JP2003089086A - Robot controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a robot controller capable of detecting a position of a workpiece using an inexpensive, simple, and compact image sensor and compensating it. SOLUTION: This robot controller is provided with a camera 101 for photographing scene images of a fixed workpiece 6 and a grip workpiece 3, a template preparation processing part 102 for generating a template image required for judging a work condition from the scene images, a target position setting part 105 for setting a target position for positioning the grip workpiece 3, a movement amount prior preparation part 106 for generating a movement amount of a robot 104 by comparing a position of the grip workpiece 3 with the target position, and a work condition template matrix 103 for preparing the template image and the movement amount, corresponding the template image and the movement amount by 1 to 1, and storing them by changing positions of the grip workpiece 6 systematically.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control device for a robot used for assembling an article, and more particularly, a control device for a robot for picking up an image of an article to be worked and correcting the position of the article based on the image. It is about.

[0002]

2. Description of the Related Art When an article is assembled by a robot, it is necessary to accurately position the article to be assembled. However, it is difficult to obtain a desired accuracy due to a dimensional error of the work itself, a deviation of a gripping position when gripping the work with a hand, and the like. Therefore, various types of devices and methods have been proposed in which the positioning error of the work is detected by various sensors and the position of the work is corrected based on the error. For example, in International Publication WO98-17444,
An apparatus for measuring and correcting the position of a work with a laser range finder is described. Further, an apparatus using a stereo image sensor, in which a work is photographed by two cameras and the position of the work is measured based on a deviation of the image, is well known.

[0003]

However, in the above-mentioned prior art, there is a problem that a laser range finder using a laser is expensive, and a black work that absorbs laser light or a mirror surface that reversely reflects it. There was also a problem that it could not be applied to finishing work. The stereo image sensor has a problem in that it takes a lot of time to calibrate and it is difficult to obtain a desired accuracy, and since two cameras have to be installed at a certain interval, a robot or other device There was also the problem that it was difficult to choose an installation position that would not interfere with. Therefore, an object of the present invention is to provide a robot controller that detects and corrects the position of a work by using an inexpensive, simple and compact image sensor.

[0004]

In order to solve the above-mentioned problems, the invention of claim 1 is a robot controller for assembling a gripping work gripped by a tip of a robot to a fixed work fixed to a workbench. A camera that captures a scene image of the work and the gripping work, a template creation processing unit that generates a template image necessary for determining the work state from the scene image, and a target that sets a target position for positioning the gripping work. A position setting unit, a motion amount pre-creation unit that generates a motion amount of the robot by comparing the position of the gripping work and the target position, and the template image and the template image while systematically changing the position of the gripping work. A motion amount is created, and the template image and the motion amount at a specific position are made to correspond one to one,
It is stored in the work state template matrix. Further, the invention of claim 2 obtains a correlation value between the scene images of the fixed work and the gripping work and a template image in the work state template matrix created in advance during the playback operation, and calculates the correlation value of the correlation value. A motion amount generation unit that derives the motion amount corresponding to the largest template image is provided, and the position of the gripping work is corrected by the motion amount. Further, the invention of claim 3 performs a correlation operation between the scene images of the fixed work and the gripping work and the template image in the work state template matrix created in advance during the playback operation,
The operation amount in the work state template matrix is weighted with the correlation value to obtain a weighted average of the operation amounts, and the operation amount correction unit uses the weighted average as a correction operation amount. The position of the gripping work is corrected by. According to a fourth aspect of the present invention, the motion amount generating section obtains a correlation value between the scene image, a predetermined search start position and a template image in the vicinity thereof.
And the second step of selecting the template having the maximum correlation value among the template images obtained in the first step as the selected template, and the correlation value of the scene image, the selected template and the template in the vicinity thereof. And a third step of selecting a template having a maximum correlation value as a new selection template, and repeating the third step until there is no template having a correlation value larger than the selection template in the vicinity of the selection template. It is a thing.

[0005]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a robot system showing a first embodiment of the present invention. In the figure,
Reference numeral 101 denotes a camera that captures a fixed work and a gripping work that is gripped by the robot 104 and assembled to the fixed work. The scene image captured by the camera 101 is sent to the template creation processing unit 102 and converted into a template image. It When converting a scene image into a template image, there are a method of setting an image area in advance and a method of manually specifying an appropriate area, but either method may be used. The position data of the gripping work is the robot 1
It is inside 04. This position data and target position setting unit 1
The target amount set in advance in 05 is compared, and the motion amount pre-creating unit 106 creates a motion amount for moving the gripping work to the target position. This amount of movement is a relative movement amount that indicates how the robot 104 at the time of creating the template image can reach the set target position. The template image created by the template creation processing unit 102 and the motion amount created by the motion amount pre-creation unit 106 are stored in the work state template matrix 103 in a one-to-one correspondence.

Next, a first embodiment of the present invention will be described by taking a fitting work which is one of the assembling works as an example. FIG. 2 is an external view of the robot system showing the first embodiment of the present invention. The same components as those shown in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted. In the figure, 1 is a robot controller, 2 is an open / close type hand that is attached to the tip of the robot 104 and holds the gripping work 3, and 4 is a cable that connects the robot controller 1 and the robot 104. 5 is a workbench on which the fitting work is performed. The fixed work 6 has a round hole into which the gripping work 3 is inserted. The fixed work 6 is fixed to the workbench 5 using a fixing jig 7. 8 is the camera 101 and the workbench 5
Numeral 9 is a cable for connecting the camera 101 and the robot controller 1 to a camera fixing jig. Also,
Σw is a work coordinate on the fixed work 6.

Next, in the robot system shown in FIGS. 1 and 2, the work state template matrix 103
The procedure for creating the will be described. FIG. 3 is a flowchart of a teaching sequence of a work state template matrix showing an embodiment of the present invention, and FIG. 4 is an outline view showing a state in which a gripping work is approached to a fixed work. The teaching sequence will be described below with reference to the flowchart. S301: The robot 104 gripping the gripping work 3 is moved to the approach position of the fixed work 6 so that a template image can be created. The approach position here is a position where the gripping work 3 contacts the fixed work 6, as shown in FIG. This approach position is the robot 1
This is the position of the state in which the gripping work 3 is pressed against the fixed work 6 by the force control 04. S302: The template creation processing unit 102 converts the scene image obtained from the camera 101 into a template image. S303: The target position set by the target position setting unit 105 is compared with the position data of the gripping work obtained from the robot 104, and the motion amount pre-creation unit 106 creates a motion amount. S304: S and the template image created in S302
The motion amount created in 303 is stored in the work state template matrix 103 in one-to-one correspondence. S305: It is checked whether all template images have been created. If they have been created, the process ends. If they have not been created, the process proceeds to S306. S306: The approach position of the robot 104 is moved by a preset amount, and the process returns to S302. In such a sequence, the approach position of the gripping work 3 with respect to the fixed work 6 is systematically changed, and the amount of movement for moving the gripping work 3 from the template image and the approach position at a certain approach position to the target position is set to 1 Work status template matrix 103 associated with one-to-one
Is created.

Next, an apparatus and a procedure for performing the fitting work will be described with reference to the work state template matrix during the playback operation. FIG. 5 is a block diagram of a robot system showing a second embodiment of the present invention. In the figure, 501
Is a motion amount generation unit, which performs a correlation operation between the scene image captured by the camera 101 and the template image of the work state template matrix 103, and calculates the motion amount corresponding to the template image having the maximum correlation value from the work state template matrix 103. Is selected, and the selected motion amount is output to the robot controller 502. In addition, the robot controller 5
Reference numeral 02 moves the gripping work 3 to the target position based on the movement amount. Details of the method of selecting the motion amount from the correlation value will be described later.

FIG. 6 is a flowchart of a fitting operation showing a second embodiment of the present invention, and FIG. 7 is a flow chart of S604 in FIG.
5 is a flowchart illustrating the detailed processing content of FIG. The sequence of the fitting operation will be described below based on these flowcharts. S601: The robot 104 moves from the standby position to the gripping work 3
Move to the position to grip. S602: The open / close type hand 2 grips the gripping work 3. S603: The robot 104 moves to the approach position before the start of the fitting work. At this time, the relative positional relationship between the gripping work 3 and the fixed work 6 is not accurately positioned because of the uncertainty of the positioning accuracy when gripping the gripping work 3 and the uncertainty of the fixing accuracy of the fixed work 6. Absent. The approach position here is the gripping work 3 as shown in FIG.
Is a position in a state of being pressed by the force control of the robot 104 until it contacts the fixed work 6. S604: A template image closest to the scene images of the gripping work 3 and the fixed work 6 obtained from the camera 101 is selected from the work state template matrix 103 according to a processing procedure described later, and the gripping work 3 and the fixed work 6 are selected.
Understand the relative positional relationship of. S605: From the relative positional relationship detected in S604, it is determined whether or not the gripping work 3 is at a regular position with respect to the fixed work 6, that is, whether or not it is in the target state. S606: The operation amount corresponding to the template image selected in S604 is calculated as the work state template matrix 103.
The robot 104 is moved by the acquired motion amount, and the process returns to S604. S607: Since the gripping work 3 has come to the regular position with respect to the fixed work 6, the gripping work 3 can be inserted into the fixed work 6. Therefore, the insertion operation by force control of the robot 104 is executed. The insertion operation by force control is a well-known matter, and therefore its explanation is omitted. S608: The hand 2 releases the grip of the grip work 3. S609: The robot 104 moves to the standby position.

The detailed sequence of S604 described above will be described with reference to FIG. S701: Template images are selected one by one from the work state template matrix 103 prepared in advance. S702: Correlate the template image selected in S701 with the scene image obtained from the work state monitoring camera 101, and output the correlation value. This correlation calculation uses methods such as residual correlation calculation and normalized correlation calculation. Note that these methods are publicly known items, and therefore description thereof will be omitted. S703: It is determined whether all the template images in the work state template matrix 103 are selected, and if there is an unselected template image, the process returns to S701, and the correlation calculation processing is performed on the unselected template images. If is selected, the process proceeds to S704. S704: Select the template image with the largest correlation value. The template image with the maximum correlation value is the template image that is the closest to the scene image. S705: A signal corresponding to the determined template image is output, and the process ends.

Next, the work state template matrix 103 used in the fitting work will be described. 8A and 8B are explanatory diagrams of a work state template matrix showing a second embodiment of the present invention, FIG. 8A is a plan view for explaining the operation amount, and FIG. 8B is a work state template matrix. It is a conceptual diagram. FIG. 8A is a plan view of the fixed work 6, and the X axis and the Y axis of the coordinates in the figure are the X axis and the Y axis of the coordinate Σw fixed to the fixed work 6 shown in FIG. The circle shown by the solid line in the figure is the outline of the round hole (the hole into which the gripping work 3 is fitted) of the fixed work 6, and the circle shown by the dotted line is the insertable area. Further, the coordinate origin, that is, the position of (X, Y) = (0, 0) is the target position setting unit
This is the target position set in 5. Here, the insertable area means an area where the fitting work by force control can be performed by using chamfering or taper if the center position of the cylinder of the gripping work 3 is within this area. . Also, in the figure, there are seven dividing lines of +3 to -3 parallel to each of the XY axes. FIG. 8B is a conceptual diagram of the work state template matrix. The work state template matrix has the gripping work 3 at the intersection of the squares of the dividing line in FIG.
3 is a list of when template images are created by placing a. In the description of the teaching sequence of the work state template matrix, the phrase that the approach position of the gripping work is systematically changed means that the position is regularly changed like the intersection of the squares.
The amount of motion corresponding to each template image in FIG. 8B is shown in the lower left of the template image. If the position corresponding to the template image is (X, Y), the motion amount is (−X,
-Y). For example, the template image of (X, Y) = (0,0) is an image in which the convex portion of the gripping work 3 and the concave portion of the fixed work 6 overlap each other, and the operation amount is (0,0). Also, (X, Y)
In the template image of = (− 3,0), the convex portion of the gripping work 3 is reflected behind the concave portion of the fixed work 6, and the motion amount is (3,0). As in the table of FIG. 8B, each template image shows the positional relationship between the gripping work 3 and the fixed work 6. In this way, the data including the plurality of template images shown in FIG. 8B and the motion amount is stored in the work state template matrix 103.

Here, referring to FIGS. 4, 6, 9, and 8, the concrete processing procedure of the fitting work sequence of the gripping work 3 and the fixed work 6 will be described again. S603: “Move to approach position before starting fitting work”
The positional relationship between the gripping work 3 and the fixed work 6 when the work is performed is shown in FIG.
Suppose you came to the position shown in. S604: In “Understanding state from scene image of camera 101”, a scene image as shown in FIG. 9 is obtained, and the scene image and the template image in the work state template matrix 103 are subjected to correlation calculation to obtain the image shown in FIG. ) Of (X, Y)
= (− 3,0) template image is selected. S605: In the determination of "target state?", It is determined that the state is not the target state. S606: “Only the amount of movement in the grasped state is the robot 104
Move amount (3,0) corresponding to the template image (-3,0) selected in "Move"
03, and the robot controller 502 sets the robot 1
04 is moved by +3 in the X direction of the Σw coordinate. S604: “Understanding the state from the scene image of the camera 101” is performed again, and (X, Y) of FIG.
A template image of (0,0) is selected. S605: The determination of “target state?” Is performed again and it is determined that the target state is set, and S607 to S609 are sequentially executed and the process ends. In this embodiment, in order to simplify the description,
In S603, it is assumed that the gripping work 3 is pressed until it contacts the fixed work 6, and X of the fixed work coordinate system Σw is pressed.
It is assumed that the end surface of the gripping work 3 is on the Y plane, and by doing so, the work state template matrix 103 is created two-dimensionally instead of three-dimensionally. It is also possible to create the work state template matrix 103 three-dimensionally, and in that case, it is also possible to three-dimensionally grasp the relative positional relationship between the gripping work 3 and the fixed work 6.

Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 10 is a block diagram of a robot system showing a third embodiment of the present invention, and is a diagram illustrating in detail the motion amount generation unit 501. In addition, about the above-mentioned component, the same code | symbol is attached and description is omitted. In the figure,
The motion amount generation unit 501 includes a correlation calculation unit 1001 and a motion amount correction unit 1002. Correlation calculation unit 1001
Performs a correlation operation from the scene image and the template image and outputs the correlation value. The motion amount correction unit 1002 calculates the corrected motion amount from the correlation value and the motion amount, and the robot control unit 5
Send to 02.

FIG. 11 is a flowchart of a fitting work sequence showing a third embodiment of the present invention, and FIG. 12 is a flowchart of FIG.
3 is a flowchart showing S1104 of No. 1 in more detail. The flowchart shown in FIG. 11 is basically the same as the flowchart shown in FIG. 6, but S604 to S606.
Instead, it is different in that S1104 to S1106 are provided. Each of these different steps will be described below. S1104: The correction operation amount is derived from the scene image and the template image obtained from the camera 101 according to the processing procedure shown in FIG. The detailed processing of FIG. 12 will be described later. S1105: It is determined whether the movement amount of the correction motion amount derived in S1104 is smaller than the radius of the insertable area shown in FIG. 8A. If the movement amount of the correction movement amount is smaller than the radius of the insertable area, the result is Yes. S1106: The corrected motion amount derived in S1104 is output to the robot controller 502. Here, the detailed sequence of S1104 will be described with reference to FIG. Figure 1
The flowchart shown in 2 is basically the same as the flowchart shown in FIG. 7, except that S1204 and S1205 are provided instead of S704 and S705.
Each of these different steps will be described below. S1204: A corrected motion amount described below is derived using the correlation value and motion amount of each template image. S1205: Output the determined correction motion amount, and terminate the correction motion amount derivation sequence. Here, the correction operation amount will be described. The correction motion amount (x, y) is determined by a weighted average formula such as the following formula (1) after obtaining the correlation value of each template image.

[0015]

[Equation 1]

However, the formula (1) is a formula for determining the two-dimensional correction motion amount, and in the case of the three-dimensional formula, the following formula (2) is used.
Will be used.

[0017]

[Equation 2]

In equations (1) and (2), L, M and N are work
The number of elements in the state template matrix. That is
Taking the work state template matrix of FIG. 8B as an example
Then, the elements in the X-axis direction are -3, -2, -1, 0, 1,
Since there are 7 of 2 and 3, L = 7. The elements in the Y-axis direction are also −
There are 7 of 3, -2, -1,0,1,2,3, so M = 7
is there. Similarly, N is the number of elements in the Z-axis direction. Also, i,
j and k are (i = 0,1, ..., L), (j = 0,1, ..., M), (k = 0,
1, ..., N), which is a number indicating the element of each axis. i
X th coordinate of X-axis_iAnd the j-th Y-axis coordinate value is Y_jWhen
And set the kth Z-axis coordinate value to Z _kAnd then S_ijIs two-dimensional (x,
y) ＝ (X_i, Y_j), S is the correlation value_ijkIs (x, y,
z) ＝ (X_i, Y_j, Z_k) Is the correlation value.

Here, a method of deriving a two-dimensional correction motion amount will be specifically described with reference to FIG. FIG. 13 is a conceptual diagram of a work state template matrix for explaining a method of deriving the correction operation amount. The numbers from 0 to 90 shown in the figure indicate the correlation value between the scene image and each template image. The correlation value between the scene image obtained immediately after S603 and each template image of the work state template matrix is obtained, and the values shown in FIG. 13 are obtained. By substituting these values into the equation (1), the following equations (3) and (4) are obtained, and the corrected motion amount (x, y) is obtained.

[002]

[Equation 3]

[0021]

[Equation 4]

If the corrected motion amount (x, y) = (2.229,0.085), the position is (−2.229, −0.085), and it is outside the insertable area shown in FIG. 8A in S1105. Is determined S11
The processing shifts to 06. In S1106, xy of the Σw coordinate
The robot 104 is moved by the vector (2.229,0.085) on the plane. After execution of S1106, S1104 is performed again to obtain the correction motion amount from the scene image. If the position obtained from the obtained correction motion amount is within the insertable area, S607.
To start the insertion operation. In this example, the formula (1) and the formula
Although the corrected motion amount is generated by using the weighted average formula as shown in (2), it goes without saying that a calculation formula such as a quadratic formula, a biquadratic formula, and a cubic formula can be applied.

Next, a fourth embodiment of the present invention will be described with reference to the drawings. FIG. 14 is a flow chart showing the fourth embodiment of the present invention, which shows the flow of S604 in detail. FIG. 15 is a conceptual diagram of the work state template matrix. The processing sequence of the fourth embodiment will be described below. S1401: Correlation calculation with a scene image is performed using a predetermined template image for starting search and a plurality of template images in the vicinity thereof, and a correlation value of each template image is derived. S1402: The template image having the maximum correlation value derived in S1401 is selected. S1403: It is compared whether or not the correlation calculation with the scene image has been performed for all of the template images in the vicinity of the template image selected in S1402. S has been completed
If it has not been carried out, the procedure proceeds to S1404. S1404: Since at least one template image for which correlation calculation has not been performed exists in the plurality of template images in the vicinity of the template image selected at S1402, the correlation value is calculated for the template image for which correlation calculation has not been performed and the scene image. Derive. S1405: The motion amount corresponding to the template image having the maximum correlation value is output.

Here, the above-mentioned sequence will be specifically described with reference to FIG. First, the search start position of the work state template matrix is set to the position (0,0). S1
In 401, the search start template (0,0) and the scene image are subjected to correlation calculation to derive the correlation value. Then (1,0), (1,1), (0,1), (−1,0) near the search start template (0,0)
1), (-1, 0), (-1, -1), (0, -1) and (1, -1) The correlation calculation of the template image and the scene image is performed, and the correlation values at 9 locations are Derived. In S1402, the template image having the maximum correlation value is selected from the nine correlation values derived in S1401. In FIG. 15, the correlation value 40 (-
Since (1,0) is the maximum, the template image of (-1,0) is selected. The template image selected in S1403
(0,0), (0,1), (−1,1), (−2,
1), (−2,0), (−2, −1), (−1, −1) and (0, −1) are checked. As a result of checking, (−2,1), (−2,0)
No correlation calculation of (−2, −1) was found to have been performed.
Will be determined. In S1404, (−2,1) and (−2,0)
Correlation calculation between the template image of (−2, −1) and the scene image is performed to derive the correlation value. In S1402, S1404
And the correlation value of (−2,1), (−2,0) and (−2, −1)
The correlation value of (−1,0) is compared, and the template image having the maximum correlation value is selected. In FIG. 15, the correlation value 85 (−2,
Since (0) is the maximum, the template image of (−2,0) is selected. In S1403, the selected template image (-
(−1,0), (−1,1), (−2,1), (−3,
1), (−3,0), (−3, −1), (−2, −1) and (−1, −1) are checked. As a result of checking, (−3,1), (−3,0)
No correlation calculation of (−3, −1) was found to have been performed.
Will be determined. In S1404, (−3,1) and (−3,0)
Correlation calculation between the template image of (-3, -1) and the scene image is performed to derive the correlation value. In S1402, S1404
And the correlation value of (−3,1), (−3,0) and (−3, −1)
The correlation value of (−2,0) is compared, and the template image having the maximum correlation value is selected. In FIG. 15, the correlation value 90 (−3,
Since (0) is the maximum, the template image of (−3,0) is selected. In S1403, (−2,0), (−2,1), (−3,1), (−3, −1) and (−2,0) which are in the vicinity of the template image (−3,0).
-1) is checked. As a result of the check, it is determined that the correlation calculation has been performed at all five places, and the result is Yes. Correlation value that maximizes the correlation value in S1405
The 90 (−3,0) template image is selected, and the motion amount (3,0
0) is output. In the above specific example, the number of times the correlation value is derived by the correlation calculation is 15. If the template image with the maximum correlation value is determined after deriving the correlation values of all the template images as in the second embodiment, the number of correlation calculations is 49 times, and the calculation amount of 35 times is reduced. Has been implemented. As the search start position, a position with a high probability of having a maximum correlation value may be selected by empirical measurement. In this embodiment, by starting from a template image at a predetermined search start position and performing a hill climbing search in which the search proceeds in a direction in which the correlation value becomes larger,
The template image having the maximum correlation value is selected, and the template image having the maximum correlation value is selected only by performing the correlation calculation on some template images and the scene images.

[0025]

As described above, according to the inventions of claims 1 and 2, since a large number of template images and the amount of motion corresponding to the template image are stored in a one-to-one manner, the position of the gripping work can be determined. Since the operation amount for correction can be determined by template matching, the image sensor can be configured with one camera. Therefore, there is an effect that the device can be configured easily, inexpensively and compactly. According to the third aspect of the invention, since the motion amount of each template image is weighted by the correlation value and the weighted average is calculated as the corrected motion amount, the position can be corrected with high accuracy. In the invention of claim 4, the search is started from the template image at the predetermined search start position, and the hill-climbing search is performed in the vicinity of the template image in the direction in which the correlation value is high. Therefore, the processing time can be shortened.

[Brief description of drawings]

FIG. 1 is a block diagram of a robot system showing a first embodiment of the present invention.

FIG. 2 is an external view of the robot system showing the first embodiment of the present invention.

FIG. 3 is a flowchart of a teaching sequence of a work state template matrix showing a first embodiment of the present invention.

FIG. 4 is an outline view showing a state where a gripping work is approached to a fixed work.

FIG. 5 is a block diagram of a robot system showing a second embodiment of the present invention.

FIG. 6 is a flowchart of a fitting operation showing a second embodiment of the present invention.

FIG. 7 is a flowchart showing detailed processing contents of S604 of FIG.

FIG. 8 is an explanatory diagram of a work state template matrix showing a second embodiment of the present invention. (A) is a plan view for explaining the operation amount, and (b) is a diagram for explaining a work state template matrix.

FIG. 9 is a scene image showing a second embodiment of the present invention.

FIG. 10 is a block diagram of a robot system showing a third embodiment of the present invention.

FIG. 11 is a flowchart of a fitting work sequence showing a third embodiment of the present invention.

FIG. 12 is a flowchart showing S1104 of FIG. 11 in more detail.

FIG. 13 is a conceptual diagram of a work state template matrix for explaining a method of deriving a correction motion amount.

FIG. 14 is a flowchart showing a fourth embodiment of the present invention.

FIG. 15 is a conceptual diagram of a work status template matrix showing a fourth embodiment of the present invention.

[Explanation of symbols] 1 robot controller, 2 open / close type hand, 3 gripping work, 4 robot controller cable,
5 work table, 6 fixed work, 7 fixed jig for fixed work, 8 camera fixed jig, 9 camera connection cable, 10
1 camera, 102 template creation processing unit, 103
Work state template matrix, 104 robot, 105 target position setting unit, 106 motion amount pre-creation unit, 501 motion amount generation unit, 502 robot control unit,
1001 correlation calculation unit, 1002 motion amount correction unit

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 3C007 AS06 BS10 KS05 KS07 KT01                       KT06 KT11 KT18 LT12                 5B057 AA05 BA02 DA07 DB02 DB05                       DB09 DC34                 5L096 AA03 AA06 BA05 CA02 DA02                       FA34 HA08 JA09

Claims (4)

[Claims] 1. A robot control apparatus for assembling a gripping work gripped by a tip of a robot to a fixed work fixed to a workbench, comprising: a camera for photographing a scene image of the fixed work and the gripping work; A template creation processing unit that generates a template image necessary for determining the state of work, a target position setting unit that sets a target position for positioning the gripping work, and a position of the gripping work and the target position are compared. A motion amount pre-creation unit that generates the motion amount of the robot, and the template image and the motion amount are generated while systematically changing the position of the gripping work, and the template image and the motion amount at a specific position. The robot control device is characterized by storing each one in a work state template matrix in a one-to-one correspondence. 2. During playback operation, a correlation value between scene images of the fixed work and the gripping work and a template image in the work state template matrix created in advance is obtained, and the template having the largest correlation value is obtained. The robot controller according to claim 1, further comprising a motion amount generation unit that derives the motion amount corresponding to an image, and corrects the position of the gripped work with the motion amount. 3. During a playback operation, a correlation calculation is performed between scene images of the fixed work and the gripping work and a template image in the work state template matrix created in advance, and the correlation image is calculated in the work state template matrix. Weighting the movement amount with the correlation value,
The weighted average of the movement amount is obtained, and a movement amount correction unit that uses the weighted average as a correction movement amount is provided, and the position of the gripping work is corrected by the correction movement amount. Robot controller. 4. The motion amount generation unit includes a first step of obtaining a correlation value between the scene image, a predetermined search start position and a template image in the vicinity thereof, and the template image obtained in the first step. Of the two, the template with the highest correlation value is selected as the selection template.
And a third step of obtaining a correlation value between the scene image, the selected template and templates in the vicinity thereof, and selecting the template having the maximum correlation value as a new selected template. The robot controller according to claim 2, wherein the third step is repeated until there is no template having a correlation value larger than that of the selected template.