JPH11175150A - Stop position deviation amount detecting device for moving body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a stop position deviation amount detecting device for moving body, on a floor surface where an arbitrary pattern is formed, which can correct position of a moving body with high precision while preventing the influence of disturbances. SOLUTION: A moving body which moves on a floor surface 9 such as a punching floor where arbitrary patterns and specific recognition marks are formed, is stopped at a specified position, and an image of the floor surface 9 is picked-up to obtain position data. Deviations in the position and posture are obtained by using many pieces of position data to enable high-precision measurement, and even when part of it can not be recognized under the influence of disturbing light, the processing can be performed without any problem.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is mounted on a moving body used for transporting luggage in a clean room or the like in which a flooring material such as a punching material or a grating material having an arbitrary pattern formed on its surface is laid. The present invention relates to an apparatus for detecting a displacement of a stop position of a moving body, which detects a displacement between a stop position and a reference stop position.

[0002]

2. Description of the Related Art In a clean room or the like of a semiconductor device factory, an unmanned transport vehicle that transports a load such as a wafer between devices and a device, a device and a stocker, and transfers the load by a transfer arm or the like is used. In this type of automatic guided vehicle, the transfer arm and the like perform work according to the teaching data taught in a state where the automatic guided vehicle is stopped at a predetermined work position (teaching position) in advance. Therefore, in order to accurately perform the work using the transfer arm or the like, the automatic guided vehicle is accurately stopped at the teaching position, or the teaching data itself is corrected by a deviation of the stop position from the teaching position. For example, it is essential to accurately correct the position of the automatic guided vehicle. By the way, in a clean room or the like as described above, a perforated floor material having through holes arranged in a predetermined pattern on the entire surface, such as a punching material or a grating material, is often used as a floor material. Accordingly, the present applicant has already filed a patent application for a device for correcting the position of the automatic guided vehicle using the pattern formed on the floor surface (Japanese Patent Application No. 09-103301). FIG. 1 shows an automatic guided vehicle position correcting apparatus and method according to the above-mentioned prior application.
This will be described with reference to FIGS.

[0003] The automatic guided vehicle A with an arm according to the above-mentioned prior application.
At 0, as shown in FIG. 11, an arm 52 having a hand 53 at the tip is mounted on the upper part of the automatic guided vehicle 51. An imaging unit 54 having a ring-shaped illumination device 55 is fixedly installed near the center of the lower part of the automatic guided vehicle 51 (a position hardly affected by lighting or the like) so as to face the floor surface 59. The image processing unit 5
6, a storage unit 57, and a teaching data correction unit 58. Further, a punching floor perforated in a predetermined pattern as shown in FIG. 13 is laid on the floor 59 as is usually performed in a semiconductor clean room or the like. As shown in FIG. 13, a plurality of orthogonal horizontal frames 62 and vertical frames 63 are formed in the punching floor 59 in parallel with perforations (punching holes) 61. Further, a recognition mark 60 is provided on the punching floor 59 near the stop position of the automatic guided vehicle 51 and near an intersection 64 between any of the horizontal frame 62 and the vertical frame 63. The unmanned guided vehicle 5 is placed in a predetermined working position in advance on the arm 52.
The operation of transferring a wafer (not shown) from the unmanned carrier 51 to a work table (not shown) while the machine 1 is stopped is taught. At the time of the actual transfer operation, the arm 52 performs the operation according to the teaching data. The image capturing section 54 captures an image of the punching floor 59 including the recognition mark 60 when the operation of the arm 52 is taught and when the arm 52 is stopped at a predetermined work position during the transfer operation. The image processing unit 56
Then, by performing image processing on the captured image captured from the imaging unit 54 after the imaging by the imaging unit 54,
Straight lines H and V are extracted from the horizontal frame 62 and the vertical frame 63 forming the intersection 64 where the recognition mark 60 is provided, and the straight line H and the straight line V in the local coordinate system on the captured image are extracted. The position coordinates of the intersection Q and the rotation angle (hereinafter, referred to as
Location information). The position information obtained when teaching the operation of the arm 52 is stored in the storage unit 57. In the teaching data correction unit 58, after the automatic guided vehicle 51 stops at a predetermined work position during the transfer operation and is processed by the image processing unit 56, the position information stored in the storage unit 57 in advance is stored. And this time the image processing unit 56
Difference from the position information obtained by the
The teaching data of the arm 52 is corrected based on the shift amount of the stop position of No. 1.

Hereinafter, the position correcting operation of the automatic guided vehicle A0 with an arm will be specifically described with reference to a flowchart shown in FIG. First, prior to the actual transfer operation, a wafer (not shown) is transferred from the automatic guided vehicle 51 to the worktable (not shown) with respect to the arm 52 while the automatic guided vehicle 51 is stopped at a predetermined work position. (Shown) is taught. At this time, the image of the punching floor 59 is captured by the imaging unit 54 (step S51). Subsequently, the image processing unit 56 performs image processing on the captured image captured from the image capturing unit 54, thereby obtaining the image shown in FIG.
As shown in FIG. 3, straight lines H and V are respectively extracted from the horizontal and vertical frames forming the intersection where the recognition mark 60 is provided, and the intersections Q of the imaging region R in the local coordinate system CS are extracted. The position coordinates (X ₀ , Y ₀ ) and the straight line H
And the rotation angle θ ₀ between the local coordinate system CS and the X axis are obtained (step S52). The position information thus obtained is stored in the storage unit 57 (step S5).
3). Next, during the actual moving operation, when the automatic guided vehicle 51 stops at a predetermined operation position, the image of the punching floor 59 is captured by the imaging unit 54 (Step S).
54). Subsequently, the image processing unit 56 performs image processing on the captured image captured from the image capturing unit 54 in the same manner as at the time of the teaching, and thereby, as shown in FIG. Are extracted from the horizontal frame and the vertical frame, respectively, forming the position coordinates (X ₁ , Y ₁ ) of the intersection Q in the local coordinate system CS of the imaging region R, and the straight line H and the straight line H. Local coordinate system CS
Eggplant rotation angle theta ₁ with the X-axis of is obtained.

Subsequently, in the teaching data correction unit 58,
Obtained in step S52 at the time of the above teaching and stored in the storage unit 57
The stored position information (position coordinates (X₀, Y₀), Rotation angle θ
₀) And the position information (position
Mark (X₁, Y₁), Rotation angle θ ₁), That is, unmanned conveyance
The amount of deviation of the stop position of the car 51 between teaching and working is calculated.
The teaching data of the arm 52 is corrected based on the
Is performed (step S56). The arm 52 is
Transfer operation in accordance with the teaching data corrected in step S56.
(Step S57). Thereafter, the movement of the automatic guided vehicle 51
Steps S54 to S57 are repeated each time. Less than
The top is the position correction device and the automatic guided vehicle
And its method. In addition, the captured image at the time of the above teaching and the above
When calculating the deviation from the captured image during work,
As in the application, the position of the intersection Q of the straight line H and the straight line V
Other than using the mark and the rotation angle, for example, the image of the floor above
Is stored as a template image.
The matching process between the captured image and each of the above captured images.
A method for calculating the deviation between the position and orientation of each captured image is also conceivable.
You.

[0006]

However, in the position correction method according to the above-mentioned prior application, a pattern (a hole 6
There is a problem that when the reference lines corresponding to the straight lines H and V cannot be defined, as in the case where 1) is not arranged in a grid pattern, it cannot be handled. Further, even when the patterns on the floor surface are arranged in a grid pattern, the punching holes 6
The position 1 includes an error at the time of manufacture, and therefore, the setting accuracy of the straight lines H and V depends on the error. Therefore, the position correction accuracy is low. There is also a problem that there is a limit in improving the accuracy because only the image is used. Furthermore, if the punching hole 61 of a notable part cannot be recognized due to disturbance caused by a lighting device installed under the floor, the position correction operation itself becomes impossible. Further, in the method based on the matching process using the template image, as shown in FIG. 15, the captured image A having the same orientation (rotation angle) as the template image can be easily detected. For captured images B having different postures (rotation angles), the matching process is repeated while rotating the template image little by little, or a large number of template images with the rotation angle slightly changed are stored in a memory in advance. In addition, it was necessary to repeat the matching process while replacing the template image. Therefore, there is a problem that an enormous amount of time is required for the matching process, and a large storage area for storing a large number of template images is required. The present invention has been made in view of the above circumstances, and an object of the present invention is to prevent the influence of disturbance on a floor surface on which an arbitrary pattern is formed, and to accurately correct the position of a moving body. It is an object of the present invention to provide a movable body position correcting apparatus that can perform the above. It is still another object of the present invention to provide a moving body stop position shift amount detection device capable of performing processing in a short time by using a small storage area when matching processing is used in the position correction.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above-mentioned object, the present invention provides a moving object which moves on a floor on which an arbitrary pattern and a predetermined recognition mark are formed, stops at a predetermined position, and The difference between the captured image at the time of the moving operation in which the image of the surface is captured and the reference captured image in which the moving body is stopped at a predetermined reference stop position in advance and the image of the floor surface is captured, A moving body stop position shift amount detecting device that determines a shift amount of the stop position of the moving body based on the shift based on a comparison based on the recognition mark of an arbitrary pattern. A reference position data acquisition unit for acquiring position data of a plurality of feature parts constituting an arbitrary pattern and the recognition mark on the reference captured image;
Moving work position data acquisition means for acquiring position data of the plurality of feature parts and the recognition mark on the moving work captured image; and providing the plurality of feature parts with the recognition mark on each of the captured images. A pair of coding means for performing common coding based on a mark, and a pair of the plurality of feature parts between the respective captured images based on codes of the plurality of feature parts obtained by the coding means. 1 and a position obtained by the reference position data acquiring means and the moving work position data acquiring means for each set of the plurality of form parts associated by the associating means. A shift calculating means for determining a difference between data and calculating a shift between the captured images based on a difference between the position data in each of the sets. It is configured as a location.

Further, a predetermined number of position data is selected from the position data of the form part obtained by the reference position data obtaining means and the moving work position data obtaining means, and a form corresponding to the selected position data is selected. If only the parts are associated by the association means, it is possible to increase the processing speed and reduce the memory capacity while maintaining a certain level of accuracy. Alternatively, the moving work position data obtaining means may include a general position data obtaining means for obtaining general position data of the plurality of feature parts, and a predetermined number of general position data obtained from the general position data obtaining means. General position data selecting means for selecting the general position data, and detailed position data obtaining means for obtaining detailed position data for the form part corresponding to the general position data selected by the general position data selecting means. With such a configuration, it is possible to further speed up the processing and reduce the memory capacity. In the case where the arbitrary pattern is constituted by a regular arrangement of the plurality of feature parts, the encoding means is configured to determine the plurality of feature parts based on the recognition mark on each of the captured images. By performing the coding based on the regularity of the above arrangement, easy and reliable coding can be performed. On the other hand, for example, when the arbitrary pattern is formed by an irregular arrangement of the plurality of feature parts, the above-described coding method is not used, and thus the recognition mark is used as a starting point on the reference captured image. The coding of each of the features on the reference captured image is performed based on a vector obtained by sequentially connecting the plurality of features, and the reference imaging is performed in accordance with the vector with the recognition mark as a starting point on the captured image during the moving operation. This can be achieved by performing common coding with the image.

In the reference position data acquiring means and the moving work position data acquiring means, a matching process is performed between a predetermined template image and the plurality of feature parts and the image of the recognition mark, so that the plurality of feature parts and the recognition mark image are matched. By recognizing the recognition mark and acquiring the position data on each of the captured images, it is possible to acquire position data with high accuracy. At this time, when a plurality of feature parts constituting the arbitrary pattern are formed in the same circle, respectively, the template image and the individual images are used by using a single image of the circular feature part as the predetermined template image. Regardless of the deviation of the attitude angle from the captured image, the matching processing can be performed at high speed and with a small memory usage.
In the reference position data acquiring means and the moving work position data acquiring means, a portion having a predetermined shape characteristic amount is determined from a binarized image obtained by binarizing each captured image with a predetermined threshold value. If the plurality of feature parts and the recognition marks are recognized by extraction and the position data on each of the captured images is acquired, the processing speed can be increased although the accuracy is not as good as the matching processing. It is possible. If one or more of the features constituting the arbitrary pattern can be specified as the recognition mark, the processing by the coding means can be performed based on the features. It is. Note that the arbitrary pattern and the predetermined recognition mark are not necessarily formed on the floor, but may be formed on a plane on which the moving body moves, for example, on a work table.

[0010]

In the moving object stop position shift amount detecting apparatus according to the present invention, the floor surface including an image of an arbitrary pattern and a predetermined recognition mark is preliminarily stopped in a state where the moving object is stopped at a predetermined reference stop position. (Reference image) is captured. The reference captured image is subjected to predetermined image processing by reference position data acquisition means, and position data on the reference captured image of the plurality of feature parts constituting the arbitrary pattern and the recognition mark is calculated. . At this time, the reference position data acquiring means recognizes the plurality of feature parts and the recognition mark by matching processing of a predetermined template image with the plurality of feature parts and the image of the recognition mark, and acquires position data. By doing so, highly accurate position data can be obtained. At this time, when a plurality of feature parts constituting the arbitrary pattern are formed in the same circle, respectively, the template image and the individual images are used by using a single image of the circular feature part as the predetermined template image. Regardless of the deviation of the attitude angle from the captured image, the matching processing can be performed at high speed and with a small memory usage. The reference position data acquisition means extracts a portion having a predetermined shape feature amount from a binarized image obtained by binarizing each captured image with a predetermined threshold, thereby obtaining the plurality of feature parts. If the recognition mark is recognized and the position data is acquired, the processing speed can be increased, although the accuracy is lower than the matching processing.

Subsequently, the coding means performs coding based on the recognition mark for each of the features on the reference captured image. On the other hand, at the time of actual moving work, an image of the floor surface (image taken at the time of moving work) including an image of the arbitrary pattern and the predetermined recognition mark is taken with the moving body stopped. The moving work picked-up image is subjected to predetermined image processing by the moving work position data acquiring means in the same manner as the reference picked-up image, and the positions of the plurality of features and the recognition marks on the reference picked-up image are obtained. Data is calculated. Then, the coding unit performs common coding with the reference captured image on the basis of the recognition mark for each of the features on the captured image during the moving operation. Here, a predetermined number of position data are selected from the position data of the form part obtained by the moving work position data acquisition means, and only the form part corresponding to the selected position data is handled by the association means. With such a configuration, it is possible to increase the processing speed and reduce the memory capacity while maintaining a certain level of accuracy. Alternatively, the processing by the position data obtaining means during moving work may be performed by first obtaining the approximate position data of the plurality of form parts by the approximate position data obtaining means, and the predetermined number of the approximate position data is selected by the approximate position data selecting means. If the position data is selected and the detailed position data acquisition means acquires detailed position data only for the form part corresponding to the approximate position data selected by the approximate position data selection means, further It is possible to increase the processing speed and reduce the memory capacity.

In the case where the arbitrary pattern is constituted by a regular arrangement of the plurality of feature parts, the encoding means sets the plurality of feature parts on the basis of the recognition mark on each of the captured images. Is performed based on the regularity of the above arrangement of the form part. This allows easy and reliable coding. On the other hand, for example, when the arbitrary pattern is formed by an irregular arrangement of the plurality of feature parts, the above-described coding method is not used, and thus the recognition mark is used as a starting point on the reference captured image. The coding of each of the features on the reference captured image is performed based on a vector obtained by sequentially connecting the plurality of features, and the reference imaging is performed in accordance with the vector with the recognition mark as a starting point on the captured image during the moving operation. This can be achieved by performing common coding with the image. Subsequently, the associating means makes a one-to-one correspondence between the plurality of feature parts between the captured images based on the code given by the coding means.
Are made, and a set of form parts is created. At this time, since correspondence is performed only with respect to the form part for which the position data is obtained in both captured images, even if some form parts cannot be recognized during the moving work due to the influence of disturbance light or the like, processing can be performed without any problem. . Subsequently, the difference between the position data obtained by the reference position data obtaining means and the position data obtained by the moving work position data obtaining means for each set of the form parts associated by the association means is obtained by the deviation calculating means. Then, based on the difference between the position data in each of the sets, a deviation between the captured images is obtained by using, for example, a least square method. Then, a deviation amount of the stop position of the moving body is detected based on the obtained deviation between the captured images. As described above, high-precision measurement can be performed by determining the position / posture deviation using a large number of sets of position data. Further, since the individual position data of each of the feature parts is used, even if the position of the feature part includes an error at the time of manufacture, the measurement accuracy is not affected.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments and examples of the present invention will be described below with reference to the accompanying drawings to facilitate understanding of the present invention. The following embodiments and examples are mere examples embodying the present invention, and do not limit the technical scope of the present invention. Here, FIG. 1 is a schematic diagram showing a schematic configuration of an automatic guided vehicle A1 equipped with a stop position shift amount detecting device according to an embodiment of the present invention, and FIG. 2 shows a processing procedure by the stop position shift amount detecting device. Flow chart, FIG.
FIG. 4 is a diagram showing an example of a captured image at the time of teaching and encoding, FIG. 4 is a diagram showing a template image used for template matching processing, FIG.
FIG. 7 is a diagram showing an example of a picked-up image and a code at the time of work, FIG. 7 is a diagram showing an example of position data of the associated punched holes, and FIG. 8 is a flowchart showing a processing procedure of step S5 according to the second embodiment FIG. 9 is an explanatory view of an associating process when the arrangement of punching holes according to the third embodiment has no regularity, and FIG. 10 is an explanatory view of a case where a specific punching hole according to the fourth embodiment is used as a recognition mark. . In the present embodiment, similarly to the above-described conventional technology, in a semiconductor clean room, between semiconductor manufacturing apparatuses or between semiconductor manufacturing apparatuses.
The automatic guided vehicle A1 with an arm for transferring and transferring wafers between stockers will be described as an example. In the automatic guided vehicle A1 with an arm, as shown in FIG.
An arm 2 having a hand 3 at the tip is mounted on an upper part of (an example of a moving body). Further, in the vicinity of the lower central portion of the automatic guided vehicle 1 (a position that is hardly affected by lighting or the like), an image pickup device having a ring-shaped lighting device 5 so as to face the floor surface 9 (an example of a predetermined plane). An image processing unit 6 (corresponding to a reference position data acquisition unit, a position data acquisition unit during moving work, and an encoding unit), a storage unit 7, and a teaching data correction unit 8 (Corresponding to associating means and deviation calculating means).
Further, as is usually performed in a semiconductor clean room or the like, a punching floor in which punching holes 11 (corresponding to a predetermined form part) are formed in a predetermined pattern as shown in FIG. Has been. Further, a recognition mark 10 as shown in FIG. 3 is provided on the punching floor 9 at an arbitrary position near the stop position of the automatic guided vehicle 1. The recognition mark 10 is provided in order to identify one location of the punched holes 11 arranged in the predetermined pattern, and is unique in a captured image due to characteristics such as shape, size, and brightness. Any mark can be used as long as it can be specified, for example, a mark attached between the punching holes 11 or a screw hole for fixing a floor material.

The above components will be described in more detail. A wafer (not shown) is mounted on the arm 2 in a state where the automatic guided vehicle 1 is stopped at a predetermined work position in advance.
Is transferred from the automatic guided vehicle 1 to a work table (not shown). At the time of the actual transfer operation, the arm 2 performs the operation according to the teaching data. In the imaging unit 4,
When the operation of the arm 2 is taught and when the arm 2 is stopped at a predetermined work position during the transfer operation, an image of the punching floor 9 including the recognition mark 10 is captured. The image processing unit 6 performs image processing on the captured image captured from the image capturing unit 4 after the image capturing by the image capturing unit 4 to obtain position data of each of the punching holes 11 and the recognition marks 10 (the image data of the captured image). (Coordinate values in the local coordinate system) are obtained. The position data obtained when teaching the operation of the arm 2 is stored in the storage unit 7. The teaching data correcting unit 8 stops the automatic guided vehicle 1 during the transfer operation and performs the processing by the image processing unit 6.
The position data of each punching hole 11 stored in advance and the punching holes 11 obtained by the image processing unit 6 this time.
Is performed on the basis of the position data of the recognition mark, and the shift amount of the stop position of the automatic guided vehicle 1 is determined based on the difference between the position data and the associated position data. Arm 2 based on the displacement amount.
Is corrected.

Hereinafter, the position correcting operation in the automatic guided vehicle A1 with an arm will be described more specifically with reference to the flowchart shown in FIG. First, prior to the actual transfer operation, a wafer (not shown) is transferred from the automatic guided vehicle 1 to the worktable (not shown) with respect to the arm 2 while the automatic guided vehicle 1 is stopped at a predetermined work position. (Shown) is taught. At this time, an image of the punching floor 9 (corresponding to a reference captured image) is captured by the imaging unit 4 (Step S).
1). Subsequently, the image processing section 6 performs image processing on the picked-up image (see FIG. 3) taken in from the image pickup section 4, thereby recognizing each punching hole 11 and the recognition mark 10 on the floor surface, and being recognized. A coordinate value (position data) in the local coordinate system CS1 in the imaging region R is calculated for each of the punched holes 11 and the recognition marks 10 (step S2). The position data can be obtained at an optimum position according to the shape of the punching hole, such as the center position of the punching hole 11, for example. Each punching hole 1
For the recognition of the recognition mark 1 and the recognition mark 10 and the calculation of the position data, for example, a method using template matching or a shape feature amount generally used in image processing can be used. The template matching method is
As shown in FIG. 4, the recognition mark 10 and the punching hole 1
1 is stored in advance as a template image,
This is a method of extracting and recognizing a portion having a high correlation value with the template image in a captured image. In the punching floor used in the present embodiment, since all the punching holes 11 are formed in a circular shape having the same size, by using a single image of the punching holes 11 as a template image as shown in FIG. Even when the captured image has a rotation angle, the matching process can be performed in a short time without rotating the template image with one template image, and a storage for storing a large number of template images can be performed. Requires no space. It should be noted that even when there are several sizes of the punching holes 11, it is possible to prepare by preparing the template images by the number.

[0016] The above-mentioned method based on the shape characteristic amount means that a captured image is binarized by a predetermined threshold value and a predetermined shape characteristic amount such as a width, a height and an area (see FIG. 5). Is a method of extracting and recognizing a part having. The above-described method using template matching has the advantage of being able to perform high-accuracy position detection because processing is performed as a gray-scale image without using a binarized image, but also has the disadvantage of requiring a large amount of calculation for processing. . Therefore, in a portion where a very high positional accuracy is not required, such as the detection of the recognition mark 10, the method using the above-mentioned shape feature is used and the punching hole 11
It is effective to use a combination of the above two methods, such as using the above-described method based on template matching for a part requiring high accuracy, such as detection of a pattern. The teaching position data obtained in step S2 is stored in the storage unit 7 (step S3).

Next, during the actual transfer operation, when the automatic guided vehicle 1 is stopped, an image of the punching floor 9 (corresponding to the image picked up during the transfer operation, see FIG. 6) is picked up by the image pickup section 4. (Step S4). Subsequently, the same processing as in step S2 is performed in the image processing unit 6 to obtain position data during work (step S5). continue,
The teaching data correction unit 8 performs one-to-one correspondence between the working position data obtained in step S5 and the teaching position data stored in the storage unit 7 (step S6). A specific example of the association will be described below. An index is added to each punching hole 11 corresponding to the position data at the time of teaching based on the coordinate value of the recognition mark 10. For example, as shown in FIG. 3, a, b, c,
.., Each column is assigned a code such as A, B, C,..., And a combination of each code of a row and a column (aA, dB
Etc.) to create an index. An index similar to the above is added to the work position data (see FIG. 6). The punching holes 11 to which the same index has been added at the time of teaching and at the time of work are associated. In the above association process, the position data of the punched hole 11 included only in the position data at the time of teaching or work is ignored. That is, the punching hole 11 for which the position data is calculated both during teaching and during work.
Subsequent processing is performed using only the position data. Therefore, even if some punching holes 11 cannot be recognized during the work due to the influence of disturbance light or the like, it is possible to cope with the problem without any problem. FIG. 7 shows an example of each associated punching hole 11. The teaching position data associated with the i-th position is (Tix, Tiy), and the work position data is (Pix, Py).
iy).

Subsequently, the teaching data correction unit 8 uses the set of position data associated in step S6 to determine the position / posture of the captured image at the time of teaching and the captured image at the time of work. That is, the deviation of the stop position of the automatic guided vehicle A1 from the predetermined stop position (the stop position during teaching) is determined (step S7). Specifically, first, the relationship between the teaching-time position data (Tix, Tiy) and the work-time position data (Pix, Piy) associated with the i-th is expressed by the following equation.

(Equation 1) Here, θ, dx, and dy are conversion parameters from the position data at the time of teaching to the position data at the time of work, θ is a rotation angle, and dx and dy are translation amounts in the x and y directions, respectively. Each set of position data (Tix, Ti
y), (Pix, Piy) are applied to the above equation (1), and the above θ, dx, dy is calculated using the least squares method, so that the positions of the captured image at the time of teaching and the captured image at the time of work are calculated.・
Posture deviation is required. As described above, since the above-described position / posture deviation is obtained using the position data of the plurality of punching holes 11 existing in the captured image, highly accurate measurement is possible. In addition, since the individual position data of the punched holes 11 is used, even if the positions of the punched holes include an error at the time of manufacturing, the measurement accuracy is not affected.
Subsequently, the teaching data of the arm 2 is corrected based on the deviation of the position / posture between the captured image at the time of teaching and the captured image at the time of work obtained at step S7 (step S8). The arm 2 performs a transfer operation according to the teaching data corrected in step S8 (step S9). Thereafter, each time the automatic guided vehicle 1 moves, the above steps S4 to S4 are performed.
9 is repeated.

As described above, in the automatic guided vehicle A1 with an arm according to the present embodiment, the position data of each punching hole 11 constituting the pattern formed on the floor surface is transmitted during teaching and operation. Each is obtained from the captured image,
The position data of the corresponding punching holes 11 are correlated between the position data at the time of the teaching and the position data at the time of the work, and the captured image at the time of the teaching and the captured image at the time of the work are used using a set of corresponding position data. Position / posture deviation is required. Therefore,
Since the deviation of the position / posture is obtained using a large number of sets of position data, highly accurate measurement is possible. In addition, since the individual position data of the punched holes 11 is used, even if the positions of the punched holes include an error at the time of manufacturing, the measurement accuracy is not affected. In addition, since the displacement of the position / posture is obtained using a large number of sets of position data, the position data of the punching hole 11 that is included only in the position data at the time of teaching or at the time of work is ignored. And the punching hole 1 for which the position data was calculated both during work
By using only the position data of No. 1, even when some punching holes 11 cannot be recognized during the work due to the influence of disturbance light or the like, processing can be performed without any problem.

[0020]

(Embodiment 1) In the position correcting operation of the automatic guided vehicle A1 with an arm according to the above-described embodiment, the accuracy of detection of the position data of each punching hole during teaching and work is associated with δ. The number n of sets of the position data described above is a large factor that determines the accuracy. The accuracy increases as the number n increases, but if the number is too large, the processing time and the memory capacity required for the processing in the associating process (step S6) and the calculation of the amount of displacement of the position and orientation (step S7) increase. There is a problem. Therefore, if the obtained position data is selected after the calculation of the work position data (step S5) and the number of position data used in the subsequent processing is limited to an appropriate number, high accuracy is maintained. In addition, it is possible to suppress an increase in the processing time and the memory used. In this case, the number of data after the selection can be calculated from the error required in the calculation of the displacement amount of the position / posture (step S7). That is, the calculation error Δ of the displacement amount of the position / posture
Is generally

(Equation 2) Since it has been statistically and experimentally confirmed that n becomes n, the allowable maximum value of Δ is Δm, and the number of data after selection is n ′, n ′ can be obtained by the following equation. n ′> (δ / Δm) ² (3) The n ′ pieces of position data may be randomly selected from the position data calculated in step S5.
It is more effective to preferentially extract position data suitable for the calculation of the position / posture deviation amount (step S7).
For example, if data near the outer periphery of the imaging region P (see FIG. 6) is preferentially selected, the accuracy with respect to the posture angle θ is improved.

(Second Embodiment) In the first embodiment, the position data is selected after the calculation of the work position data (step S5). However, the position data selection processing is performed in the step S5. If it is performed in the middle of the process, the effect of reducing the processing time and the memory used can be further increased. For example, the process of step S5 is divided into a process of calculating approximate position data and a process of calculating detailed position data, and a data selection process is performed between these processes. This will be specifically described with reference to the flowchart shown in FIG. After the image of the floor surface is captured in step S4, first, the captured image is subjected to a thinning-out or averaging process to reduce the image data, and each punched hole 11 is recognized from the reduced image data to be roughly outlined. Is calculated (step S5a). For example, if the number of vertical and horizontal pixels of an image is reduced to 1 / m by reduction, the processing time is approximately 1 / m ⁴ (in the case of template matching),
Alternatively, it can be greatly reduced to 1 / m ² (in the case of the shape feature amount). Thereafter, the position data selection processing as described in the first embodiment is performed (step S10), and the vicinity of the punching hole 11 corresponding to the selected position data is determined.
Detailed position data is calculated using the image data that has not been subjected to the above-described reduction processing (step S5b). With the above processing, it is possible to further suppress the processing time and the increase in the memory used while maintaining high accuracy.

(Embodiment 3) In the above embodiment, in the associating process of step S2, an index is added to each punching hole 11 based on the coordinate value of the recognition mark 10, so that the same is used for teaching and working. The punched holes 11 to which the indexes are added are associated with each other. However, this method is applicable only when the punching holes 11 are regularly arranged vertically and horizontally. Therefore, each punching hole 11
The procedure of the associating process that can be applied even when the arrangement of the patterns has no regularity will be described with reference to FIG. The position data of each punching hole 11 at the time of teaching is rearranged in an order close to the recognition mark Mt (see FIG. 9A). With respect to the position data at the time of teaching, the position data of the punching hole 11 closest to the recognition mark Mt is selected and set as the starting point Ts. The vector from the recognition mark Mt to the starting point Ts is defined as V0. With respect to the position data at the time of work (FIG. 9B), position data near the position extended from the recognition mark Mp by the amount of the vector V0 is searched, and this is set as a starting point Ps. At this time, the search range is set within the range of the radius r0, and if there is one position data in this range, the position data is set as the starting point Ps. If there is no corresponding position data in the search range, the process returns to the above, and the next closest position data is set to the new starting point T.
The above processing is performed as s. This is repeated until the starting point Ps is determined. Regarding the teaching position data, the vector from the starting point Ts to the other position data Ti is Vi (FIG. 9A).
reference). Regarding the work position data (FIG. 9B), near the position extended from the starting point Ps by the amount of the vector Vi, the position data Pi corresponding to the Ti is searched with the search range as the radius rp. If Pi is found, Ti and Pi
Are associated with each other. The above process is repeated for all Ti. The above processing can be applied when the deviation of the posture angle is small among the deviations of the position and posture between the punching floor and the automatic guided vehicle during teaching and work. When the deviation of the attitude angle is large, it can be dealt with by calculating the approximate deviation of the posture angle from the deviation of the angle between the vector V0 and the vector MpPs determined as described above, and performing the above processing based on the calculated deviation. By performing the above-described associating process, it is possible to cope with the case where the punching holes 11 are arranged vertically and horizontally irregularly.

(Embodiment 4) In the above embodiment, the recognition mark 1 provided separately from the punching hole 11 on the floor is provided.
Although an example using 0 has been shown, the punching hole 11 can be used as the recognition mark as long as it can be uniquely identified on the captured image. For example, in the example shown in FIG. 10, the interval between the punching holes 11 in the imaging region R is large, and the range R of the stop error of the automatic guided vehicle 1 is R.
Since it is guaranteed that there is only one specific punching hole 11a in s, it is possible to use the punching hole 11a as the recognition mark. In the above embodiment and each example, an automatic guided vehicle (moving body)
Has been described on the floor (plane) on which the pattern and the recognition mark are formed, but the moving body and the plane are not limited to the automatic guided vehicle and the floor.
For example, when a pattern and recognition marks are formed on a worktable (plane) and a camera is attached to the tip (moving body) of a robot arm that moves on the worktable, positioning is performed. Also, the present invention can be applied to a case where a pattern and a recognition mark are formed on a workpiece and the working device is positioned with respect to the workpiece.

[0024]

As described above, according to the present invention, the moving object moving on the floor on which an arbitrary pattern and a predetermined recognition mark are formed is stopped at a predetermined position, and an image of the floor is taken. The difference between the captured image at the time of the moving operation and the reference captured image obtained by capturing the image of the floor surface by previously stopping the moving body at a predetermined reference stop position is determined by the above-described arbitrary pattern on the captured image. The arbitrary pattern is formed from the reference captured image in a moving body stop position shift amount detecting device which is obtained by comparison based on the recognition mark and detects the shift amount of the stop position of the moving body based on the shift. Reference position data acquisition means for acquiring position data of the plurality of feature parts and the recognition mark on the reference captured image, and the plurality of feature parts and the recognition mark from the movement work captured image. A moving work position data acquiring means for acquiring position data on the moving work captured image, and a common code for each of the plurality of feature parts on the basis of the recognition mark on each of the captured images. Encoding means, and associating means for performing one-to-one correspondence of the plurality of feature parts between the respective captured images based on the codes of the plurality of feature parts obtained by the encoding means, For each set of the plurality of form parts associated by the associating means, a difference between the position data obtained by the reference position data obtaining means and the position data obtaining means during moving work is obtained, and A shift calculating means for calculating a shift between the respective captured images based on the difference in the position data. By displacement of the position and orientation using a set of position data is determined, it is possible to measure with high accuracy. Further, since the individual position data of each of the feature parts is used, even if the position of the feature part includes an error at the time of manufacture, the measurement accuracy is not affected. In addition, since the displacement of the position / posture is obtained using a large number of sets of position data, the position data of the above-mentioned form part, which is included only in the position data of the reference captured image or the captured image during the moving work, is ignored. However, by using only the position data of the form part for which the position data is calculated in both of the captured images, even if some form parts cannot be recognized during the moving work due to the influence of disturbance light or the like, processing can be performed without any problem.

Further, a predetermined number of position data are selected from the position data of the form part obtained by the reference position data obtaining means and the moving work position data obtaining means, and a form corresponding to the selected position data is selected. By associating only parts with the associating means, it is possible to increase processing speed and reduce memory capacity while maintaining a certain level of accuracy. Alternatively, the moving work position data obtaining means may include a general position data obtaining means for obtaining general position data of the plurality of feature parts, and a predetermined number of general position data obtained from the general position data obtaining means. General position data selecting means for selecting the general position data, and detailed position data obtaining means for obtaining detailed position data for the form part corresponding to the general position data selected by the general position data selecting means. With such a configuration, it is possible to further speed up the processing and reduce the memory capacity. In the case where the arbitrary pattern is constituted by a regular arrangement of the plurality of feature parts, the encoding means is configured to determine the plurality of feature parts based on the recognition mark on each of the captured images. By performing the coding based on the regularity of the above arrangement, easy and reliable coding can be performed.

On the other hand, for example, when the arbitrary pattern is constituted by an irregular arrangement of the plurality of feature parts,
Since the above-described coding method is not used, the code of each feature on the reference captured image is determined based on a vector connecting the plurality of features in order from the recognition mark on the reference captured image. This can be dealt with by attaching the reference mark image in common with the reference picked-up image in accordance with the vector starting from the recognition mark on the picked-up image at the time of the moving work. In the reference position data acquiring means and the moving work position data acquiring means,
By matching a predetermined template image with the plurality of feature parts and the image of the recognition mark, the plurality of feature parts and the recognition mark are recognized, and position data on each of the captured images is acquired. Thus, highly accurate position data can be obtained. At this time, when a plurality of feature parts constituting the arbitrary pattern are formed in the same circle, respectively, the template image and the individual images are used by using a single image of the circular feature part as the predetermined template image. Regardless of the deviation of the attitude angle from the captured image, the matching processing can be performed at high speed and with a small memory usage. In the reference position data acquiring means and the moving work position data acquiring means, a portion having a predetermined shape characteristic amount is determined from a binarized image obtained by binarizing each captured image with a predetermined threshold value. If the plurality of feature parts and the recognition marks are recognized by extraction and the position data on each of the captured images is acquired, the processing speed can be increased although the accuracy is not as good as the matching processing. It is possible. In addition, if one or more of the features constituting the arbitrary pattern can be specified as the recognition mark, the processing by the coding means can be performed with reference to the features. Becomes
Note that the arbitrary pattern and the predetermined recognition mark are not necessarily formed on the floor, but may be formed on a plane on which the moving body moves, for example, on a work table.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a schematic configuration of an automatic guided vehicle A1 equipped with a stop position deviation amount detection device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a processing procedure by the stop position shift amount detecting device.

FIG. 3 is a diagram showing an example of a captured image at the time of teaching and an example of coding;

FIG. 4 is a view showing a template image used for template matching processing.

FIG. 5 is a diagram illustrating an example of a shape feature amount.

FIG. 6 is a view showing an example of a picked-up image and a sign at the time of work.

FIG. 7 is a diagram illustrating an example of position data of an associated punching hole.

FIG. 8 is a flowchart illustrating a processing procedure of step S5 according to the second embodiment.

FIG. 9 is an explanatory diagram of an associating process when the arrangement of punching holes according to the third embodiment is not regular;

FIG. 10 is an explanatory diagram when a specific punching hole according to the fourth embodiment is used as a recognition mark.

FIG. 11 is a schematic diagram showing a schematic configuration of an automatic guided vehicle A0 equipped with a stop correction device according to a conventional example.

FIG. 12 is a flowchart showing a processing procedure by the conventional stop correction device.

FIG. 13 shows a position at the time of teaching by the conventional stop correction device.
Location information (position coordinates (X ₀, Y₀), Rotation angle θ₀) Calculation
Explanatory drawing which shows an example of a method.

FIG. 14 shows a position at the time of work by the conventional stop correction device.
Location information (position coordinates (X ₁, Y₁), Rotation angle θ₁) Calculation
Explanatory drawing which shows an example of a method.

FIG. 15 is an explanatory diagram of a template matching process according to a conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Automatic guided vehicle (an example of a moving body) 2 ... Arm 3 ... Hand 4 ... Imaging part 5 ... Ring-shaped illumination device 6 ... Image processing part (reference position data acquisition means, position data acquisition means during moving work, and numbering 7 ... Storage unit 8 ... Teaching data correction unit (corresponding to associating unit and deviation calculating unit) 9 ... Floor surface (punching floor) 10 ... Recognition mark 11 ... Punching hole (predetermined form part)

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Koji Miura 150 Motoyashiki, Miyamachi, Toyohashi-shi, Aichi Prefecture Shinko Electric Co., Ltd. Toyohashi Office

Claims (10)

[Claims] 1. A moving work captured image obtained by stopping a moving body moving on a floor on which an arbitrary pattern and a predetermined recognition mark are formed at a predetermined position and capturing an image of the floor,
The deviation from a reference captured image obtained by capturing the image of the floor surface by previously stopping the moving body at a predetermined reference stop position is determined by comparing the discrimination mark of the arbitrary pattern on each captured image with reference to the recognition mark. A moving body stop position displacement amount detecting device for detecting the displacement amount of the moving body stop position based on the displacement and obtaining a plurality of feature parts constituting the arbitrary pattern from the reference captured image; A reference position data acquiring means for acquiring position data of the mark on the reference captured image; and a moving of the plurality of feature parts and the recognition mark on the captured image of the moving work from the captured image of the moving work. Moving work position data obtaining means for obtaining position data; and a code for performing a common code on each of the plurality of form parts based on the recognition mark on each of the captured images. Means for performing a one-to-one correspondence of the plurality of feature parts between the respective captured images based on codes of the plurality of feature parts obtained by the signing means; The difference between the position data obtained by the reference position data obtaining means and the position data obtaining means at the time of moving work is determined for each set of the plurality of form parts associated by the attaching means, and the position data in each set is obtained. And a displacement calculating means for calculating a displacement between the captured images based on the difference between the stop positions. 2. A method according to claim 1, wherein a predetermined number of position data are selected from the position data of said form part obtained by said reference position data obtaining means and said moving work position data obtaining means, and a form corresponding to the selected position data is selected. 2. The moving-body stop position deviation amount detecting device according to claim 1, wherein the association is performed by the associating means only for the units. 3. The moving work position data acquiring means,
A rough position data obtaining means for obtaining rough position data of the plurality of form parts, a rough position data selecting means for selecting a predetermined number of position data from the rough position data obtained by the rough position data obtaining means; 2. The stop position of a moving body according to claim 1, further comprising: detailed position data acquisition means for acquiring detailed position data on the form part corresponding to the approximate position data selected by the approximate position data selection means. Deviation detector. 4. The method according to claim 1, wherein the arbitrary pattern is constituted by a regular array of the plurality of feature parts, and the coding means is configured to determine the plurality of feature parts based on the recognition mark on each of the captured images. The moving-body stop position displacement amount detecting device according to any one of claims 1 to 3, wherein coding is performed based on the regularity of the array. 5. The coding means for coding each feature on the reference captured image based on a vector connecting the plurality of features in sequence starting from the recognition mark on the reference captured image. The moving object stop position shift amount detecting device according to any one of claims 1 to 3, wherein the moving object is subjected to common signing with the reference picked-up image in accordance with the vector starting from the recognition mark on the picked-up image during the moving operation. . 6. The reference position data acquiring means and the moving work position data acquiring means, by performing a matching process between a predetermined template image and the plurality of feature parts and the image of the recognition mark. 6. The moving-body stop position displacement amount detection device according to claim 1, wherein the recognition mark is recognized, and position data on each of the captured images is acquired. 7. The moving body according to claim 6, wherein the plurality of feature parts constituting the arbitrary pattern are each formed in the same circle, and a single image of the circular feature part is used as the predetermined template image. Stop position shift amount detection device. 8. A method according to claim 1, wherein said reference position data obtaining means and said moving work position data obtaining means use a binarized image obtained by binarizing each of the captured images with a predetermined threshold value to obtain a predetermined shape characteristic amount. The stop position shift amount of the moving object according to any one of claims 1 to 7, wherein the plurality of form parts and the recognition mark are recognized by extracting a part having the stop position, and position data on each of the captured images is acquired. Detection device. 9. The processing according to claim 1, wherein the encoding unit performs processing based on one or more of the form parts that can be specified as the recognition mark among the form parts constituting the arbitrary pattern. The moving position stop amount detecting device for a moving body according to any one of the above. 10. A moving work captured image obtained by stopping a moving body moving on a predetermined plane on which an arbitrary pattern and a predetermined recognition mark are formed at a predetermined position and capturing an image of the plane, A deviation from a reference captured image obtained by stopping the moving body at a predetermined reference stop position and capturing the image of the plane is determined by comparison with the recognition mark of the arbitrary pattern on each captured image as a reference. In the moving body stop position shift amount detecting device that detects the shift amount of the stop position of the moving body based on the shift, a plurality of feature parts constituting the arbitrary pattern and the recognition mark are determined from the reference captured image. Reference position data acquisition means for acquiring position data on the reference captured image, and the moving image of the plurality of features and the recognition mark from the moving image. Moving work position data obtaining means for obtaining the position data above, coding means for performing common coding on each of the plurality of forms based on the recognition mark on each of the captured images, An associating means for performing one-to-one correspondence of the plurality of feature parts between the respective captured images based on the codes of the plurality of feature parts obtained by the encoding means; For each set of the plurality of attached feature parts, a difference between the position data obtained by the reference position data obtaining means and the position data obtaining means during moving work is obtained, and based on the difference between the position data in each set. And a shift calculating means for calculating a shift between the captured images.