JP7166582B2 - Vehicle surrounding recognition sensor inspection system, vehicle surrounding recognition sensor inspection method - Google Patents

Description

The present invention relates to an inspection system and the like for inspecting the installation orientation of a surrounding recognition sensor mounted on a vehicle.

2. Description of the Related Art In recent years, vehicles such as automobiles equipped with automatic brakes for collision avoidance and automatic driving have become popular. Vehicles detect conditions around the vehicle, such as oncoming vehicles and the road environment, with an on-vehicle sensor such as a millimeter-wave radar, and apply brakes. The premise is that the irradiation axis (sensor axis) of the in-vehicle sensor irradiates the vehicle in an appropriate direction (usually directly in front of the center). do.

However, if the vehicle is deformed or distorted due to an accident or the like, the sensor axis of the vehicle-mounted sensor may be displaced. However, there is a problem that it is difficult for the repair shop to accurately detect the displacement of the sensor axis. Similarly, at the vehicle inspection site, it is difficult to accurately detect the displacement of the sensor axis, so it has been difficult to establish inspection standards.

SUMMARY OF THE INVENTION In view of such circumstances, it is an object of the present invention to provide a system or the like for detecting the installation state of a surrounding recognition sensor mounted on a vehicle with high accuracy.

To achieve the above objects, the present invention provides a vehicle surroundings recognition sensor inspection system for inspecting the installation state of a surroundings recognition sensor mounted on a vehicle and capable of recognizing the orientation of objects existing around the vehicle, wherein the vehicle is scheduled to stop. A target unit that is installed in a state of being positioned in advance around an area and is recognized by the surrounding area recognition sensor, a vehicle measuring device that measures a vehicle that is stopped in the planned vehicle stop area, and a vehicle measuring device that is connected to the surrounding area recognition sensor. a sensor output acquisition device for acquiring a sensor output from the surrounding recognition sensor; and a computing device connected to the vehicle measuring device and the sensor output acquiring device, wherein the computing device receives the output from the vehicle measuring device. A direction of the target portion recognized by the surrounding recognition sensor using sensor output data obtained from an axle acquisition processing unit that calculates the axle of the vehicle based on the obtained vehicle measurement data, and the sensor output acquisition device. A sensor recognition angle acquisition processing unit that acquires a value (hereinafter referred to as a target direction value) regarding the an amount calculation processing unit, and a sensor axis deviation amount calculation processing unit that calculates an angle deviation amount of the sensor reference axis of the surroundings recognition sensor (hereinafter referred to as a sensor angle deviation amount) from at least the target direction value and the axle angle deviation amount. and a vehicle surroundings recognition sensor inspection system.

In relation to the vehicle surroundings recognition sensor inspection system, the computing device calculates an azimuth angle variation amount of the target portion caused by an offset of the surroundings recognition sensor (hereinafter referred to as an offset-caused angle deviation amount). A calculation processing unit is provided, and the sensor axis deviation amount calculation processing unit calculates the sensor angle deviation amount from the target direction value, the axle angle deviation amount, and the offset factor angle deviation amount.

In relation to the vehicle surrounding recognition sensor inspection system, the axle acquisition processing unit includes a feature region extraction processing unit that extracts feature region data including feature values for calculating the axle from the vehicle measurement data. and calculating the axle using the characteristic region data.

In relation to the vehicle surroundings recognition sensor inspection system, the vehicle measurement data includes outline data of the vehicle in a plan view, and the feature area data includes a side mirror area of the vehicle.

In relation to the vehicle surroundings recognition sensor inspection system, the axle acquisition processing unit has a feature value determination processing unit that determines a feature point for calculating the axle from the vehicle measurement data. The point is used to calculate the axle.

In relation to the vehicle surroundings recognition sensor inspection system, the vehicle measurement data includes outline data of the vehicle in a plan view, and the characteristic points include outline change points in side mirrors of the vehicle. and

In relation to the vehicle surroundings recognition sensor inspection system, the vehicle measurement data includes a contour shape of the entire circumference when the vehicle is viewed from above, and the axle acquisition processing unit performs and calculating the axle.

To achieve the above object, the present invention provides a vehicle surroundings recognition sensor inspection method for inspecting the installation state of a surroundings recognition sensor mounted on a vehicle and capable of recognizing the orientation of an object existing around the vehicle, the method comprising: A target portion is installed in a pre-positioned state around the region, the target portion is recognized by the surrounding recognition sensor, a vehicle stopped in the planned vehicle stop region is measured by a vehicle measuring device, and the Get the sensor output from the surrounding recognition sensor,

An axle of the vehicle is calculated based on vehicle measurement data obtained from the vehicle measurement device, and a direction of the target portion recognized by the surrounding recognition sensor using sensor output data obtained from the sensor output acquisition device. (hereinafter referred to as the target direction value), calculates the amount of deviation of the angle of the axle with respect to the stop reference axis of the planned stop area (hereinafter referred to as the amount of deviation of the axle angle), and obtains at least the target direction value and the axle angle A method for inspecting a vehicle surroundings recognition sensor, characterized in that a deviation amount of an angle of a sensor reference axis of the surroundings recognition sensor (hereinafter referred to as a sensor angle deviation amount) is calculated from the deviation amount.

ADVANTAGE OF THE INVENTION According to this invention, the outstanding effect that the installation state of the surrounding recognition sensor mounted in a vehicle can be detected with high precision can be produced.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is (A) a front view, (B) a partial cross-sectional plan view, and (C) a partial cross-sectional side view showing the overall configuration of a vehicle surroundings recognition sensor inspection system according to an embodiment of the present invention; It is the (A) front view and the (B) side view which show the imaging part of the same vehicle surroundings recognition sensor test|inspection system by enlarging. It is a block diagram showing a functional configuration of a computing device of the same vehicle surroundings recognition sensor inspection system. It is a figure which shows the three-dimensional data imaged with the imaging part of the same vehicle surroundings recognition sensor test|inspection system. (A) is a diagram showing feature regions extracted from the same three-dimensional data, and (B) is a diagram showing a technique for analyzing feature points. (A) to (C) are explanatory diagrams for explaining a calculation method executed by a calculation device of the same vehicle surroundings recognition sensor inspection system. (A) to (C) are explanatory diagrams for explaining a modification of the calculation method executed by the calculation device of the vehicle surroundings recognition sensor inspection system. (A) to (C) are explanatory diagrams for explaining a modification of the calculation method executed by the calculation device of the vehicle surroundings recognition sensor inspection system. FIG. 4 is a plan view showing a calibration technique for the double ambient recognition sensor inspection system; FIG. 11 is a plan view showing a modification of the dual surroundings recognition sensor inspection system; FIG. 11 is a plan view showing a modification of the dual surroundings recognition sensor inspection system;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

<Overall composition>
FIG. 1 shows the overall configuration of a vehicle surroundings recognition sensor inspection system (hereinafter referred to as an inspection system) 1 of this embodiment. The inspection system 1 inspects the installation state of the surrounding recognition sensor E mounted on the vehicle D. FIG. This inspection system 1 includes a target section 10 , a vehicle measurement device 20 , a sensor output acquisition device 40 and a calculation device 60 . It should be noted that here, as the surrounding recognition sensor mounted on the vehicle, it is assumed that a millimeter-wave radar that detects a vehicle running ahead is used, but a camera-type forward detection sensor and other sensors can also be used. Applicable.

The target unit 10 is a target object recognized by the surroundings recognition sensor E, and is installed in a state of being positioned in advance around the planned vehicle stop area 15 in which the vehicle D is to be stopped. In this embodiment, a reference axis J extending in the longitudinal direction of the vehicle is set in the planned vehicle stop area 15, and an origin O is also set at a predetermined position on the reference axis J. As shown in FIG. Along this reference axis J, a target unit 10 is installed, for example, 5 m ahead of the front end of the vehicle. The target unit 10 is, for example, a corner reflector having a triangular pyramidal depression, and can efficiently reflect the radar waves emitted from the vehicle D. As shown in FIG. The stop position of the vehicle D in the planned vehicle stop area 15 is such that the axle (vehicle center line) F coincides with the reference axis J as much as possible, and the position in the front-rear direction near the left and right side mirrors (door mirrors) of the vehicle D is Match the origin O. From the viewpoint of visibility from the driver's seat, there is a vehicle D in which the positions of the left and right door mirrors are shifted from each other. ), the line segment W connecting the door mirrors should pass through the origin O. However, as will be described later in detail, a certain amount of stopping error of the vehicle D is allowed. The axle F here means a center line extending in the front-rear direction at the center of the vehicle D in the width direction.

Further, although the present embodiment exemplifies the case where the target section 10 is arranged linearly along the reference axis J, the present invention is not limited to this. Assuming a vehicle D that is accurately stopped in the planned vehicle stop area 15, it can be arranged on the extension line of the sensor reference axis (which can take various directions) of the surrounding recognition sensor E of the vehicle D. This extension line is defined as a target reference axis for arranging the target section 10 . The reference axis J and the target reference axis may be translated (offset) or have a predetermined angular difference.

The vehicle measurement device 20 measures the vehicle D stopped in the planned vehicle stop area 15 . The purpose of this measurement is to detect the amount of deviation of the vehicle D from stopping. At a repair shop or vehicle inspection site, it is difficult to stop the vehicle D with a high degree of accuracy. Therefore, the vehicle measuring device 20 allows the inspection to be performed while allowing the deviation of the vehicle D to stop. The vehicle measuring device 20 is installed above the vehicle D and has an imaging unit 22 that acquires posture data of the vehicle D, and a frame unit 38 that is configured by a support or a bar that holds the imaging unit 22 . If the inspection space already has a ceiling of a building, the frame part 38 may be omitted and the imaging part 22 may be directly fixed to the ceiling.

As shown enlarged in FIG. 2, the imaging unit 22 is interposed between a base 24, a three-dimensional scanning device 26, and the base 24 and the three-dimensional scanning device 26 to rotate the three-dimensional scanning device 26. and a camera 30 installed on the base 24 .

The three-dimensional scanning device 26 can measure the shape (three-dimensional position) of an object by laser irradiation. Here, 3D-LiDAR (Light Detection and Ranging) is adopted, and more specifically, the product M8-1 from Quanergy Systems is used. The three-dimensional scanning device 26 has the advantage of being less affected by the external environment such as sunlight as compared with a camera. Note that this 3D-LiDAR is a full-circle rotary line scanner, and as shown in FIG. Scanning is possible in the width direction. Furthermore, in this embodiment, as shown in FIG. 2B, the entire vehicle D can be scanned by rotating the entire three-dimensional scanning device 26 in the longitudinal direction of the vehicle D using the table device 28. I have to.

The camera 30, like the three-dimensional scanning device 26, acquires an image of the vehicle D viewed from above. In this embodiment, data from the three-dimensional scanning device 26 is mainly used to perform various data processing. Therefore, the image of the camera 30 is also used for calibrating the three-dimensional scanning device 26 and for visually confirming the state of the vehicle D by the user (inspector). However, the imaging unit 22 of the present invention is not limited to the three-dimensional scanning device 26, and various methods such as a two-dimensional scanner, a distance sensor, and a camera can be adopted.

The three-dimensional scanning device 26 is installed directly above the origin 0, as shown in FIG. The installation height is preferably a place where the contour from the vehicle side to the front and rear ends can be measured without the laser beam being blocked by the ceiling (roof) of the vehicle D. For example, the height is 3 m or more, and here it is about 4 m. there is A plurality of imaging units 22 (three-dimensional scanning devices 26) may be prepared and installed at positions spaced apart in the front-rear direction and vehicle width direction to achieve wide-range measurement. It is also preferable to dispose the imaging unit 22 in the vicinity directly above a shape-like feature area/feature point (details will be described later). Further, for example, if the frame device 20 is improved so that the three-dimensional scanning device 26 can be slid (moved) in the longitudinal direction (or width direction) of the vehicle D, the imaging device 22 can be installed at a relatively low position. can also scan contours all around the vehicle.

The sensor output acquisition device 40 is electrically connected to the surroundings recognition sensor E mounted on the vehicle D and acquires the sensor output from the surroundings recognition sensor E. FIG. This output is, for example, angle data in the horizontal direction. When an object on the sensor reference axis R for the surroundings recognition sensor E is detected, the output is 0 degrees. Negative angle values to the left. This sensor output is sent to computing device 60 . In this embodiment, for example, a data scan tool G-Scan2 manufactured by Intersupport Co., Ltd. is used as the sensor output acquisition device 40, which is connected to the vehicle D to acquire the sensor output. Although the case where the sensor reference axis R coincides with the axle F of the vehicle D is exemplified here, it may be parallel (offset) with the axle F in some cases. Further, for example, in the case of a surrounding recognition sensor E for detecting an object on the side of the vehicle, the sensor reference axis R may have a preset angle with respect to the axle F. Further, this sensor reference axis R means a software main axis that is set from the viewpoint that the output data of the surrounding recognition sensor E is 0 degrees. may have different hardware arrangements.

The calculation device 60 is a so-called computer and is connected to the vehicle measurement device 20 and the sensor output acquisition device 40 . The computing device 60 is composed of a CPU, a RAM and a ROM, a storage device such as a hard disk, and the like. The CPU is a central processing unit that executes inspection programs and implements various functions. The RAM is used as a work area and storage area for the CPU, and the ROM and storage device store the operating system and programs executed by the CPU. Various data are accumulated in the storage device.

<Inspection processing>
Next, the flow of inspection processing by the inspection system 1 will be described, including the functional configuration of the computing device 60. FIG. As shown in FIG. 3, the computing device 60 includes, as functional configuration, an axle acquisition processing unit 62, a sensor recognition angle acquisition processing unit 70, an axle angle deviation calculation processing unit 74, a sensor axis deviation calculation processing unit 76, an offset factor It has an angle deviation amount calculation processing section 78 and a vehicle type feature database section 80 .

<Registration of feature information by vehicle type>
The vehicle-specific feature database unit 80 is implemented by a storage device, and stores vehicle-specific feature information data. For example, the feature information data includes positional information of feature areas or feature points that have shape features that exist in positions separated in the left-right or front-rear direction of the vehicle in the contour and shape data of the vehicle when viewed in plan. be This is because, if there are at least two unique points on the vehicle D, the axle can be specified based on those two points.

Therefore, it is preferable that the feature information data also include information (axle information) about the relative position or relative angular relationship of the axle F with reference to the plurality of feature areas or feature points. Even in the case of a vehicle D in which characteristic regions or characteristic points cannot be set at symmetrical positions, the axle F can be calculated based on a plurality of characteristic points by referring to this axle information. In this embodiment, left and right area data including left and right side mirrors are accumulated as specific characteristic area data. A range (that is, a range near the side mirrors) is set. In addition, in this embodiment, as specific feature point data, data related to feature points (unevenness) that are easy to distinguish from the shape of the left and right side mirrors are accumulated. It becomes a specific position in the direction and the left-right direction. Further, in this embodiment, the axle information based on the characteristic points of the side mirrors is registered for each vehicle type. Therefore, by referring to the feature information data and extracting (analyzing) the measurement data, the data area of the side mirror can be automatically extracted. Further, if the characteristic point can be detected from the data area of the side mirror, the axle F can be calculated based on the axle information.

It should be noted that, depending on the type of vehicle, it may be easier to extract shape feature points from other parts than from left and right side mirrors. That is, the present invention is not limited to the side mirrors, and may set the range of other characteristic regions. For example, the left and right tail lamp areas, the left and right door knob areas, the four corner areas of the vehicle, and the like may be used. Furthermore, not limited to a plurality of characteristic regions, for example, boundary lines of window frames, line segments around the bonnet, line segments around the body, line segments around the roof, line segments around the trunk, body surface A single feature region may be used when the predetermined length or area itself has a shape feature such as a design line to be created. This is because the axle can be calculated based on this.

In addition, as other information, position information of the surrounding recognition sensor from the feature points (for example, two feature points of the side mirrors), position information of the center of the vehicle from the feature points, and position information of the surrounding recognition sensor from the center of the vehicle. Location information and the like may be accumulated for each vehicle type. Also, the permissible value (pass/fail judgment criteria) of the amount of axis deviation of the surrounding recognition sensor may be accumulated. This is because the allowable amount of misalignment differs for each vehicle type. Furthermore, as other information, the sensor reference axis set in the surrounding recognition sensor (this is defined as the "ideal sensor reference axis") is parallel (offset) to the axle F, If it has an angle, information about this ideal sensor reference axis (eg, relative angle and offset amount) may be stored. For example, information for calculating an "ideal sensor reference axis" based on feature points and feature regions may be registered. Similarly, information on the target reference axis, which means the placement position of the target portion for inspection, may be accumulated for each surrounding recognition sensor to be inspected. For example, information for calculating a "target reference axis" based on feature points and feature areas may be registered. Also, information about the relative difference (offset value or angle difference) between the "ideal sensor reference axis" and the "target reference axis" may be accumulated.

<Axle Acquisition Processing>
The axle acquisition processing unit 62 calculates the axle F of the vehicle D based on the vehicle measurement data obtained from the vehicle measurement device 20 . Specifically, the axle acquisition processing section 62 has a feature area extraction processing section 64 and a feature value determination processing section 66 . The characteristic region extraction processing unit 64 extracts characteristic region data including characteristic values for calculating the axle F from the vehicle measurement data.

FIG. 4 visualizes vehicle measurement data obtained from the vehicle measurement device 20 . This vehicle measurement data becomes three-dimensional data (three-dimensional data) of the surface shape of the vehicle when the vehicle is viewed from information, and includes contour (outer edge) data of the vehicle D in plan view. The characteristic area extraction processing unit 64 extracts stereoscopic data in the vicinity of the left and right side mirror areas from the vehicle measurement data, as shown in FIGS. 5A and 5B, for example. For example, the position information of the left and right side mirrors accumulated in the vehicle-specific feature database unit 80 may be referred to, and the specified range (range near the side mirrors) may be extracted from the vehicle measurement data. Since the vehicle D will be stopped within the scheduled vehicle stop area 15 with a certain amount of accuracy, the side mirror area registered in the vehicle-specific feature database unit 80 is set wide. If you do, no extraction error will occur. The computing device 60 preferably has a comparison display processing section that superimposes (or contrasts) the image captured by the camera 30 and the data area extracted by the feature area extraction processing section 64 on a monitor. As a result, the measurement person confirms the image captured by the camera 30 on the monitor. The person may select the extraction area while checking the image of the vehicle D, and the characteristic area extraction processing unit 64 may receive the selection instruction to re-extract the stereoscopic data around the left and right side mirror areas. good. Here, for example, the case of using the data of the vehicle-specific feature database unit 80 is illustrated, but the feature region extraction processing unit 64 is provided with a machine learning function, and vehicle measurement data of various vehicle models and left and right side mirror data are used. By learning the range, it may be possible to automatically extract characteristic regions and characteristic points such as the positions of left and right side mirrors.

The feature value determination processing unit 66 refers to the side mirror area extracted from the vehicle measurement data to determine a shape feature point (singular point) for calculating the axle F. For example, as indicated by the solid-line arrow in FIG. 5(B), a group of points positioned on the outermost side in the width direction of the side mirror area is traced in one direction, and the direction of the traveling vector that connects the consecutive points changes along the way. A feature point T may be a point at which there is an abrupt change. The determination method of this feature point T is not particularly limited, but it is important for accurate calculation of the axle F to be able to extract the same feature point T from the left and right side mirrors. Therefore, a method capable of reliably determining the characteristic point T should be adopted in consideration of the shape characteristics of the side mirror. For example, a feature point T2 at the base of the side mirror may be employed, as indicated by the dotted line arrow. As already mentioned, feature points at locations other than the side mirrors may be employed.

The computer 60 preferably superimposes (or contrasts) the image captured by the camera 30 and the feature points T determined by the feature value determination processing unit 66 on the monitor by the comparison display processing unit. If the feature point T determined by the feature value determination processing unit 66 clearly contains an error, the person who measures the vehicle checks the image of the vehicle D on the monitor, manually designates the feature point T, and performs this designation instruction. can be set as the feature point T by the feature value determination processing unit 66 receiving the .

After recognizing a pair of feature points T from the left and right side mirrors through the above steps, the axle acquisition processing unit 62 refers to the axle information in the vehicle-specific feature database unit 80, and based on the position of the feature point T, the axle Calculate F. If the shape of the left and right side mirrors is completely symmetrical, the vehicle width line W is determined by the line segment connecting TT, and the vertical bisector of this vehicle width line W is determined. The axle F may be determined (see FIG. 5(A)).

<Sensor Recognition Angle Acquisition Processing>
As shown in FIG. 6A, the sensor recognition angle acquisition processing unit 70 uses the sensor output data obtained from the sensor output acquisition device 40 to obtain a value ( Get the target direction value γ).

<Axle Angle Deviation Amount Calculation Processing>
As shown in FIG. 6A, the axle angle deviation calculation processing unit 74 calculates the angle deviation of the axle F relative to the stop reference axis J in the planned stop area 15 (axle angle deviation α). Since the vehicle measurement device 20 is calibrated in advance using the vehicle stop reference axis J, it is possible to determine the angular deviation amount α from the vehicle stop reference axis J in the vehicle measurement data. Therefore, the axle angle deviation amount calculation processing section 74 compares the data of the axle F obtained by the axle acquisition processing section 62 with the known stop reference axis J to calculate the axle angle deviation amount α.

<Basic Concept of Sensor Axis Deviation Amount Calculation Processing>
The sensor axis deviation calculation processing unit 76 calculates the angle of the sensor reference axis R of the surrounding recognition sensor E with respect to the ideal sensor reference axis I (which coincides with the axle F in this embodiment) from at least the target direction value γ and the axle angle deviation amount α. A deviation amount (hereinafter referred to as a sensor angle deviation amount C) is calculated. FIG. 6A exemplifies the case where the ideal sensor reference axis I (axle F) and the sensor reference axis R coincide (when the sensor angle deviation amount C is 0), but FIG. As shown in (C), even if the axle angle deviation amount α does not change, if the ideal sensor reference axis I (axle F) and the sensor reference axis R deviate, the deviation angle is reflected in the target direction value γ. It can be seen that

<Calculation of Sensor Axial Deviation Considering Offset Factor Angular Deviation Amount>
In calculating the sensor angular deviation amount C with higher accuracy, first, the offset factor angular deviation amount calculation processing unit 78 calculates a line segment U connecting the surrounding recognition sensor E and the target unit 10 and a target reference axis V (this embodiment In the form, the angle difference β generated between the vehicle stop reference axis J and (coincided with the vehicle stop reference axis J) is calculated. This angle difference β is defined as "offset factor angle deviation amount β". This offset factor angular deviation amount is the azimuth angle of the target unit 10 from the surroundings recognition sensor E, which is caused only by the offset amount (sensor offset amount) G of the surroundings recognition sensor E with respect to the target reference axis V (stopping reference axis J). is the amount of change. As already described, the "position of the surrounding recognition sensor E" here is not limited to the position as hardware, but the software of the surrounding recognition sensor E set on the program of the surrounding recognition sensor E. Including the case of the virtual position above.

In addition, in this embodiment, the case where the surroundings recognition sensor E exists in the front end of the vehicle on the axle F is illustrated. Therefore, by calculating the position where the front edge of the vehicle D intersects the axle F, the position of the surrounding recognition sensor E can be determined directly. Therefore, from the position of the surrounding recognition sensor E and the position of the target unit 10, a line segment U connecting the two is geometrically calculated, and the angle at which the line segment U intersects with the stop reference axis J is calculated. A factor angle deviation amount β is obtained. The method of determining the position of the surrounding recognition sensor E refers to the relative position information of the surrounding recognition sensor E from the feature point T registered in the vehicle-specific feature database unit 80, and the feature point T of the actual vehicle measurement data. Alternatively, the actual position of the surrounding recognition sensor E may be calculated directly. Further, the determination method is not limited to these methods, and for example, position information of the surrounding recognition sensor E based on the axle F, the vehicle center L, etc. is registered in the vehicle-specific feature database unit 80, and this data is referred to. Alternatively, the position of the surrounding recognition sensor E may be calculated directly from the vehicle measurement data. Of course, the position of the surrounding recognition sensor E may be calculated by extracting the line of the front end of the vehicle from the vehicle measurement data and using that line as a reference. Of course, the target reference axis V and the ideal sensor reference axis I can also be calculated from the feature points T and the like by referring to the vehicle-specific feature database section 80 .

As shown in FIG. 6A, when the sensor angular deviation amount C is 0, that is, when the ideal sensor reference axis I (axle F) and the sensor reference axis R match, the origin O, the surrounding recognition sensor E, the target From the cosine theorem of the triangle connecting the part 10, the following (Equation 1) is established.
(Formula 1) γ = α + β

Next, as shown in FIG. 6B, it is assumed that the sensor reference axis R is deviated from the ideal sensor reference axis I (axle F) by the sensor angle deviation amount C. FIG. At this time, the output value of the target direction value γ reflects (includes) the sensor angular deviation amount C in excess, so the following (Equation 2) is established.
(Formula 2) γ-C = α + β

Therefore, the actual sensor angle deviation amount C is obtained by converting (Equation 2) into the following (Equation 3).
(Formula 3) C = γ-(α + β)

As described above, the sensor axis deviation amount calculation processing unit 76 can calculate the sensor angle deviation amount C based on the target direction value γ, the axle angle deviation amount α, and the offset factor angle deviation amount β. If the sensor angle deviation amount C is calculated in consideration of the offset factor angle deviation amount β in this way, the calculation accuracy of the sensor angle deviation amount C is dramatically improved even if the accuracy of the stop position of the vehicle D is poor. However, if the operator has high accuracy in stopping the vehicle D and, as a result, the offset factor angular deviation amount β can always be kept small, the offset factor angular deviation amount β can be omitted.

<Verification when the vehicle center is offset>
In an actual measurement site, as shown in FIG. 6C, the vehicle D may stop with the vehicle center L offset in the direction perpendicular to the target reference axis V (here, the vehicle stop reference axis J). be. This offset amount (vehicle center offset amount) Z of the vehicle center L is inherent in the sensor offset amount G. FIG. In other words, the offset factor angular deviation amount calculation processing unit 78 calculates the sensor offset amount G including the vehicle center offset amount Z. FIG.

As the vehicle center offset amount Z increases, the sensor offset amount G increases accordingly. As a result, the offset factor angle deviation amount β increases, and at the same time, the target direction value γ detected by the surrounding recognition sensor E also increases. That is, due to the increase or decrease in the vehicle center offset amount Z, the offset factor angular deviation amount β and the target azimuth value γ increase or decrease by the same amount, so that they are canceled by the above (Equation 3). As a result, even if the vehicle center L is offset as shown in FIG. 6C, the actual sensor angle deviation amount C can be calculated by (Equation 3).

<When the target reference axis V and the stop reference axis J are different, or when the ideal sensor reference axis I and the axle F are different>
FIG. 7 shows a calculation example when the target reference axis V and the stop reference axis J are different, and the ideal sensor reference axis I and the axle F are different. As can be understood from FIGS. 7A to 7C, if the target reference axis V and the ideal sensor reference axis I can be set in the calculation device 60, the sensor angle deviation amount C can be calculated by the completely same procedure as in FIG. It turns out that it can be calculated. In FIGS. 6 and 7, the basic target reference axis V and the ideal sensor reference axis I are originally aligned. Thus, the case where the same axle angle deviation amount α occurs between the basic target reference axis V and the ideal sensor reference axis I is exemplified. However, the present invention is not limited to this, and for example, as shown in FIG. obtain. In this case, the sum of the axle angle deviation amount α and the initial angle difference x in the above equations 1 to 3 may be replaced with the axle angle deviation amount α. That is, the above formulas 1, 2 and 3 can be replaced with the following formulas 4, 5 and 6. These equations 4, 5, and 6 are necessary, for example, when the target portion 10 cannot be arranged on the extension line of the ideal sensor reference axis I due to the inspection space.
(Formula 4) γ = (α + x) + β
(Formula 5) γ-C = (α + x) + β
(Formula 6) C = γ-(α + x + β)

<Pass/Fail Judgment>

Note that the computing device 60 may include a pass/fail determination processing unit, although not particularly illustrated as a functional configuration here. The pass/fail determination processing unit makes a pass/fail determination based on whether or not the sensor angular deviation amount C exceeds a predetermined threshold value. Since this threshold value may differ depending on the vehicle type, it is also preferable to refer to the vehicle-specific feature database unit 80 and adopt the threshold value for each vehicle type.

<Initial setting or calibration>

Initial setting or calibration of the inspection system 1 will be described. As shown in FIG. 9, a calibrating unit P that can be scanned by the vehicle measuring device 20 is arranged in the longitudinal and lateral directions of the reference axis J in the planned vehicle stop area 15 and at the origin 0. As shown in FIG. When the imaging unit 22 is the three-dimensional scanning device 26, the calibration unit P is preferably a three-dimensional object, such as a post. When the imaging unit 22 is a camera, a marker or the like that can specify the position by image recognition is preferable. In this state, if the vehicle measuring device 20 detects the calibration portion P, the reference axis J can be newly set or calibrated easily. If the target reference axis has an offset or angle with respect to the reference axis J, it is desirable to calibrate this target reference axis as well.

As described above, according to the inspection system 1 of the present embodiment, it is possible to calculate the sensor angle deviation amount C with extremely high accuracy with a simple operation at a repair shop, a vehicle inspection site, etc. It is possible to adjust the installation posture. In particular, regarding the stop position of the vehicle D, an angle error and an offset error can be allowed, so the burden on the measurer can be greatly reduced. In particular, focusing on the side mirrors, which are present in all vehicles and tend to include characteristic points in shape, and calculating the axle F from the analysis, it is possible to obtain the axle F of various vehicle models with high accuracy. It becomes possible.

In the above embodiment, the imaging unit is a three-dimensional scanner, and the axle F is calculated from a three-dimensional shape. You can calculate. At this time, for example, as shown in FIG. 10, the axle F may be calculated by fixing a marker M to the vehicle D with an adhesive, a magnet, or the like, and recognizing the image of the marker M. FIG. This idea can also be applied to a three-dimensional scanner, and by installing a three-dimensional marker M that subsequently creates a shape singularity in the front-rear direction of the vehicle D or on the side mirror, the three-dimensional scanner can The axle F may be calculated by recognizing M. In other words, the feature point is not limited to the vehicle D itself, and may be a feature point added after the fact. Furthermore, the marker M is not limited to being attached to the upper surface of the vehicle D. For example, as shown in FIG. You may arrange|position the marker M so that it may contact|connect the side surface of . Here, a case is shown in which a vertically extending pillar is erected from the ground so as to contact a corner (recess) formed by the base of the side mirror and the side surface of the vehicle D, and the upper end of the pillar is used as the marker M. there is Such a marker M also serves as a feature point in terms of shape for the vehicle D. As shown in FIG.

Furthermore, in the above-described embodiment, the case where the axle F is calculated based on the shape feature points existing in the bilaterally symmetrical positions of the vehicle D has been exemplified, but the present invention is not limited to this. For example, the axle can be calculated by comprehensively determining the contour shape (including area information) of the entire periphery. At this time, a computer machine learning function (for example, deep learning) may be used. Further, for example, the center of gravity in the longitudinal direction, the center of gravity in the width direction, and the overall center of gravity may be calculated from the planar image of the vehicle D, and the axle F may be calculated from this information on the center of gravity. It is also possible to measure the vehicle D from the sides, the front and rear sides, and the bottom side.

Furthermore, in the above embodiment, the case of inspecting the axial deviation of the sensor that measures the vehicle ahead of the vehicle D is illustrated, but the present invention is not limited to this, and can be applied to side sensors and rear sensors. It is also possible to test multiple sensors simultaneously.

It should be noted that the inspection system of the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made without departing from the gist of the present invention.

1 vehicle circumference recognition sensor inspection system 10 target unit 20 vehicle measurement device 40 sensor output acquisition device 60 calculation device 62 axle acquisition processing unit 70 sensor recognition angle acquisition processing unit 74 axle angle deviation calculation processing unit 76 sensor shaft deviation calculation processing unit 78 Axle offset calculation processing unit 80 Vehicle-specific feature database unit

Claims (8)

A vehicle surroundings recognition sensor inspection system for inspecting the installation state of a surroundings recognition sensor mounted on a vehicle and capable of recognizing the orientation of an object existing around the vehicle,
a target unit that is positioned in advance around the planned vehicle stop area and that is recognized by the surrounding recognition sensor as the orientation of the object;
a vehicle measuring device that measures a vehicle that is stopped in the planned vehicle stop area;
a sensor output acquisition device connected to the surrounding recognition sensor and acquiring a sensor output from the surrounding recognition sensor;
a computing device connected to the vehicle measurement device and the sensor output acquisition device;
The computing device
an axle acquisition processing unit that calculates an axle of the vehicle based on vehicle measurement data obtained from the vehicle measurement device;
a sensor recognition angle acquisition processing unit that uses sensor output data obtained from the sensor output acquisition device to acquire a value related to the direction of the object by the target unit recognized by the surrounding recognition sensor (hereinafter referred to as a target direction value); ,
an axle angle deviation amount calculation processing unit that calculates the amount of deviation of the angle of the axle with respect to the stop reference axis in the planned stop area (hereinafter referred to as the axle angle deviation amount);
a sensor axis deviation amount calculation processing unit that calculates an angle deviation amount of a sensor reference axis of the surrounding recognition sensor (hereinafter referred to as a sensor angle deviation amount) from at least the target direction value and the axle angle deviation amount;
has
The computing device includes an offset factor angular deviation amount calculation processing unit that calculates an azimuth angle variation amount of the target portion caused by an offset of the surrounding recognition sensor (hereinafter referred to as an offset factor angular deviation amount),
The vehicle surrounding recognition sensor inspection system, wherein the sensor axis deviation amount calculation processing unit calculates the sensor angle deviation amount from the target direction value, the axle angle deviation amount, and the offset factor angle deviation amount. The computing device further comprises a vehicle-specific feature database section in which feature information data including shape features for each vehicle type is accumulated,
The axle acquisition processing unit calculates the axle of the vehicle by referring to the characteristic information data corresponding to the vehicle type of the vehicle to be measured,
The vehicle surroundings recognition sensor inspection system according to claim 1 . The axle acquisition processing unit has a feature region extraction processing unit that extracts feature region data including feature values for calculating the axle from the vehicle measurement data, and uses the feature region data to Calculating the axle,
The vehicle surrounding recognition sensor inspection system according to any one of claims 1 and 2 . The vehicle measurement data includes contour data of the vehicle in plan view,
The feature area data includes the side mirror area of the vehicle,
The vehicle surroundings recognition sensor inspection system according to claim 3 . The axle acquisition processing unit has a feature value determination processing unit that determines a feature point for calculating the axle from the vehicle measurement data, and uses the feature point to calculate the axle. characterized by
The vehicle surrounding recognition sensor inspection system according to any one of claims 1 to 4 . A vehicle surroundings recognition sensor inspection system for inspecting the installation state of a surroundings recognition sensor mounted on a vehicle and capable of recognizing the orientation of an object existing around the vehicle,
a target unit that is positioned in advance around the planned vehicle stop area and that is recognized by the surrounding recognition sensor as the orientation of the object;
a vehicle measuring device that measures a vehicle that is stopped in the planned vehicle stop area;
a sensor output acquisition device connected to the surrounding recognition sensor and acquiring a sensor output from the surrounding recognition sensor;
a computing device connected to the vehicle measurement device and the sensor output acquisition device;
The computing device
an axle acquisition processing unit that calculates an axle of the vehicle based on vehicle measurement data obtained from the vehicle measurement device;
a sensor recognition angle acquisition processing unit that uses sensor output data obtained from the sensor output acquisition device to acquire a value related to the direction of the object by the target unit recognized by the surrounding recognition sensor (hereinafter referred to as a target direction value); ,
an axle angle deviation amount calculation processing unit that calculates the amount of deviation of the angle of the axle with respect to the stop reference axis in the planned stop area (hereinafter referred to as the axle angle deviation amount);
a sensor axis deviation amount calculation processing unit that calculates an angle deviation amount of a sensor reference axis of the surrounding recognition sensor (hereinafter referred to as a sensor angle deviation amount) from at least the target direction value and the axle angle deviation amount;
has
The axle acquisition processing unit has a feature value determination processing unit that determines a feature point for calculating the axle from the vehicle measurement data, calculates the axle using the feature point,
The vehicle measurement data includes contour data of the vehicle in plan view,
The vehicle surroundings recognition sensor inspection system, wherein the characteristic points include contour change points in the side mirrors of the vehicle. The vehicle measurement data includes a contour shape of the entire periphery when the vehicle is viewed from above,
The axle acquisition processing unit calculates the axle based on the contour shape of the entire circumference,
The vehicle surrounding recognition sensor inspection system according to any one of claims 1 to 6 . A vehicle surroundings recognition sensor inspection method for inspecting the installation state of a surroundings recognition sensor that is applicable to a plurality of types of vehicles of different vehicle types and that is mounted on the vehicle and capable of recognizing the orientation of an object existing around the vehicle. There is
A target portion is installed in a state of being positioned in advance around the planned vehicle stop area, and the orientation of the object by the target portion is recognized by the surrounding recognition sensor,
measuring the vehicle stopped in the vehicle stop planned area from the outside by a vehicle measuring device fixed to the outside of the vehicle;
Acquiring a sensor output from the surrounding recognition sensor by a sensor output acquisition device detachably connected to the surrounding recognition sensor;
calculating an axle of the vehicle based on vehicle measurement data measured from the outside of the vehicle obtained from the vehicle measurement device;
Using the sensor output data obtained from the sensor output acquisition device, a value related to the direction of the object recognized by the surrounding recognition sensor (hereinafter referred to as a target direction value) by the target unit and the offset generated by the surrounding recognition sensor Acquire the amount of azimuth angle variation of the target part (hereinafter referred to as offset factor angle deviation amount) ,
calculating the amount of deviation of the angle of the axle with respect to the stop reference axis of the planned stop area (hereinafter referred to as the amount of deviation of the axle angle);
An angle deviation amount of the sensor reference axis of the surroundings recognition sensor (hereinafter referred to as a sensor angle deviation amount) is calculated from at least the target direction value, the axle angle deviation amount, and the offset factor angle deviation amount ,
Vehicle surrounding recognition sensor inspection method.

Images