase the angle of view as the vehicle speed decreases.
When a passenger is notified of the presence of an object to be detected, there is a limit to how quickly the passenger can react after being notified. By reducing the angle of view of the detection range R as the vehicle speed increases, objects located far away are made relatively large within the detection range R, making it easier to recognize the images of distant objects at an early stage.
At low speeds, the amount of information change in the captured image GC is generally relatively small. By increasing the angle of view of the detection range R as the vehicle speed decreases, it becomes easier to recognize objects over a wider range.
For example, the angle of view when the vehicle speed is a predetermined value is set to the threshold size SS, when the vehicle speed is greater than the predetermined value, the angle of view is reduced to the threshold size SS or less, and when the vehicle speed is smaller than the predetermined value, the angle of view is set to the threshold size SS. In this case, the angle of view is made larger than the threshold size SS. The predetermined value and the threshold size SS may be set to appropriate values as long as the speed and angle of view allow appropriate image recognition of the object.

<Second condition>
The detection range determining data DR is formed into data for correcting the horizontal inclination of the detection range R according to the horizontal inclination angle.
Since the camera 21 tilts as the vehicle 1 tilts from side to side, the tilt of the camera 21 can be corrected by making the detection range R tilted in the opposite direction by the tilted amount. Thereby, the detection range R can be set between an image when the vehicle 1 is tilted left and right and images before and after the tilt so that the image does not tilt. Thereby, when the object to be detected is image recognized from the image of the detection range R, it becomes easier to improve the object detection accuracy and shorten the detection time.

<Third condition>
Based on the vehicle speed, steering angle, and left/right inclination angle of the vehicle 1, the course of the vehicle 1 can be estimated by a known method, and it is also possible to estimate whether the vehicle 1 is turning.
The detection range determination data DR is formed to set the center position PA of the detection range R so as to include the course corresponding to the turning destination.
In this case, the center position PA is set on the road G1 on which the vehicle 1 travels. When the vehicle 1 is turning, that is, when the road G1 is a curve, the road G1 is not located in front of the camera 21, and the object to be detected in front may not be detected. Therefore, when the vehicle 1 is turning, the detection range determination data DR sets the center position PA of the detection range R to a course corresponding to the turning point rather than in front of the camera 21, as shown in FIG. 5 described later. Correct the position to include. Furthermore, the detection range determination data DR rotates the detection range R in the opposite direction by the left and right tilt angle. With these, a part of the detection range R can be overlapped on the course, and when an object to be detected is image recognized from the detection range R, it becomes easier to detect the object at the turning destination.

The output unit 15 is a device that inputs the recognition result of the image recognition unit 43 and outputs information based on the recognition result. The output unit 15 of this embodiment includes a first notification unit 51 that notifies the occupant of the vehicle 1 of the recognition result, and a second notification unit 52 that notifies other vehicles and the like of the recognition result.
The first notification unit 51 performs a process of causing a display device or an audio output device provided in the vehicle 1 to notify the presence of the object to be detected, for example, by display or audio as a recognition result.

The second notification unit 52 includes a communication device that realizes wireless communication with other vehicles existing around the vehicle 1. The second notification section 52 transmits information indicating the recognition result of the image recognition section 43 to other vehicles or the like via a communication device. The other destination vehicle is, for example, a vehicle running behind the vehicle 1, but is not limited to a vehicle behind the vehicle 1.
Note that either the second notification unit 52 or the first notification unit 51 may be omitted from the output unit 15. Further, the configuration is not limited to the above configuration, and the configuration of the output unit 15 and the output destination may be changed as appropriate.

FIG. 3 is a flowchart showing the basic operation of the driving support device 10. The flow shown in FIG. 3 is a process that is repeatedly executed when the vehicle 1 is running or the like.
First, the driving support device 10 starts acquiring vehicle information DM detected by the vehicle sensor 12 and acquiring an image GC captured by the camera 21 (step S1). Next, in the driving support device 10, the detection range determination unit 41 determines the detection range R based on the vehicle information DM and using the detection range determination data DR (step S2: detection range determination processing).

In order to make the explanation easier to understand, in FIG. 3, the detection range determination unit 41 sets a rectangular detection range R smaller than the shooting range of the shot image GC at the center position of the shot image GC, which corresponds to the front center of the vehicle 1. This example shows a case where a decision has been made. This detection range R corresponds to an area where another vehicle G2 (see FIG. 2) traveling in front of the vehicle 1 exists.

Next, the driving support device 10 uses the image generation unit 42 to generate a recognition image GT consisting of an image of the detection range R from the photographed image GC (step S3: image generation processing). In the driving support device 10, the image recognition unit 43 performs image recognition processing for recognizing the object to be detected on the recognition image GT (step S4).
A wide range of known image recognition techniques can be applied to this image recognition, and for example, the object to be detected is determined by image recognition of color and shape. In the case of the example shown in FIG. 3, the recognition image GT includes another vehicle G2 running in front of the vehicle 1, so the vehicle G2 is detected as the object to be detected by the process of step S4.

If the object to be detected is detected (step S5: YES), the driving support device 10 performs a process of notifying the recognition result through the output unit 15 (step S6). On the other hand, if the object to be detected is not detected (step S5: NO), the driving support device 10 temporarily ends the flow shown in FIG. 3. The above is the basic operation of the driving support device 10.

The detection range determination process in step S2 will be further explained.
FIG. 4 is a flowchart showing the detection range determination process. FIG. 5 is a diagram schematically showing the road G1, the photographed image GC, and the detection range R when the vehicle 1 is turning to the right. As shown in FIG. 5, when the vehicle 1 is turning to the right, the vehicle 1 is tilted to the right, so the photographed image GC is tilted by an angle θK (corresponding to the horizontal tilt angle) with respect to the horizontal plane. It has become a thing.

As shown in FIG. 4, the detection range determination unit 41 estimates the course of the vehicle 1 based on the vehicle speed, steering angle, and left/right inclination angle detected by the vehicle sensor 12 (step S11), and Based on this, the center position PA of the detection range R is calculated (step S12). In the case of the driving situation shown in FIG. 5, a position on the road G1 on which the vehicle 1 is traveling and which is closer to the right side in the turning direction within the photographing range of the photographed image GC is calculated as the center position PA. Thereby, the range including the turning destination location on the road G1 is set as the detection range R.

Here, FIG. 6 schematically shows the road G1, the detection range R, and the center position PA when the vehicle 1 is traveling straight. In FIG. 6, the detection range R and center position PA during high speed driving are shown as the detection range RH and center position PAH, and the detection range R and center position PA during low speed driving are shown as the detection range RL and center position PAL. ing. As shown in FIG. 6, the center position PAL during low-speed driving is a position close to the front of the vehicle 1 on the road G1 on which the vehicle 1 travels. Furthermore, the center position PAL during high-speed travel is a forward position far from the vehicle 1 on the road G1 on which the vehicle 1 travels.

Next, the detection range determination unit 41 calculates the angle θA of the detection range R based on the left and right inclination angles (step S13). The angle θA is an angle that corrects the inclination of the detection range R due to the left and right inclination of the vehicle 1. In the case of the driving situation shown in FIG. 5, the angle θA is an angle that corrects the angle θK that is the left and right inclination angle. The angle is −θK. Note that in the case of the driving situation shown in FIG. 6, the left-right inclination angle is zero, so the angle θA is zero.

Next, the detection range determining unit 41 calculates the angle of view of the detection range R based on the vehicle speed (step S14).
FIG. 7 is a diagram showing an example of the angle of view of the captured image GC and the angle of view of the detection range R.
As shown in FIG. 7, two types of angles of view (XR, YR) and (XR1, YR1) are illustrated as the angle of view (horizontal angle of view, vertical angle of view) of the detection range R. The angle of view of the detection range R is smaller than the angle of view (XS, YS) of the captured image GC. Also, among the angles of view (XR, YR) and (XR1, YR1) of the detection range R, the angle of view (XR, YR) is an example of the angle of view when the vehicle speed is smaller than a predetermined value, The angle of view (XR1, YR1) is an example of the angle of view when the vehicle speed is higher than a predetermined value.

The angle of view (for example, the angle of view (XR, YR)) when the vehicle speed is lower than the predetermined value is greater than or equal to the predetermined threshold size SS. Further, the angle of view (for example, the angle of view (XR1, YR1)) when the vehicle speed is higher than the predetermined value is less than or equal to the threshold size SS.
In this embodiment, the angle of view of the detection range R is set to a smaller angle of view as the vehicle speed is higher than the predetermined value, and is set to be smaller than the angle of view of the captured image GC. Therefore, the recognition image GT generated by the image generation unit 42 becomes image data with a smaller amount of data than the photographed image GC. Note that although FIG. 5 illustrates a case where the angle of view of the detection range R has a shape similar to the angle of view of the photographed image GC, it is not limited to a similar shape.

If the angle of view of the detection range R is increased as the vehicle speed becomes smaller than the predetermined value, the angle of view of the detection range R becomes close to the angle of view of the photographed image GC at extremely low speeds, and the recognition image GT and the photographed image GC There is a possibility that the amount of data will hardly change.
Therefore, the detection range determining unit 41 determines whether the angle of view of the detection range R calculated based on the vehicle speed is larger than a predetermined threshold size SS (step S15), and determines whether the angle of view of the detection range R is larger than the threshold size SS. If the size is larger than the size SS (step S15; YES), a thinning coefficient KG is calculated so that the angle of view when the vehicle speed is smaller than a predetermined value is fixed to the threshold size SS.

The thinning coefficient KG is a value that defines the amount of thinning of pixels in an image, and is set to a value such as 1/3, 1/4, . . . , 1/n (n is an arbitrary integer). The thinning coefficient is calculated such that the larger the angle of view of the detection range R, the larger the amount of thinning of pixels in the image.
For example, when the thinning coefficient KG is 1/2, the recognition image GT is created with approximately half the number of pixels by averaging adjacent pixel data of the recognition image GT to generate one pixel data. , and the amount of data of the recognition image GT can be reduced. In this way, by performing the process of thinning out the number of pixels in the detection range R, the resolution of the recognition image GT corresponding to the detection range R can be converted to a low resolution.

In the following step S16, a thinning coefficient KG is calculated. In this case, in step S17, the detection range determination unit 41 outputs detection range specification information DT that specifies the detection range R and the thinning coefficient KG to the image generation unit 42. As a result, the image generation unit 42 selects an image for recognition based on the captured image GC and the detection range specifying information DT, which is an image in the detection range R in the captured image GC and whose number of pixels has been reduced by the thinning coefficient KG. It is generated as an image GT.
Note that if the angle of view of the detection range R is smaller than the threshold size SS, the thinning coefficient KG is not calculated. In this case, in step S17, the detection range determining section 41 outputs detection range specifying information DT for specifying the detection range R to the image generating section 42. As a result, the image generation unit 42 generates a recognition image GT in which the number of pixels is not reduced. The above is the content of the detection range determination process.

As described above, the driving support device 10 uses the vehicle sensor 12 that detects the vehicle information DM of the vehicle 1, the surroundings detection sensor 11 that detects the surrounding situation of the vehicle 1, and the surroundings based on the detection results of the vehicle sensor 12. A detection range determination unit 41 that determines the detection range R of the situation; an image generation unit 42 that generates a recognition image GT consisting of an image of the detection range R based on detection information from the surrounding detection sensor 11; and a recognition image GT. On the other hand, it is provided with an image recognition unit 43 for image recognition of the object to be detected, and the recognition image GT is an image with a smaller amount of data than the photographed image GC obtained from the detection information of the surrounding detection sensor 11.
According to this configuration, by making the detection range R of the surrounding situation smaller than the detection range of the surrounding detection sensor 11, it is possible to obtain a recognition image GT with a reduced amount of data. Thereby, it is possible to reduce the amount of calculation when recognizing the object to be detected from the recognition image GT. Since the amount of calculation can be reduced, it is possible to improve the object detection speed, and it is also possible to create a configuration that does not require hardware resources such as high-performance arithmetic processing devices, which is advantageous for cost reduction.

Note that the vehicle sensor 12 corresponds to the "first sensor" of the present invention, and the surrounding detection sensor 11 corresponds to the "second sensor" of the present invention. Note that the vehicle sensor 12 and the surroundings detection sensor 11 may be changed as appropriate. For example, the surroundings detection sensor 11 may include a radar that is one of the devices that detect the surroundings.

Further, the vehicle sensor 12 includes a vehicle speed sensor 31, a steering angle sensor 32, and a tilt angle sensor 33 that detect the vehicle speed, steering angle, and left/right tilt angle of the vehicle 1, respectively. The detection range R is varied based on the detection result. According to this configuration, since the detection range R is varied based on the vehicle speed, steering angle, and left/right inclination angle of the vehicle 1, the detection range R can be made small to reduce the amount of calculations during image recognition, while the vehicle 1 is traveling It becomes easier to determine the detection range necessary for object recognition according to the state, and it becomes easier to recognize an appropriate object.

Although a mode has been described in which the vehicle sensor 12 includes all of the vehicle speed sensor 31, steering angle sensor 32, and tilt angle sensor 33, the vehicle sensor 12 is not limited to this, and includes the vehicle speed sensor 31, the steering angle sensor 32, and the tilt angle sensor It may be configured to include at least one of 33. For example, the process of determining the detection range R may be performed based on only one of the vehicle speed, steering angle, and left/right inclination angle of the vehicle 1. The vehicle sensor 12 also includes other sensors such as pitch sensors that detect the pitch angle of the vehicle 1, and determines the detection range R or corrects the detection range R using the detection results of the other sensors. You can do it like this.

Further, the detection range determination unit 41 reduces the angle of view of the detection range R when the vehicle speed is higher than a predetermined value, and increases the angle of view of the detection range R when the vehicle speed is lower than the predetermined value. According to this configuration, when the vehicle speed is higher than a predetermined value, the angle of view of the detection range R is made small, so that the detection range R can be set so that distant objects can be detected at high speeds. When the vehicle speed is lower than the predetermined value, the angle of view of the detection range R is increased, making it easier to recognize objects over a wider range. As described above, the predetermined value can be changed as appropriate. Further, the size of the angle of view of the detection range R is not limited to the above-mentioned embodiment, and may be determined as appropriate.

In addition, since the detection range determination unit 41 thins out the number of pixels in the detection range R according to the vehicle speed, when the detection range R increases depending on the vehicle speed, by thinning out the number of pixels in the detection range R, the recognition image GT The amount of data can be reduced, and the amount of calculation can be reduced. In this configuration, the detection range R becomes smaller when the speed is high, so the number of pixels within the detection range R is not thinned out, and when the speed is low, the detection range R is widened and the number of pixels within the detection range R is thinned out. . Therefore, the resolution of the recognition image GT can be maintained even at high speeds where blurring etc. are likely to occur in the photographed image GC, making it easier to appropriately recognize objects.

Furthermore, the detection range determination unit 41 estimates the course of the vehicle 1 based on the detection result of the vehicle sensor 12, and overlaps a part of the detection range R on the estimated course. According to this configuration, it is possible to obtain the recognition image GT on the course of the vehicle 1 while reducing the amount of data of the recognition image GT by reducing the detection range R, and to detect objects that affect the running of the vehicle 1. It becomes easier to recognize properly. In this case, according to the detection result of the steering angle sensor 32 included in the vehicle sensor 12, the detection range R can be shifted according to the steering angle to include the course, making it easier to detect objects at the turning destination.
In this embodiment, a case has been described in which the course of the vehicle 1 is specified based on the vehicle speed, the steering angle, and the left/right inclination angle of the vehicle 1. The information on either the left or right inclination angles may not be used, or the course of the vehicle 1 may be estimated using detection results from other sensors.

Furthermore, since the detection range determination unit 41 corrects the inclination of the detection range R according to the left and right angle of inclination, even if the vehicle 1 leans left and right, the recognition image GT of the detection range R is prevented from leaning left and right. can. This makes it easier to improve object detection accuracy and shorten detection time while reducing the amount of calculations.

Furthermore, when the vehicle 1 is turning, the detection range determination unit 41 sets the detection range R to include the position of the road G1 corresponding to the turning destination. Since the road G1 curves in the turning direction during a turn, there is a possibility that the position of the road G1 at the turning destination cannot be included in the detection range just by correcting the detection range according to the left/right inclination angle. When the vehicle 1 is turning, by setting the detection range R to include the position of the road G1 corresponding to the turning destination, objects that affect the running of the vehicle 1 can be easily recognized while reducing the amount of calculation. .
Note that the position of the road G1 corresponding to the turning destination corresponds to the "course corresponding to the turning destination" of the present invention.

The above-described embodiment is merely an example of one aspect of the present invention, and can be arbitrarily modified and applied without departing from the spirit of the present invention.

[Configuration supported by the above embodiment]
The above embodiment supports the following configurations.

(Configuration 1) A first sensor that detects vehicle information of a vehicle, a second sensor that detects surrounding conditions of the vehicle, and detects surrounding objects based on the detection information of the first sensor and the detection information of the second sensor. The recognition arithmetic processing device includes a detection range determination unit that determines a detection range of the surrounding situation based on the detection result of the first sensor, and an image of the detection range based on detection information of the second sensor. The recognition image includes an image generation unit that generates a recognition image, and an image recognition unit that recognizes an object to be detected from the recognition image, and the recognition image includes the recognition image obtained from the detection information of the second sensor. An arithmetic processing device characterized by having a smaller amount of data than an image of the surrounding situation.
According to this configuration, by making the detection range of the surrounding situation smaller than the detection range of the second sensor, it is possible to obtain a recognition image with a reduced amount of data. Thereby, it is possible to reduce the amount of calculation when recognizing the object to be detected from the recognition image. Since the amount of calculation can be reduced, it is possible to improve the speed of object detection, and it is also possible to eliminate the need for hardware resources such as high-performance arithmetic processing devices, thereby reducing the cost required for object detection.

(Configuration 2) The first sensor includes at least one of a vehicle speed sensor, a steering angle sensor, and a tilt angle sensor that respectively detect a vehicle speed, a steering angle, and a left/right tilt angle of the vehicle, and the detection range determining unit includes: The arithmetic processing device according to configuration 1, wherein the detection range is varied based on the detection result of the first sensor.
According to this configuration, the detection range is varied based on at least one of the vehicle speed, steering angle, and left/right inclination angle, so that the detection range can be made smaller to reduce the amount of calculations during image recognition, while the vehicle is It becomes easier to determine the detection range necessary for object recognition according to the state, and it becomes easier to recognize an appropriate object.

(Configuration 3) The detection range determining unit reduces the angle of view of the detection range when the vehicle speed is higher than a predetermined value, and increases the angle of view of the detection range when the vehicle speed is lower than a predetermined value. The arithmetic processing device according to configuration 2, characterized in that:
According to this configuration, when the vehicle speed is higher than a predetermined value, the angle of view of the detection range is reduced, so that a distant object can be detected at high speeds, and when the vehicle speed is lower than the predetermined value, the detection range is Since the angle of view of the range is increased, it becomes easier to recognize objects over a wider range.

(Configuration 4) The arithmetic processing device according to Configuration 3, wherein the detection range determining unit thins out the number of pixels in the detection range according to the vehicle speed.
According to this configuration, when the detection range increases depending on the vehicle speed, the data amount of the recognition image can be reduced by thinning out the number of pixels in the detection range, and the amount of calculation can be reduced.

(Structure 5) The arithmetic processing device according to any one of Structures 2 to 4, wherein the detection range determination unit corrects the inclination of the detection range according to the horizontal inclination angle.
According to this configuration, since the inclination of the detection range is corrected according to the left-right inclination angle, even if the vehicle inclines left and right, the recognition image of the detection range can be prevented from inclining left and right. This makes it easier to improve object detection accuracy and shorten detection time while reducing the amount of calculations.

(Configuration 6) Configurations 2 to 5, wherein the detection range determination unit estimates the course of the vehicle based on the detection result of the first sensor, and overlaps a part of the detection range on the estimated course. The arithmetic processing device according to any one of the above.
According to this configuration, it is possible to obtain a recognition image on the path of the vehicle while reducing the amount of data for the recognition image by reducing the detection range, making it easier to appropriately recognize objects that affect the running of the vehicle. Become.

(Structure 7) The arithmetic processing device according to Structure 6, wherein when the vehicle is turning, the detection range determination unit sets the detection range to a range that includes a course corresponding to a destination of the turn.
According to this configuration, since the detection range is set to include the course corresponding to the destination of the turn, it becomes easier to appropriately recognize objects that affect the running of the vehicle while reducing the amount of calculation.

1 Vehicle 10 Driving support device (computation processing device)
11 Surrounding detection sensor (second sensor)
12 Vehicle sensor (first sensor)
13 Arithmetic unit 14 Storage unit 15 Output unit 21 Camera 31 Vehicle speed sensor 32 Rudder angle sensor 33 Tilt angle sensor 41 Detection range determination unit 42 Image generation unit 43 Image recognition unit R Detection range GT Recognition image GC Photographed image (image of surrounding situation) )

Claims

A first sensor (12) that detects vehicle information of the vehicle, a second sensor (11) that detects the surrounding situation of the vehicle (1), and a detection information of the first sensor (12) and the second sensor ( 11) In the arithmetic processing device that recognizes surrounding objects based on the detection information,
a detection range determination unit (41) that determines a detection range of the surrounding situation based on the detection result of the first sensor (12);
an image generation unit (42) that generates a recognition image (GT) consisting of an image of the detection range (R) based on the detection information of the second sensor (11);
an image recognition unit (43) for image recognition of an object to be detected for the recognition image (GT);
The arithmetic processing device is characterized in that the recognition image (GT) has a smaller amount of data than the surrounding situation image (GC) obtained from the detection information of the second sensor (11).
The first sensor (12) is at least one of a vehicle speed sensor (31), a steering angle sensor (32), and a tilt angle sensor (33) that respectively detect the vehicle speed, steering angle, and left/right tilt angle of the vehicle (1). including
The arithmetic processing device according to claim 1, wherein the detection range determination unit (41) varies the detection range (R) based on the detection result of the first sensor (12).
The detection range determination unit (41) reduces the angle of view of the detection range (R) when the vehicle speed is higher than a predetermined value, and reduces the angle of view of the detection range (R) when the vehicle speed is lower than a predetermined value. The arithmetic processing device according to claim 2, characterized in that the angle of view is increased.
The arithmetic processing device according to claim 3, wherein the detection range determining unit (41) thins out the number of pixels in the detection range (R) according to the vehicle speed.
The arithmetic processing device according to any one of claims 2 to 4, wherein the detection range determination unit (41) corrects the inclination of the detection range (R) according to the left-right inclination angle.
The detection range determination unit (41) estimates the course of the vehicle (1) based on the detection result of the first sensor (12), and overlaps a part of the detection range (R) on the estimated course. The arithmetic processing device according to any one of claims 2 to 5, characterized in that:
7. The detection range determination unit (41) sets the detection range (R) to a range that includes a course corresponding to a turning destination when the vehicle (1) is turning. arithmetic processing unit.