JP2520104B2 - Boundary detection device for autonomous vehicles - Google Patents

Description

TECHNICAL FIELD The present invention relates to a boundary detection device for an automated work vehicle.

[Conventional technology]

In the above-mentioned automatic traveling work vehicle of this kind, for example, in a work vehicle such as a lawn mowing work vehicle or a reaper harvesting machine, the outer circumference of the planned work range is artificially processed in advance as a processed work site and surrounded by this processed work site. After traveling through the unprocessed work site inside the ground while working along the boundary between the unprocessed work site and the treated work site, one travel from the one end to the other end of the unprocessed work site The control is performed such that the vehicle automatically turns while working in a predetermined range by turning in the direction of the next stroke. In the middle of each stroke, traveling is controlled based on the boundary between the unprocessed work place and the processed work place in the lateral direction of the aircraft and the positional relationship with the aircraft, and at the end of the travel, the boundary in the longitudinal direction of the aircraft. The direction to the next stroke is controlled based on the positional relationship between the vehicle and the aircraft.

To detect the boundary in the machine width direction and the boundary in the machine front-back direction, for example, in a work site where grass is planted, it looks dark because there is uncut grass in the unprocessed work site,
It can be considered that the unprocessed work site and the processed work site are different in brightness from each other, such that light is reflected by the cut grass to make the processed work site look brighter. For example, as disclosed in Japanese Patent Laid-Open No. 52-153531, in order to detect the boundary in the lateral direction of the machine body, in order to detect the boundary part, and in order to detect the boundary in the longitudinal direction of the machine body, It is conceivable that an image pickup means for picking up an image of the boundary portion is separately provided to detect the boundaries of both types. By the way, in this conventional example, the positional relationship with respect to the boundary of the body is determined by detecting the brightness on the screen of the camera as the image pickup means by the pair of photoelectric elements. It should be noted that a camera for picking up an image of the ground away from the boundary is provided separately from a camera for picking up an image of the boundary to obtain the reference brightness.

Further, another conventional technique in which an image pickup means is separately provided for each boundary detection in the machine body width direction and the machine body front-rear direction is described in, for example, Japanese Patent Application Laid-Open No. 52-152026. The image of the boundary point input to the camera is received by a plurality of photoelectric elements arranged in the direction intersecting the boundary direction, and the boundary position is detected by the position of the light receiving element corresponding to the location of the change in brightness. There is.

[Problems to be Solved by the Invention]

In the above-mentioned conventional means, as the image pickup means for picking up the image of the boundary portion, two image pickup means for individually picking up images respectively corresponding to the boundaries in the machine body width direction and the machine body front-rear direction are provided. There is a disadvantage. Further, when performing image processing for boundary detection based on the imaged image information of the two image pickup means, for example, rather than an intermediate portion of each stroke from one end to the other end of the unprocessed work site, that is, a stroke end portion. If both of the above boundary detections are performed while traveling on the front side along the boundary in the machine width direction, processing for boundary detection in the machine front-rear direction is performed in addition to the boundary detection processing time in the machine width direction. It takes time, and there is a disadvantage that the boundary information in the lateral direction of the airframe necessary for performing an appropriate steering operation cannot be quickly obtained. If only the boundary detection in the machine width direction is performed while traveling in the middle part of the above-mentioned travel, there is a problem that the image pickup means for boundary detection in the machine front-back direction is not used at all. .

The present invention has been made in view of the above circumstances,
The purpose is to simplify the device configuration and effectively utilize it, and to cope with the necessity of each boundary information in a state in which erroneous detection of the boundaries of both types in the machine longitudinal direction and the machine width direction is suppressed. SUMMARY OF THE INVENTION It is an object of the present invention to provide a boundary detection device for an automated guided vehicle that can appropriately detect the boundary.

[Means for solving problems]

The feature of the boundary detection device for an automated guided vehicle according to the present invention is that one image pickup means for picking up an image of a work ground condition in a predetermined range on the front side in the traveling direction of the machine body, and a lateral width of the body based on the imaged image information of the image pickup means. Image processing means for detecting the boundary between the unprocessed work site and the processed work site in the direction and the boundary between the unprocessed work site and the processed work site in the machine front-rear direction, and the image processing means is provided. When detecting a boundary in the machine width direction, the captured image information is differentiated along the direction corresponding to the machine width direction based on the brightness, and the machine width based on the differentiated information. In the case of detecting a boundary in a direction and detecting a boundary in the machine front-rear direction, the captured image information is differentiated along a direction corresponding to the machine front-rear direction based on brightness, and the differentiation processing is performed. Based on the information It is configured to detect a boundary in the machine longitudinal direction, and is configured to be switchable between a processing state in which the boundary in the machine lateral width direction is detected and a processing state in which a boundary in the machine longitudinal direction is detected. The action and effect are as follows.

[Action]

That is, based on the image information captured by one image capturing unit, the image processing unit differentially processes the captured image information based on the brightness, and based on the differentially processed information, the boundary in the lateral direction of the machine body. And a boundary in the front-rear direction of the machine body is detected by switching which of the two boundaries is being processed.

In other words, when detecting the boundary in the machine width direction,
The captured image information is differentiated along a direction corresponding to the lateral direction of the machine body, and a boundary in the lateral direction of the machine body is detected based on the differentiated information. Further, when detecting the boundary in the machine front-rear direction, the captured image information is differentiated along the direction corresponding to the machine front-rear direction, and the boundary in the machine front-rear direction is determined based on the differentiated information. To detect.

In order to detect the boundary in the machine width direction, adding a description about differentiating the captured image information along the direction corresponding to the machine width direction, if the differentiation process is performed along the direction corresponding to the machine width direction, The information corresponding to the change in brightness in the front-back direction of the machine is removed, and the information indicating the change in brightness in the lateral direction of the machine is obtained. of course,
The place where the change in brightness is large corresponds to the boundary, and the image processing means detects the boundary based on the information obtained by the differential processing, that is, the information indicating the change in brightness.

In addition, in order to detect the boundary in the front-back direction of the machine body, the description will be made regarding that the captured image information is differentiated along the direction corresponding to the front-back direction of the machine body. Then, the information corresponding to the change in brightness in the lateral direction of the machine is removed, and the information indicating the change in brightness in the front-rear direction of the machine is obtained. Of course, a portion having a large change in brightness corresponds to the boundary, and the image processing means detects the boundary on the basis of the differentiated information, that is, the information indicating the change in brightness.

In short, the image processing means, when detecting the boundary in the machine lateral direction and the boundary in the machine longitudinal direction from the captured image information from one imaging means, when detecting the boundary in the machine lateral direction, By differentiating along the direction corresponding to the machine width direction, it is unnecessary for detecting the boundary in the machine width direction and may cause erroneous detection. The information obtained by removing the information corresponding to the change in the brightness in the front-rear direction of the aircraft will be obtained, and the boundary will be detected from the information. Similarly, when the boundary in the front-rear direction of the aircraft is detected, By differentiating along the direction corresponding to the direction, it is unnecessary to detect the boundary in the front-back direction of the aircraft and may cause erroneous detection. The information obtained by removing the information corresponding to is obtained, and the boundary is detected from the information.

Furthermore, for example, when traveling along each boundary in the machine lateral width direction, which is set so as to reach from one end to the other end of the unprocessed work site, the travel portion on the front side of the travel end is traveled. When it is, it switches to a processing state that only performs boundary detection in the machine width direction so as to quickly obtain boundary information in the machine width direction, and when traveling near the end of the stroke,
Switch to a processing state in which boundary detection in the machine width direction and boundary detection in the machine front-rear direction is performed, or a processing state in which only boundary detection in the machine front-rear direction is performed is switched to appropriately detect that the stroke end has been reached. To do so.

〔The invention's effect〕

Therefore, the boundary in the lateral direction of the machine body and the boundary in the longitudinal direction of the machine body are detected by appropriately switching which of the two boundaries is detected from the imaged image information of one imaging means. While simplifying the device configuration as compared with providing separate image pickup means for boundary detection, at the same time, when only one boundary detection is performed, the other image pickup means for boundary detection is not used at all. The problem can be avoided and the device can be effectively used. Moreover, each boundary is detected based on the information obtained by aligning the captured image information with the direction of each boundary based on the brightness and differentiating along the differentiating direction. Therefore, it is possible to detect both boundaries in a state in which information that is unnecessary for detecting each boundary and that may cause erroneous detection is removed and erroneous detection is suppressed. Front-back direction and In practical use, the boundaries of both types in the lateral direction of the body can be appropriately detected according to the necessity, depending on the running state of which place in the work site the vehicle is traveling. It has come to obtain a boundary detection device for an automatic traveling work vehicle having a great advantage of.

〔Example〕

Hereinafter, an embodiment in which the present invention is applied to a lawn mowing work vehicle as an automatic traveling work vehicle will be described with reference to the drawings.

As shown in FIG. 7 and FIG. 8, the lawn mowing device (2) is moved up and down in the middle part of the machine body (V) configured to be able to steer both front wheels (IF) and rear wheels (1R). A video camera (3) is provided as an imaging means for freely suspending and imaging a work area condition in a predetermined range in front of the traveling direction of the machine body (V), and image information captured by the camera (3) is subjected to image processing described later. Boundary detection device as means (A)
And the boundary between the unprocessed work site (B) and the processed work site (C) in the lateral direction of the vehicle body and the longitudinal direction of the vehicle body.
_{(L 1), (L 2} ) to detect the positional relationship between the body (V) with respect to the boundary by the boundary detection unit (A) (L _1), based on the detection result of (L _2), the travel control device Aircraft (V) by (H)
To perform steering operation for automatically traveling along the boundary (L ₁ ), and to perform steering operation and speed change operation for automatically traveling to the next stroke after the end of one stroke. Configured.

The camera (3) captures an image of a work area in a predetermined range on the front side of the machine body from a diagonally upper side to a lower side of the work frame at a front end portion of a support frame (4) extending upward and forward of the machine body (V). is provided with, in a state where the body (V) is along the boundary (L ₁₎ in its width direction, the boundary (L ₁₎ is to be located in the vertical center of the imaging field of view of the camera (3) There is.

When detecting the boundary (L ₁ ) in the lateral direction of the vehicle body (V), the reference is the lower end of the imaging field of view of the camera (3) as a reference, and a reference that vertically cuts the vertical center of the imaging field of view of the camera (3). A deviation (β) of the detection boundary (L ₁ ) position with respect to the line (L ₀ ) is used as positional deviation information in the lateral direction of the machine body (V), and a reference line (longitudinal direction) that vertically crosses the vertical center of the imaging field of view of the camera (3) ( The inclination (α) of the detection boundary (L ₁ ) with respect to L ₀ ) is detected as the displacement information in the airframe (V) direction.

On the other hand, in the case of detecting the boundary (L ₂ ) in the front-back direction of the machine body (V), as will be described in detail later, in order to store the captured image information after 90-degree coordinate conversion, the image captured by the camera (3). A deviation (β) with respect to a reference line (L ₀ ) that crosses the vertical center of the visual field becomes distance information between the aircraft (V) and the boundary (L ₂ ), and the vertical center of the visual field of the camera (3). The inclination (α) with respect to the reference line (L ₀ ) that crosses the vertical axis is the inclination of the vehicle body (V) with respect to the boundary (L ₂ ) in the longitudinal direction of the vehicle body (V). When the boundary (L ₂ ) in the front-rear direction of the machine body (V) is detected, the value of the detection deviation (β) does not immediately serve as the position information at the end of the stroke, so the detection deviation (β) is detected. Value and the fuselage (V) of the camera (3)
The distance from the lower end of the imaging field of view to the aircraft (V) is set in advance based on the attachment distance and the angle of the imaging direction, and the actual distance between the boundary (L ₂ ) and the aircraft (V) is set based on this set distance information. Will be calculated.

By the way, since the image picked up by the camera (3) is binarized based on the difference in brightness between pixels adjacent to each other, it is necessary to remove the influence of the local change in brightness and average it. As a means therefor, a soft focus filter (8) for blurring a captured image is provided on the front part of the camera (3). Then, the degree of blurring of the filter (8) is such that the imaging field of view of the camera (3) looks down diagonally over the work area in front of and below the aircraft (V). The image spreads like a fan in the front direction, and the pixel density of the portion of the captured image that is distant with respect to the airframe (V) is coarser than that of the portion near the airframe (V). The whole captured image is uniformly averaged by continuously changing so as to become coarser toward.

A strobe device (9) that emits light in synchronization with the image pickup of the camera (3) is provided as means for removing the effect of a large change in brightness such as the shadow of unprocessed grass caused by natural light.
Is provided.

As shown in FIG. 9, the driving force from the engine (E) is transmitted to each of the front wheels (1F) and the rear wheels (1R) via the transmission (4), and the transmission position detecting potentiometer ( The motor (5) is driven so as to travel at a predetermined traveling speed so that the shift position detected by R ₃ ) becomes a predetermined position.

Further, the front wheel (1F) and the rear wheel (1R) are configured to be power-steered by hydraulic cylinders (6F, 6R) respectively, and a steering angle detection potentiometer ( R ₁ ),
The electromagnetic valves (7F) (7R) for operating the hydraulic cylinders (6F) (6R) are driven so that the steering angle detected by (R ₂ ) matches the target steering angle. When correcting the inclination (α) in the longitudinal direction of the vehicle (V) with respect to the boundary (L ₁ ) in the lateral direction of the vehicle, in order to change the direction of the vehicle (V), the front wheels (1F) and the rear wheels ( 1R) is relatively steered in the opposite direction to perform turning steering, and the body (V) with respect to the boundary (L ₁ )
When correcting the deviation (β) in the width direction, the front wheel (1F) is used to move in parallel without changing the direction of the aircraft (V).
By performing parallel steering in which the rear wheels (1R) are steered to face in the same direction, the body (V) is efficiently controlled so as to follow the boundary (L ₁ ).

The configuration and operation of the boundary detection device (A) will be described below.

As shown in FIG. 1, the boundary detection device (A) is an A / D converter that quantizes image information (S ₀ ) captured by the camera (3) with 8-bit resolution and converts it into a digital value. (10), the digitized image information is converted into its coordinates (x,
y) is converted by 90 degrees and sampled and stored as image information having 32 × 32 constituent pixels, the image memory (11), the image information (S ₂ ) stored in the image memory (11) and the A / An image switching unit as an input mode switching unit for switching which image information of the image information (S ₁ ) output from the D conversion unit (10) (however, the number of constituent pixels is processed as 32 × 32) (12) The image signal (F ₁ ) output from the switch (12) is differentiated based on the brightness change in the machine width direction (x coordinate direction) to determine the direction and the magnitude of the brightness change. A differential processing unit (20) for converting into corresponding edge image data (F ₂ ), and a positive / negative value based on the differential sign (positive, negative) of the edge image data (F ₂ ) differentially processed by the differential processing unit (20). comparing the one code side of the edge image data (F ₂₎ and setting the threshold value (Fref) Binarization processing section for binarizing based on the magnitude of the brightness difference, and extracts an image portion brightness change is large by the (30), 2 in the binarization processing unit (30)
A Hough transform unit (40) that approximates the binarized image data (F ₃ ) that has been binarized as a continuous straight line and specifies the boundaries (L ₁ ) and (L ₂ ) on the image. ), A boundary parameter calculation unit (50) for converting the detection boundaries (L ₁ ) and (L ₂ ) into the following equation (i) approximated as a straight line representing the positional relationship between the vehicle and the vehicle (V), A control unit (60) that controls the operation of the processing unit, and an ITV controller (1 that controls the light emission of the strobe (9) and the imaging of the camera (3).
3), the differentiation processing section (20) and the binarization processing section (30)
And a CRT controller (14) for controlling the operation of a monitor television (15) displaying a processed image by the CRT.

y = ax + b (i) where a: boundaries (L ₁ ) and (L ₂ ) with respect to the reference line (L ₀ ) in the image coordinate system
, B: Boundary (L ₁ ), (L ₂ ) with respect to the reference line (L ₀ ) in the image coordinate system
X is the coordinate value of the pixel in the machine width direction (V), and y is the coordinate value of the pixel in the front-rear direction of the machine body (V).

The overall operation will be described below based on the flowchart shown in FIG.

That is, the entire image is averaged by the blurring filter (8) to be captured, and the captured image signal (S ₀ ) is A / D.
It is converted into a digitized original image signal (F ₁ ) in a rectangular coordinate system (x, y) composed of 32 × 32 pixels per screen.

Next, it is checked whether the boundary (L ₁ ) in the lateral direction of the aircraft on the way of travel or the boundary (L ₂ ) in the longitudinal direction of the aircraft, which is the end of the travel, is detected. L ₂ ) is detected, the address coordinates of the image memory (11) are
When the image information (S ₂ ) stored by converting the coordinates of the image signal (S ₁ ) by 90 ° is used by converting the image signal (S ₁ ) and giving the boundary (L ₁ ) on the way, Image signal (S ₀ )
The image data (S ₁ ) having the same coordinate system as the above is used to perform differential processing on the image data by the differential processing section (20), and only the one having a positive or negative differential code direction is binarized. ) Is binarized.

As described below, the selection of whether the differential sign is positive or negative is binarized, as described below, the unprocessed work site (B) exists on either the left or right side of the machine body (V). It is set based on the crab. That is, for example,
As an example of the case where the working method is turning around in which each side of the work area is turned 90 degrees in the counterclockwise direction while traveling inward, the machine body (V) is processed to the right. It is in a state of traveling adjacent to the completed work site (C). In other words, the change in lightness causes a large positive change from dark to light at the boundary (L ₁ ) when scanning from the left side to the right side on the image, and when the change between adjacent pixels is differentiated, the differential value The one with a positive sign (+) corresponds to a certain directionality of the boundary (L ₁ ) when the brightness change is seen from the unprocessed work site (B) side to the processed work site (C) side, Negative (-) is regarded as noise and can be removed.

Next, this differentially processed image data is binarized, and further, based on the following equation (ii), the Cartesian coordinate system (x, y) is transformed into the polar coordinate system (ρ, θ) and Hough transformed, that is, The frequency of the binarized image data having the same polar coordinate (ρ, θ) at each polar coordinate (ρ, θ) is counted as a two-dimensional histogram to obtain the maximum value (Dmax).

ρ = xcos θ + ysin θ (ii) where 0 ° ≦ θ ≦ 180 ° x: coordinate value of the lateral (W) pixel of the vehicle (V) y: coordinate value of the forward / backward pixel of the vehicle (V)

Then, from the value of the polar coordinate (ρ, θ) of the frequency (D) that is the maximum value (Dmax), the equation (i) which is the boundary parameter corresponding to one piece of boundary information is determined, and the equation (i) is determined. From the inclination (a) and the deviation (b) as the coefficient of, the actual inclination (α) of the vehicle (V) with respect to the boundary (L ₁ ) and the deviation (β) in the lateral direction in the traveling control device (H) are obtained. Converting the steering wheel, the front wheel (1F) and rear wheel (1R) are steered so that the inclination (α) and deviation (β) are zero respectively, and the aircraft (V) follows the boundary (L ₁ ). It controls to run automatically.

By the way, when detecting the boundary (L ₁ ) in the lateral direction of the aircraft, the image pickup information from the camera (3) is directly input to the differential processing unit (20) for processing, and the boundary (L ₂ ) in the longitudinal direction of the aircraft is detected. When detecting, the vertical direction (y) of the original image information
In the left-right direction (x), in the left-right direction (x) in the up-down direction (y)
To the image memory (11) and read them in the original coordinate system (x, y), thereby converting the coordinates (x, y) of the imaging information by 90 degrees and performing the differential processing. Since it is input to the unit (20) and processed, the boundary (L ₁ ) and (L ₂ ) detection in different directions with respect to the aircraft (V) can be used without changing the processing units after the differentiation processing unit (20). You can do it.

When detecting the boundary (L ₂ ) in the front-back direction of the machine, if the maximum value (Dmax) of the frequencies matching the same polar coordinates (ρ, θ) is not greater than or equal to the set threshold (Dref), the machine (V)
It can be judged that has not reached the vicinity of the boundary (L ₂ ) at the end of the stroke. On the other hand, when the maximum value (Dmax) is equal to or larger than the set threshold value (Dref), the slope (a) of the equation (i)
Based on the deviation (b) and the distance between the lower end of the imaging field of view and the front end of the vehicle body (V), the distance from the vehicle body (V) to the stroke end boundary (L ₂ ) can be obtained, and the turning point can be accurately determined.

The operation of each unit will be described below with reference to the drawings shown in FIGS.

First, the configuration and operation of the control section (60) will be described in a third section.
A description will be given based on the block diagram shown in the figure.

The control unit (60) is designed so that the entire operation is performed in synchronization by the clock signal (CL) generated by the clock generation unit (61a), and the CPU 1 as a control processor that controls the operation of each unit. (62a) and CPU2 (62b) as a numerical operation processor are provided.
It is configured for high speed processing. Then, the calculation results of each control signal and data are transmitted / received to / from each processing unit via the bus buffer (63). Similarly,
Data is exchanged with the travel control device (H) by the SIO port (64a) for serial interface and the SI.
Boundary as done online via the communication interface (64b) connected to the O port (64a) and offline via the switch interface (64c) for offline and the PIO port (64d) for parallel interface. The detector (A) can also be activated.

In addition, CPU1 (62a) and CPU2 (6
The data transfer between the 2b) and the data transfer via the path buffer (63) and the control of the operation are performed by the input / output controller (65) and at the same time as each of the ports (64).
The control of data transmission / reception from a) and (64d) is also performed through the CTC controller (66) by the control signal from the input / output controller (65). In FIG. 3, (61b) is a clock signal generator that generates clock signals for operating the SIO port (64a) and CTC controller (66), and ROM (67) is the operating program of the boundary detection device (A). A stored memory, a RAM (68) is a memory that temporarily stores each data and is used as a buffer for calculation data, etc., and a LUT (69) is the image information used by the Hough transform unit (40) in orthogonal coordinates (x, y) to Hough transform polar coordinates (ρ,
The calculation data for high-speed conversion into θ), that is, the data for obtaining a plurality of polar coordinates (ρ, θ) representing straight lines passing through the rectangular coordinates (x, y) and having different inclinations are stored in advance as a table. It is a memory.

Next, the operation of the differentiation processing section (20) will be described based on the block diagram shown in FIG. 4 and the explanatory diagrams shown in FIGS. 5 (a) and 5 (b).

The video signal (S ₀ ) which is the original image information from the camera (3) is supplied to the video amplifier (10a) and the A / D converter (10
b) directly via the A / D converter (10), or
Image information stored in the image memory (11) after 90-degree coordinate conversion is passed through the input form switching unit (12) to the 32
Digital image data (F ₁ ) represented by × 32 pixels, each pixel having an 8-bit length, is input to the differentiation processing section (20).

Then, from the left to the right with the upper left of the imaging field of view of the camera (3) as the origin, the rank 1 raster (x for one y coordinate is
Import image data (F ₁ ) for 32 pixels of coordinates 3
When the image data (F ₁ ) for the raster (i−2, i−1, i) is input, as shown in FIGS. 5 (a) and 5 (b), the positions on the left and right sides of the pixel of interest are displayed. 2 pixels and 4 pixels located on both the upper and lower sides of the left and right pixels, a total of 6 pixels, the brightness difference between corresponding pixels in the left-right direction (x-axis direction) is calculated, and the differential value is obtained. A process is performed, and this differentiation process is repeated for each of 32 rasters (y-coordinates of 32 pixels) to calculate the lightness change of all image information of 32 × 32 pixels and its directionality. That is, differential processing is performed using a mask having directionality in the left-right direction (x-axis direction) to extract edges of the image that change from dark to light.

That is, the control unit (60) causes the bus buffer (21)
An image signal (S ₁ ) that does not undergo the 90-degree coordinate conversion in synchronization with the timing of an operation control signal from an input / output controller (22) whose operation is controlled based on information such as a control signal input via Alternatively, the brightness information (D _j-2 ) of one pixel (j-2) in one raster (i-2) of the digitized original image data (F ₁ ) which is the image signal (S ₂ ) converted by 90 degree coordinates
Is input to the first register (REG ₁ ). In the next step, the contents of the first register _{(REG 1) (D j-} 2) is transferred to the second register (REG _2), the next pixel in the first register (REG ₁₎ at the same time (j-1 ) Brightness information (D _j-1 ) is input. In the next step, the first register (REG ₁ )
The contents of (D _j-1) is transferred to the second register (REG _2), the second register (REG ₂₎ of the transferred current pixel (j) with respect to two pixels before the pixel (j -2) brightness information (D _j-2 ) and the current pixel (j) are subtracted by a subtractor (23)
Is calculated by

Then, the result is sequentially transferred through the data buffer (24a) to the first raster memory (25a) that stores the brightness difference data for one raster (i-2), and the brightness difference data for 32 pixels in the x coordinate direction is sequentially transferred. However, the next raster (i-1) output from the subtractor (23) has the same x coordinate as the pixel (j,
The data between i-2) and (j, i-1) is added to the first adder (ALU ₁ )
And the result is transferred to the second raster memory (25b) via the data buffer (24b). Furthermore,
Simultaneously with the transfer of the brightness difference data to the second raster memory (25b), the transfer data, that is, the raster (i-2) two before and the raster (i-1) one before the current raster (i). The pixels (j, i-2) having the same x coordinate of
Brightness difference data between (j, i-1) and the first adder (ALU ₁ )
Pixel (j, i-1) whose x coordinate of the previous raster (i-1) and the current raster (i) added by
The brightness difference data between (j, i) is added by the second adder (ALU ₂ ) and stored in the third raster memory (25c).

Therefore, each raster memory (25a), (25b), (25
In c), the brightness difference data with the following contents will be stored respectively.

First raster memory ← D (x, i) Second raster memory ← D (x, i-1) + D (x, i) Third raster memory ← D (x, i-2) + D (x, i-1) ) + D (x, i−1) + D (x, i) (where x = j, j = 1 to 32) The data stored in the third raster memory (25c) is the final operation. Since it is data and the data of one raster before is added twice, only the data of the differential code (either positive or negative) preset by the differential code selection unit (26) is selected and stored. To be done. And
It is converted by the multiplier (27) into the data of the level set by the gain setting switch (28), and is input via the data buffer (24c) every time the digitized original image data (F ₁ ) for one pixel is input. It is converted to 8-bit / 1-pixel edge image data (F ₂ ) and output to the next binarization processing unit (30).

The processing described above is sequentially and repeatedly performed for each pixel, so-called pipeline processing is performed, and image data is differentiated in real time. In FIG. 4, (24d) is the edge image data (F
₂ ) is a data buffer for substituting the maximum value (FF in hexadecimal notation) that can be expressed in 8-bit length when overflow occurs as a result of the operation.

Next, based on the block diagram shown in FIG. 6, a binarization processing unit for binarizing the image information converted into the 8-bit / pixel edge image data (F ₂ ) by the differentiation processing unit (20). The configurations and operations of the (30), the Hough transform section (40), and the boundary parameter computing section (50) will be described.

That is, like the differential processing section (20), the entire operation is synchronized with the operation of the differential processing section (20) by the control signal from the control section (60) input via the bus buffer (41). The entire operation is controlled by the input / output controller (42).

The edge image data (F ₂ ) is transferred to the comparator (3) of the binarization processing unit (30) via the data buffer (43a).
1) is compared with the set threshold value (Fref) that is input via the buffer (32), and there are 2
Valued. This binary image data (F ₃ ) is input to the comparator (51) of the boundary parameter calculation section (50) via the data buffer (43b) connected to the data bus (BUS ₃ ) and Frequency counter (44)
The frequency (D) counted in is stored in the data buffer (43
The maximum value register (50) of the boundary parameter calculation unit (50) that specifies the boundary by adding to the contents of the same address position of the frequency memory (45) via d) and detecting the maximum value (Dmax). 52a) and is compared with the value stored in the maximum value register (52a) by the comparator (51), and the larger frequency is stored again as a new maximum value (Dmax). This maximum value (Dma
The address of the frequency memory (45) having the frequency (D) of x) is stored in the maximum value address register (52b) at the same time, and after the Hough conversion of all pixels (32 × 32) is completed, the maximum value (Dmax ), The coefficient (a, b) of the equation (i) is obtained from the coordinate (ρ, θ) via the equation (ii). Further, in FIG. 6, (33b) is an image memory for storing the digitized image data (F ₁ ), (33c) is an image memory for storing the edge image data (F ₂ ) and is a video RA.
An image of each processing state can be displayed on the monitor television (15) via the M (34) and the CRT controller (14).

Then, the obtained coefficient (a, b) of the equation (i) is sent to the control unit (60) via the input / output controller (42) and the bus buffer (41), and the boundary (L ₁ ) The data is used for determining the presence or absence of (L ₂ ) and the subsequent processing.

[Another embodiment]

In the above embodiment, the boundary (L ₂ ) in the longitudinal direction of the aircraft (V)
Based on whether or not the detection is performed, the coordinates of the input image information to the differential processing section (20) are switched between those converted by 90 degrees and those not converted by 90 degrees. Each time the image information is processed, the coordinates of the input image are converted by 90 degrees to detect the boundary (L ₁ ) in the width direction of the machine body (V) and the boundary (L ₂ ) in the front and rear direction of the machine body (V). May be alternately repeated. In that case, if the coordinate-converted image and the non-converted image are alternately input for each field of the image captured by the camera (3), necessary information can be simultaneously obtained for each frame, and one frame worth of information can be obtained at the same time. Image information for detecting boundaries (L ₁ ) and (L ₂ ) in different directions can be obtained during the imaging process.

Alternatively, the vehicle body (V) is provided with a means for detecting the traveling distance information, and based on the detected traveling distance information and the planned traveling distance information of one stroke, the vehicle body (V) is detected depending on whether or not it has reached the vicinity of the travel end. ) boundary in the longitudinal direction (L ₂₎ may be switched to the detection process.

Further, in the above embodiment, in the Hough transform processing equation (ii), the angle change is processed within the range of 0 ° ≦ θ ≦ 180 °, but the boundaries (L ₁ ), (L ₂ ) and the aircraft (V ₂ ) And the positional relationship
Since the directions of (L ₁ ) and (L ₂ ) do not change significantly from the traveling direction of the aircraft (V), the angle range for Hough conversion is limited to, for example, a range narrower than 0 degrees to 180 degrees. By doing so, the processing speed can be improved and the influence of noise can be effectively removed by improving the accuracy or processing only a narrow specific range.

Further, in the above-mentioned embodiment, the captured image information is sampled into 32 × 32 pixels, but it may be sampled more finely or coarsely if necessary.

In addition, in order to obtain the positional relationship between the boundaries (L ₁ ) and (L ₂ ) and the body (V), the reference line (L ₀ ) that vertically crosses the imaging field of view illustrated in the above embodiment is used as a reference. , The lower end of the imaging field of view,
The upper end, the center, or the coordinate origin position of the image information may be used as a reference.

[Brief description of drawings]

The drawings show an embodiment of a boundary detection device for an automated guided vehicle according to the present invention, FIG. 1 is a block diagram showing the configuration of the boundary detection device, FIG. 2 is a flowchart showing the overall operation thereof, and FIG. FIG. 4 is a block diagram showing the configuration of the control unit, FIG. 4 is a block diagram showing the configuration of the differential processing unit, FIG.
(B) is an explanatory view of the processing operation, FIG. 6 is a block diagram showing the configuration of the Hough conversion unit, FIG. 7 is a schematic plan view of a lawn mowing work vehicle, FIG. 8 is a side view thereof, and FIG. 9 is lawn mowing. It is a block diagram which shows the whole structure of the control system of a working vehicle. (3) ... Imaging means, (A) ... Image processing means, (B)
…… Unprocessed work site, (C) …… Processed work site, (L ₁ ),
(L ₂ ) …… Boundary.

Claims (1)

(57) [Claims] 1. An image pickup means (3) for picking up an image of a work ground condition in a predetermined range on the front side in the traveling direction of the machine body, and unprocessed in the lateral direction of the machine body based on imaged image information of the image pickup means (3). The boundary (L ₁ ) between the work site (B) and the processed work site (C), and the boundary (L ₂ ) between the unprocessed work site (B) and the processed work site (C) in the machine front-rear direction. The image processing means (A) for detecting is provided, and when the image processing means (A) detects the boundary (L ₁ ) in the machine lateral direction,
The captured image information is differentiated along the direction corresponding to the lateral width direction of the body based on the brightness, and the boundary (L ₁ ) in the lateral width direction of the body is detected based on the differentiated information. When detecting the boundary (L ₂ ) in the front-back direction,
The imaged image information is differentiated along the direction corresponding to the front-rear direction of the body based on the brightness, and the boundary (L ₂ ) in the front-rear direction of the body is detected based on the differentiated information. In addition, the automatic traveling work is configured to be switchable between a processing state in which the boundary (L ₁ ) in the lateral direction of the machine body is detected and a processing state in which the boundary (L ₂ ) in the longitudinal direction of the machine body is detected. Border detection device for cars.