JPH0646886B2 - Steering control device for automated vehicle - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of use] The present invention captures an image of a work ground condition in a predetermined range in the forward direction of a vehicle body by an image capturing device, and captures the captured image information based on the brightness of each pixel. Binarizing means for binarizing, boundary detecting means for detecting a line segment on the captured image information corresponding to the boundary between the unprocessed work site and the processed work site based on the information of the binarizing means, and The present invention relates to a steering control device for an automatic traveling work vehicle, which includes control means for steering the vehicle body so as to automatically travel along the boundary based on the boundary detection result by the boundary detection means.

[Conventional technology]

Such a steering control device is used to automatically drive a work vehicle, which is to perform work while traveling along a boundary between an unprocessed work site and a processed work site, along the boundary. .

Then, conventionally, using a photointerrupter type or a light reflection type non-contact type sensor or a contact type sensor,
I was trying to detect the boundaries. (For example, JP-A-60-186
However, in the prior art, since only one point of the boundaries existing linearly along the vehicle traveling direction is detected, for example, when the boundaries are partially uneven,
In some cases, the detected information may be unsuitable as steering control information for the vehicle, such that it is better to move the vehicle straight, but the driver may unnecessarily steer the vehicle. there were. Moreover, if there is a part of the boundary that is not clear, there is a risk of erroneous detection, and erroneous steering control will be carried out accordingly, resulting in partial unevenness at the boundary. There has been a demand for a boundary detection means that can detect appropriate information even if it exists or an unclear part of the boundary exists.

Therefore, the applicant of the present invention detects the boundary as a continuous line by binarizing image information obtained by imaging a predetermined range of the work site on the front side in the vehicle body traveling direction based on the brightness of each pixel. Has been previously proposed (Japanese Patent Application No. 59-
No. 211637).

[Problems to be solved by the invention]

The boundary detecting means for detecting the boundary by binarizing the image information based on the brightness of each pixel is intended to determine the boundary based on the difference in brightness between the unprocessed work site and the processed work site. is there.

Therefore, if the unprocessed work site and the processed work site are not clearly separated due to the difference in lightness value in the entire area of the image, there is an inconvenience that many false detections occur. That is, for example, if there is a dark portion in the processed work site or a bright portion in the unprocessed work site, it is not possible to distinguish it from the future boundary.

The present invention has been made in view of the above circumstances, and an object of the present invention is to satisfy the above-mentioned demand while effectively using an image pickup unit, and further, to perform image processing required by using the image pickup unit. Another object of the present invention is to provide a steering control device for an automated guided vehicle equipped with a boundary detection means that can be efficiently operated at high speed.

[Means for solving problems]

In the characteristic configuration of the steering control device for an automated guided vehicle according to the present invention, the binarizing means calculates a differential value corresponding to a change in lightness of each pixel of the captured image information in a direction along the lateral direction of the vehicle body. Of the discrimination code by the differential value computing means, the differential value code discriminating means for discriminating the positive / negative sign of the differential value, the code selecting means for selecting one of the positive / negative symbols of the differential value, and the discrimination code by the differential value code discriminating means. It is characterized in that it comprises extraction means for extracting a pixel in which the change in the differential value of the code selected by the code selection means is equal to or more than a predetermined value, and its action and effect are as follows.

[Action]

That is, a predetermined range on the front side in the traveling direction of the vehicle body is imaged, a differential value corresponding to a change in brightness of each pixel of the captured image information in the direction along the vehicle lateral width direction is obtained, and the sign selecting means of the differential value is used. Only the differential value showing the sign (positive or negative) on the selected side is extracted, and further the extracted differential value is equal to or more than a predetermined value, so that the image information is binarized. The least squares method and the binary image information
The straight line corresponding to the boundary is detected by the Haugh transform.

The code selected by the code selection means at this time is determined by the positional relationship between the unprocessed work site and the processed work value in the vehicle body traveling direction.

That is, for example, when the processed work site is adjacent to the right side in the vehicle body traveling direction as shown in FIG. 4 (b), the captured image shows a dark change in brightness as shown in FIG. 2 (a). → Because it shows a negative (-) differential value that becomes clear, the sign selecting means is configured to select negative when the processed work site is adjacent to the right side of the vehicle body traveling direction. When the processed work site is adjacent to the left side in the vehicle body traveling direction, positive is selected.

However, the sign of the above-mentioned differential value shows the case where the brightness change defines light → dark as a positive (+) differential value and dark → bright as a negative (−) differential value.

By the way, since the above-mentioned captured image information corresponds to the situation of the work area in the predetermined area on the front side in the vehicle body traveling direction, the boundary line detected as described above is a boundary existing along the vehicle body traveling direction. However, it corresponds to a boundary portion having a length along the vehicle body traveling direction.

Further, since the vehicle body travels along the boundary, the boundary exists along the vehicle traveling direction, and
Focusing on the point that the boundary can be detected by the change in brightness along the width direction of the vehicle body, and effectively utilizing that point, the differential value corresponding to the change in brightness along the width direction of the vehicle body of each pixel is calculated. However, since the differential value of the brightness change of each pixel in the direction along the vehicle body front-rear direction is also calculated and calculated, noise in the direction along the vehicle body front-rear direction can be detected. The components can be removed in advance and binarized, and the calculation and the image processing in the direction along the vehicle front-rear direction can be omitted.

Moreover, among the differential values corresponding to the brightness change in the width direction of the vehicle body, only the differential value showing one sign is detected while being calculated, so that it is included in the differential value showing the other sign. It is possible to remove the generated noise component in advance and binarize it, and it is also possible to omit the calculation processing and the image processing with the differential value indicating the other sign.

That is, the noise component in the direction along the vehicle body front-rear direction and the noise component included in the differential value indicating the other sign can be removed in advance from the captured image information and binarized.

〔The invention's effect〕

Therefore, from the captured image information, the noise component in the direction along the front-rear direction of the vehicle body and the noise component included in the differential value indicating the other sign can be removed in advance and binarized. Even if there is information that is difficult to remove after binarization, that is, information that is easily misjudged as a boundary, in other words, even if there is a dark portion in the processed work site or in the unprocessed work site. Even if there is a bright-looking portion, it is possible to efficiently remove in advance the information that is erroneously determined and that is included in the differential value that indicates the direction along the vehicle body front-rear direction and the other sign, and as a result, information that is easily erroneously determined. Thus, it is possible to avoid obtaining wrong boundary information and obtain more accurate boundary information.

Moreover, since each processing can be speeded up by omitting extra calculation processing and image processing, the invention can be applied to a vehicle body running at high speed. In other words, it is possible to avoid the disadvantage that the vehicle body has to run at a low speed to detect the boundary.

Therefore, it has become possible to obtain a traveling control device for an automatic traveling work vehicle capable of detecting excellent information on the steering control of the vehicle body without causing a disadvantage on the traveling surface of the vehicle body.

〔Example〕

 Embodiments of the present invention will be described below with reference to the drawings.

As shown in FIG. 5 and FIG. 6, a disk is provided in the middle portion of the vehicle body (1) configured to steer all of the front wheels (2), (2) and the rear wheels (3), (3). A lawn mowing device (4) with a built-in cutting blade is suspended vertically so that it can be moved up and down.
(B) and a copy sensor as a boundary detecting means having a structure described below for detecting a boundary (L) between the already cut land (C) as a processed work site
(A) is provided, and steering control is performed based on the boundary detection result by the scanning sensor (A), thereby forming a lawnmower work vehicle as an automatic work vehicle capable of automatically traveling along the boundary (L). There is.

As shown in FIG. 1, a monitor camera (5) as an image pickup means is included in the scanning sensor (A) so that the image pickup field of view is on the front side in the traveling direction of the vehicle body (1) and travels. Provided at the tip of the sensor support frame (6) extending toward the front upper direction of the vehicle body (1) so that the lawn (D) within a predetermined range around the target boundary (L ₀ ) to be centered, The image captured by the monitor camera (5) is image-processed by the control device (8), and a straight line corresponding to the boundary (L) (however, may be described as a detection boundary (L) in the following description). Is detected. Incidentally, the control device (8) also performs processing of steering control operation, specifically, the target boundary.
The deviation amount (X) of the detection boundary (L) with respect to (L ₀ ) and the deviation angle (θ) are calculated so that the detection boundary (L) and the target boundary (L ₀ ) match, that is, the detection boundary (L). (L) so that it is located at the center of the visual field in the front-back direction with respect to the vehicle body (1) of the monitor camera (5),
The steering operation is controlled to correct the traveling direction.

Below, the boundary (L) from the image captured by the monitor camera (5)
The configuration for detecting is described in detail with reference to the block diagram shown in FIG. 1, the explanatory diagram of the image signal shown in FIG. 2, and the flowchart shown in FIG.

That is, the monitor camera is controlled by the control device (8).
A differential value calculating means for calculating a differential value corresponding to a change in brightness of each pixel in a direction along the vehicle body width direction based on the captured image information (hereinafter referred to as the original image signal (S ₀ )) by (5). ,
Differential value sign determination means for determining the positive / negative sign of the differential ground,
Code selection means for selecting one of the positive and negative signs of the differential value, and a pixel having a differential value showing a code selected by the code selection means out of the differential values from the differential value calculation means is a predetermined value or more. Binarizing means having an extracting means for performing the same, and image information from the binarizing means (hereinafter, the binarized image (S
Boundary detecting means for obtaining a straight line corresponding to the boundary (L) is configured based on ( ₀ )), and the processing corresponding to each means is sequentially executed as described below.

However, the original image signal (S ₀ ) is subjected to an averaging process in advance, and the averaged image signal (S ₁ ) is differentiated by the differential value computing means.

Further, in the present embodiment, a change in lightness from light to dark is defined as a positive (+) differential value, and a change in darkness to light is defined as a negative (-) differential value.

The code selected by the code selecting means is the vehicle body (1).
It is determined by the positional relationship between the unprocessed work site and the processed work value with respect to the traveling direction of the, and as shown in FIG. 3, the position of the processed work value (C) is either left or right, For example, by manually inputting an initial value, the sign of the differential value at the time of starting work is selected and set in advance to a sign from positive (+) or negative (-), and the vehicle body (1) is turned by 180 degrees. In addition, the sign of the selected sign is automatically reversed.

That is, at the start of work, when the direction of the position of the processed work site (C) is initially input as, for example, the right side, the sign to be selected by the sign selecting means is initially set to negative (-).

Further, the original image signal (S ₀ ) is temporarily stored in the frame memory (7).

Then, the original image signal (S ₀ ) (shown in FIG. 2 (a)) stored in the frame memory (7) is calculated, for example, by averaging the brightness of each pixel divided by a predetermined number of surrounding pixels. An averaged image that is averaged by performing processing such as sequentially replacing it with values and blurred as shown in FIG.
Get (S ₁ ).

As shown in FIGS. 5 and 6, the field of view (D) of the monitor camera (5) is such that the working value from above the vehicle body (1) is obliquely looked down from above the vehicle body (1). 1) It spreads out in a fan shape in front of the original image signal (S ₀ )
Has a coarser pixel density in a portion farther from the vehicle body (1) (upper side in FIG. 2) than in a portion closer to the vehicle body (1) (lower side in FIG. 2). Therefore, if the averaging is performed uniformly as it is, the image is over-averaged above the lower part of the image, and as a result, the lower part is noisy and the upper part is an unnecessarily blurred image. The processing is performed so that the averaged density becomes coarser.

Next, the process for obtaining the binarized image (S ₂ ) as shown in FIG. 2C is executed from the averaged image (S ₁ ) as shown in FIG. 2B.

That is, the differential value corresponding to the brightness change in the lateral direction of the vehicle body is calculated from the averaged image (S ₁ ), and the relationship between the traveling direction of the vehicle body (1) and the work site situation is calculated from the differential value. When traveling on the unprocessed work site (B) adjacent to the processed work site (C), the sign of the differential value is darker from the left side to the right side in FIG.
By extracting only the negative (-) differential value that becomes clear, and extracting and binarizing the pixel of which the differential value is a predetermined value or more from the extracted negative (-) differential value information, Figure 2 (C)
A binarized image (S ₂ ) having a bright boundary (L) between the uncut area (B) and the already cut area (C) is obtained as shown in FIG.

Therefore, in the image information shown by (L ') in FIG. 2B, the light-> dark change portion where the differential value in the vehicle width direction is positive (+) is removed in advance as noise.

Incidentally, as shown in FIG. 4, the traveling direction of the vehicle body (1) is such that when the vehicle travels back and forth sequentially by advancing each adjacent stroke by 180 degrees at the end of the stroke, the vehicle body (1) and the treated work area ( Since the positional relationship with (C) is reversed right and left for each stroke, the sign of the differential value for discriminating the lightness change is switched between positive and negative at the end of the direction change, that is, as shown in (b) in FIG. When the processed work site (C) is adjacent to the right side of the vehicle body (1) (in the case of a captured image as shown in Fig. 2 (a)), the change in brightness becomes dark → bright.
When only the information of the differential place of (-) is extracted and the processed work site (C) is adjacent to the left side of the vehicle body (1) as shown in (a) of Fig. 4 (Fig. 2 (d)) (In the case of the captured image as shown in FIG. 2), only the information of the positive (+) differential value that changes the lightness from light to dark is extracted so that the unnecessary noise component before binarization is removed in advance. I have to.

Then, the straight line corresponding to the boundary (L) is obtained from the binarized image (S ₂ ) shown in FIG. 2C by using the least square method or the Haugh transform.

In addition, the least squares method is to find a straight line that minimizes the accumulation of distances from each pixel, and the Haugh transform is
The straight line passing through the largest number of pixels among the straight lines passing through each pixel is obtained, and the straight line passing through the binarized image (S ₂ ) shown in FIG.
When a straight line is obtained by processing the data including L and L ′ by the least square method or the Haugh transform, the straight line becomes the straight line corresponding to the boundary (L).

Therefore, the data (L, L in the binarized image (S ₂ ) shown in FIG.
By removing the noise part (L ′) from L ′), the straight line corresponding to the boundary (L) can be obtained, that is, the brightly visible part (L ′) in the unprocessed work site is the original boundary (L). Can be distinguished from. It should be noted that even in the dark-looking part of the processed work site, the bright-looking part of the unprocessed work site
By processing in the same manner as (L ′), it is possible to distinguish from the original boundary (L) (details omitted).

After the boundary line detection process is completed, the steering control operation process is executed as described above.

That is, as shown in FIG. 5, the deviation amount (X) and the deviation angle (θ) of the detection boundary (L) with respect to the target boundary (L ₀ ) along which the vehicle body (1) should follow are calculated.

Then, based on the shift amount (X) calculated by the above calculation, the front and rear wheels (2) and (3) are steered in the same direction, and the vehicle body (1) is parallelized without changing its direction. After moving, based on the deviation angle (θ), the front and rear wheels
Steering operation is performed so that (2) and (3) are relatively opposite to each other, and the direction of the vehicle body (1) with respect to the target boundary (L ₀ ) is corrected to detect the detection boundary (L) and the target boundary (L ₀ ). ₀ ) is controlled so as to coincide with the field of view of the monitor camera (5), that is, the vehicle body (1) is controlled along the target boundary (L ₀ ).

Therefore, even if a discontinuous boundary, a locally curved boundary, or an unclear boundary with uneven grass is detected, it can be accurately corrected as a continuous boundary (L). As described above, the steering control can be performed without performing unnecessary steering operation or shifting the traveling direction to the outside of the original boundary.

In FIG. 1, (9) and (10) are hydraulic cylinders for actually steering the front wheels (2) and (2) and the rear wheels (3) and (3), respectively (11) and (12). Are electromagnetic valves for driving the hydraulic cylinders (9), (10), respectively, and (13) is a motor for operating the gear shift position of the continuously variable transmission (14). Further, (R ₁ ) and (R ₂ ) are for detecting the steering amounts of the front wheels (2) and (2) and the rear wheels (3) and (3), respectively, and feeding them back to the control device (8). Similarly, the potentiometer (R ₃ ) is a potentiometer for detecting the shift position of the transmission device (14) and feeding it back to the control device (8).

By the way, the average image signal (S ₁ ) is obtained by averaging the original image signal (S ₀ ), because a large lightness change or uncut area ( This is because it is difficult to detect a bald area or an uneven area where the grass is small and difficult to distinguish from the already-cut area (C) in B), and the detection error of the boundary (L) is reduced in advance.

Then, in order to perform this averaging process, an optical filter for blurring may be used instead of the arithmetic process, or the camera (5) may be simply used in a so-called out-of-focus state. Further, for example, by using a so-called soft focus filter as an optical filter, by gradually softening the focus from above to below the field of view, the field of view (D) of the camera (5) It is possible to simultaneously correct changes in image density due to distortion.

In addition, when selecting the sign of the differential value based on the positional relationship between the unprocessed work site and the processed work site with respect to the traveling direction of the vehicle body (1), the sign is automatically inverted every 180 ° direction change. In addition to selection, it may be artificially set each time. By the way, in the case of a traveling type that travels inward from the outer periphery of the work site to the outer periphery of the unprocessed work site (B) in sequence, or a front-rear reverse travel form that repeats forward and backward by performing width adjustment for each stroke, (1) and treated work site (C) or boundary (L)
Since the left-right positional relationship with and does not change, it is only necessary to set the sign of the differential value once at the start of work.

It should be noted that reference numerals are added to the claims for convenience of comparison with the drawings, but the present invention is not limited to the structures of the accompanying drawings by the entry.

[Brief description of drawings]

The drawings show an embodiment of a steering control device for an automated guided vehicle according to the present invention, and FIG. 1 is a block diagram showing the configuration of a control system, and FIG. 2 (a), (b), (c), ( D) is an illustration of the image signal,
FIG. 3 is a flow chart showing the operation of the control device, FIG. 4 is an explanatory view of the positional relationship between the vehicle body and the boundary, FIG. 5 is a plan view showing the overall configuration of the lawn mowing work vehicle, and FIG. 6 is a side view thereof. Is. (1) …… Car body, (5) …… Imaging means, (A) …… Boundary detection means, (B) …… Unprocessed work site, (C) …… Processed work site, (L)
……boundary.

Claims (1)

[Claims] 1. A binarization for picking up an image of a work ground condition in a predetermined range in the forward direction of a vehicle body (1) by an image pickup means (5) and binarizing the picked-up image information based on the brightness of each pixel. Means, unprocessed work site (B) based on the information of the binarization means
Boundary detection means (A) for detecting the line segment on the captured image information corresponding to the boundary (L) between the processed work site (C) and the boundary (L) detection result by this boundary detection means (A) Based on the car body
(1) is a steering control device for an automatic traveling work vehicle, comprising a control means for steering so as to automatically travel along the boundary (L), wherein the binarization means is the captured image. Differential value computing means for computing a differential value corresponding to a change in brightness of each pixel of information in a direction along the vehicle body lateral direction, differential value code determining means for determining a positive / negative sign of the differential value, and a positive / negative sign of the differential value. A code selecting means for selecting either one, and an extraction for extracting a pixel of which the differential value change of the code selected by the code selecting means of the discrimination codes by the differential value code determining means is a predetermined value or more. A steering control device for an automated guided vehicle equipped with means.