JPH085021B2 - Workpiece positioning method - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for gripping a work and accurately positioning the work at a predetermined set position using an industrial robot.

2. Description of the Related Art Conventionally, in positioning and setting a work at a predetermined setting position using an industrial robot,
The work is conveyed to an accurate position by the conveying means, then the conveyed work is grasped by the hand of the robot, and the grasped work is moved to a predetermined set position to be positioned and positioned. I do. Thus, in various production lines, the assembly work can be automated by using the industrial robot. In such a prior art, since the work can be accurately carried to the predetermined position by the carrying means, the structure of the carrying means becomes complicated.

Further, in this prior art, even if the work can be accurately conveyed by the conveying means, blur may occur at the moment when the work is grasped by the hand of the robot, at which time the robot grasps the work. The position deviates from a predetermined reference gripping state. Therefore, in such a case, it is impossible to position the work with high accuracy.

When the position of the work conveyed and supplied by the conveying means is not constant, the supplied work is detected by using a visual sensor or the like, the position of the supplied work is captured, and the position of the work thus detected is detected. Based on the position information, the robot grips the work and positions it at a desired set position. Also in such a prior art, when the work is grasped by the hand of the robot, blurring may occur, and thus it is impossible to position the work at the set position with high accuracy.

SUMMARY OF THE INVENTION An object of the present invention is to provide a work positioning method capable of accurately positioning a work carried by a carrying means to a preset position by a robot. To do.

Means for Solving the Problem The present invention has a means for determining a posture of a workpiece on a conveyor in a predetermined rough posture in the middle of a conveyor path of a conveyor which has a horizontal conveying direction and conveys a workpiece. Further, a detection means for detecting the work is provided on the downstream side of the conveyor in the conveying direction with respect to the means for determining the posture, and the work is detected by the detection means.
The robot is arranged to grip the work by the hand of a robot arranged on one side of the conveyor, and the two-dimensional camera is arranged at a fixed position on the other side of the conveyor downstream of the detecting means. This camera is a method of positioning a work having a field of view facing the one side, wherein the work is held in a standard gripping state by the hand of the robot, and the hand is placed so that the work comes within the field of view of the camera.
With the camera, a workpiece held at a predetermined first reference position RT of the hand and stopped at a predetermined first reference posture T of the hand is grasped by the stopped hand in the reference grasping state. The first reference position Gs of the work and the first reference posture θs of the work
The detection result is stored in the memory, and the hand is moved and positioned so that the workpiece stopped in this reference gripping state is installed at the desired setting position of the workpiece and in the desired posture of the workpiece. The predetermined second reference position R0 of the hand and the predetermined second reference posture 0 of the hand are stored in the memory, the work is conveyed by the conveyer, and the work is conveyed on the conveyer by the means for determining the posture. Establish a predetermined posture, detect the conveyed work, grasp the detected work with the robot's fingertip, and hold the fingertip at the first reference position RT and the first reference point of the fingertip. The work is brought in the posture T and stopped, and the work in the visual field held by the stopped hand is imaged by the camera to obtain the second detection result which is the position G and the posture θ of the work. , Based on the first and second detection results, the position after correction by correcting the second reference position R0 of the hand so that the work is set at the desired setting position and in the desired posture
In addition to calculating RD, the second reference posture 0 of the hand is corrected to calculate the corrected posture D, and the corrected position RD and the corrected posture D are
A method for positioning a work, which includes positioning a hand.

Operation According to the present invention, first, the hand of the robot holding the work in the reference holding state is stopped at the first reference position RT of the hand and the first reference posture T of the hand, and the first image of the work is captured by the camera. The first reference result Gs and the first detection result, which is the first reference posture θs of the work, are stored in the memory, and the second reference of the hand is set so that the work is installed at the desired setting position and in the desired posture. The position R0 and the second reference attitude 0 are stored in the memory.

In the middle of the convey path of the conveyer for conveying the work in the horizontal conveying direction, a means for determining the posture of the work on the conveyer is provided and a detecting means for detecting the work is provided. When the workpiece conveyed by the robot is roughly postured by the posture determining means and detected by the detection means, the robot grasps the workpiece and the fixed position on the other side of the conveyor. The work is brought by the robot into the field of view facing the one side of the two-dimensional camera provided in the. That is, it is possible to set the workpiece, which has been conveyed to a rough position by a conveyor, at the desired setting position and in the desired posture, and in order to enable this, the work is conveyed to the rough position. In a state in which the hand has been grasped by the hand, the hand is brought to the first reference position RT of the hand and at the first reference position T of the hand, stopped, and imaged by the camera to be the position G of the work. The posture θ is detected, and the position RD after the hand is corrected so that the work is set at the desired setting position and in the desired posture.
And the post-correction posture D of the hand to calculate the corrected position RD and the post-correction posture D,
Position your hands.

The means for determining the posture in which the workpiece on the conveyor is roughly set in advance is a partition member 120 in an embodiment described later, and as another embodiment of the present invention, instead of the partition member 120, another structure is used. Then, the work on the conveyor may be set in a predetermined rough posture.

Embodiment FIG. 1 is a perspective view of an embodiment of the present invention. Conveyor
The workpiece 102 is conveyed by the 100 in the horizontal conveying direction 101 at an arbitrary position and posture. The work 102 transported by this transport means 100 uses an industrial robot 105,
The work stage 106 is accurately and accurately positioned at a predetermined setting position 107. In order to facilitate gripping the workpiece 102 conveyed by the conveying means 100 by the hand 115 of the robot 105, this conveyor 100 has
A partition member 120 for roughly determining the posture of the work 102 is provided. This partition member 120 is
The passage vibration 121 is configured to become narrower toward the downstream side of 1. As a result, the work 102 is guided by the partition member 120, and a rough posture is determined.

A detection unit 122 including a combination of a light source and a light receiving element is provided on the downstream side of the partition member 120. Detection means 122
When the workpiece 102 is detected by the processing circuit 124,
The robot 105 is controlled to grasp the work 102 shown at the position 102a by the hand 115. This robot 105 is in position
After gripping the work 102 brought to 102a, the hand 115 is placed in a predetermined position in a predetermined position so that the work 102 enters the visual field 111 of the television camera 110, which is a two-dimensional camera provided at a fixed position. Bring it and stop it.
In FIG. 1, at a predetermined position indicated by reference numeral 102b,
The state where the gripped work 102 is brought is shown. The work thus moved to the predetermined position 102b in the field of view 111 is imaged by the camera 110, and the position and orientation of the centroid of its outer shape is calculated and detected by image processing. The detected position and orientation of the centroid is compared with the reference position and orientation of the centroid of the workpiece that has been obtained in advance. Based on this comparison result, the deviation amount with respect to the set position 107 of the work to be positioned in the reference gripping state where the reference position and orientation of the center of the work are obtained is calculated by correction, and the work is accurately located at the set position 107. It can be positioned and moved with high accuracy. The above-mentioned background plate 123
Is a light-shielding plate for blocking ambient light from behind and making it easy to capture the shape of the work 102 when the TV camera 110 images the work at the position 102b.

Before work, attach the work 102 to the hand 115 of the robot 105.
To hold the field of view 111
The hand 115 is moved to a predetermined reference position RT of the hand 115 and a predetermined reference posture T of the hand 115 so that the work comes inside, and then stopped, and an image is picked up by the camera 110. The reference position Gs and the reference attitude θs of the centroid (in the following description, sometimes referred to as the centroid instead of the centroid) are detected, and the first detection result must be stored in the memory of the processing circuit 124. The operation at this time will be described with reference to FIG. The process moves from step a1 to step a2, and the work 102 is moved to the supply position 102a. Step a3
Then, the work 102 is gripped by the hand 115 of the robot 105. The hand 115 may be a vacuum suction pad for gripping the work 102, or may be a structure for gripping the work 102 with a plurality of grip claws.
When gripping the work 102 with this hand 115,
102 has been grasped by the hand 115 in the reference grasping state,
Position Rc of robot 115 in robot coordinate system at this time
(X, y, z) and posture c are stored and stored.

At step a4, the hand 115 is moved so that the work 102 is within the visual field 111 of the television camera 110, and stopped at T at a predetermined reference position RT at a predetermined reference position RT of the hand 115,
In this way, the work 102 is moved to the predetermined position 102b, and the position RT (x, y, z) and the posture T of the hand 115 of the robot 105 at this time in the robot coordinate system are stored.

At the next step a5, the image of the work 102 is picked up and the image processing is performed to determine the barycentric position of the work in the camera coordinate system.
Gs (x, y) and the posture θs are detected. This position Gs (x,
y) is used as a reference position of the workpiece 102 as described later, and the posture θs is used as a reference posture of the workpiece 102 as described later.

At step a6, the reference position Gs of the center of gravity and the reference posture θs in the camera coordinate system of the television camera 110 are converted into the reference position Hs (x, y, z) and the reference posture s of the center of gravity of the robot coordinate system. Store. The correspondence between the camera coordinate system and the robot coordinate system will be further described. Since the TV camera 110 captures the work 10 in a two-dimensional manner, the coordinate values of points on the two-dimensional plane in the field of view 111 of the TV camera 110 are determined by the robot. It is necessary to convert to coordinate values in a three-dimensional space in the coordinate system. For this correspondence, the following three coordinate systems are used as coordinate systems, that is, the robot coordinate system O.
−XYZ, vision coordinate system o _v −x _v y _v , camera coordinate system o _c −x _c y
There is _c , and the relationship is shown in FIG.

The camera coordinate system is defined on a two-dimensional plane that is perpendicular to the optical axis of the camera 110 and at a certain distance, and
The vertical and horizontal directions of the TV screen directly correspond to the camera coordinate system x _c axis and y _c axis.

The vision coordinate system is defined on the plane of the camera coordinate system, and the coordinate axes need not be defined parallel to the camera coordinate system.

When the camera coordinate system and the vision coordinate system have the relationship shown in FIG. 4, they can be converted by the following equation.

x _c = x _v · cos θ−y _v · sin θ + a (1) y _c = x _v · sin θ + y _v · cos θ + b (2) where θ is the rotation angle of the axis, and (a, b) is the shift amount of the origin. To do.

In FIG. 3, the vision coordinate system o in the camera field of view 111
_v −x _v y _v and x separated by a length L (absolute value) from the origin o _v
Place a paper that draws a point x1 on the axis and place it in the camera field of view.

In order to define the vision coordinate system, the cursor is moved to the origin o, the point x1, and any one point p1 on the paper while observing the screen imaged by the TV camera, and each coordinate is stored.
Thereby, the rotation angle θ and the shift amount (a, b) of the origin in the first and second equations are defined.

In order to define the robot coordinate system, the robot's hand (working point) is set to the origin o _v , point x1, and any one point p2 on the paper.
And the three-dimensional coordinate value of each point viewed from the robot coordinate system is stored.

By this, the two-dimensional plane of the vision coordinate system seen from the robot coordinate system, the origin, and the x _v axis and y _v axis can be defined. Moreover, since the point x1 that is separated from the origin by L is taught, the scale conversion of the length can be performed.

As described above, the position and orientation of the work 102 captured by the TV camera 110 can be converted into values in the robot coordinate system, and the work 102 can be handled.

In step a7, the workpiece 102 is moved by the hand 115 of the robot 105 in the reference gripping state, the workpiece 102 is moved, and the workpiece 102 is preset to a predetermined desired setting position 107 of the workpiece 102 on the work stage 106. The robot moves and positions in the desired posture, and the reference position Ro of the hand 115 in the robot coordinate system of the hand 115 of the robot 105 at that time
(X, y, z) and its reference posture o are stored, and the series of operations is completed in step a8.

At the time of work, the operation shown in FIG. 5 is performed. Moving from step b1 to step b2, the work 102 is moved to the approximate supply position 102a by the conveyor 100, and step b3
Then, position the hand 115 of the robot 105 at the position Rc (x, y, z), and position the hand 115 at c.
In this state, the work 102 on the supply device 102a is gripped.

At step b4, set the position of the hand 115 of the robot 105 to RT
Position it at (x, y, z) and set the posture of the tip 115 to φT.
, And thus bring the workpiece 102 to position 102b within field of view 111 of camera 110. Therefore, in step b5, the work 102 is imaged using the camera 110, and the gravity center position G (x, y) and the posture θ of the image taken by the work 102 in the camera coordinate system of the television camera 110 are calculated and detected.

Therefore, in step b6, the center of gravity position H of the work 102 detected by the TV camera 110 in the robot coordinate system is the coordinates G (x, y) of the center of gravity position in the TV camera coordinate system and the posture θ.
Convert to (x, y, z) and its posture.

Next, at step b7, the hand 115 of the robot 105 is
The position RD (x, y, z) after correction of the hand 115 when setting and positioning the work 102 on the work stage 106 in the robot coordinate system is calculated, and its posture D is calculated, and after the correction obtained in this way The hand 115 is positioned at the position RD (x, y, z) and the posture D, and the work 102 is accurately positioned at the desired setting position 107 in the desired posture.

Here, RD (x, y, z) after correction and posture D after the correction
Is calculated by the third and fourth equations.

RD (x, y, z) = Ro (x, y, z) -H (x, y, z) + Hs (x, y, z) (3) D = o- + s (4) And the position of the transport means 100 is approximately 1
It becomes possible to position the workpiece 102 conveyed to 02a at the set position 107 with high accuracy.

Step a5 of FIG. 2 and step b5 of FIG. 5 described above.
In order to detect the position and orientation of the center of gravity by imaging the work 102 with the TV camera 110, if there are scratches or holes on the surface of the work, it becomes a noise component when capturing the outer shape of the work 102. , The position and orientation of the center of gravity may not be accurately detected. If there are scratches or holes on the surface of the work 102, delicate shadows are formed on the work 102 at these portions, and these shadows change variously depending on the position and orientation of the work 102. Therefore, the binary image obtained by binarizing the input image often has different shapes at the scratches and holes depending on the position and orientation of the work 102. Since the position and orientation of the center of gravity are calculated from the coordinate values of all pixels whose values are logical “1” in the binary image, the values of the center of gravity position and orientation may change in the work 102 having a scratch or a hole. is there.

Therefore, even if the surface of the work 102 has scratches or holes, it is necessary to remove the noise component and capture only the outer shape of the work 102. As one method for that purpose, the processing circuit 124 performs the operation shown in FIG. Step d1 is shifted to step d2, the image signal picked up by the television camera 110 is input, and step d
At 3, in response to the image signal, a secondary image of the work 102 is obtained as shown in FIG. 7 (1). In step d4, i = 1 is set, and in the next step d5, one pixel enlargement processing is performed. That is, each pixel shown by the circle in FIG. 8 (1) is made to be the pixel added by being enlarged by the dashed circle in FIG. 8 (2) by the one-pixel enlargement processing. The pixel enlargement process is performed in this manner. Accordingly, the position image of the work 102 shown by the broken line in FIG. 7 (2) is enlarged as shown by the solid line in FIG. 7 (2), and at this time, scratches or holes on the surface of the work 102 are generated. The image of the work 102 in which the noise is removed is obtained because the part of is filled up. At this time, if the scratches or holes are not filled by the 1-pixel enlargement process, the 1-pixel enlargement process may be performed up to an appropriate number of times n times. In step d6, such 1
It is determined whether the pixel enlargement processing has been performed n times for all the pixels. After the one-pixel enlargement process in step d5 described above is performed for all the pixels, the process proceeds to step d7, i = 1, and this time, one-pixel reduction process is performed in step d8. That is, as shown in FIG. 7 (3), the outer shape of the image of the work 102 is returned to the original size, and at this time, the scratches and the holes remain filled as described above, and are not included in this image. Absent. When the reduction processing is performed the same number of times as the enlargement processing, n times, in step d9, the process proceeds to step d10, and the imaging calculation processing of the outer shape of the workpiece 102 without noise components due to scratches or holes is completed. As described above, the position and orientation of the center of gravity are obtained based on the image of the work 102 from which the noise components have been removed.

Further, as another method of obtaining the position and orientation of the center of gravity from the outer shape, there is a method of removing noise components from the contour line image. That is, from the binary image of FIG. 9 (1), a contour line image including a scratch or a hole as shown in FIG. 9 (2) is obtained. Of these, those contour lines whose length is less than or equal to the threshold value are ignored, and a contour line image having only the outer shape as shown in FIG. 9C is obtained.

In FIG. 9 (3), the center of gravity and the posture are determined by finding the portion sandwiched between the pixels Pi, m and the pixels Pi, n that form the contour line in the horizontal direction from the upper end to the lower end of the image. Can be calculated.

As another embodiment of the present invention, the positions and orientations of other characteristic points may be obtained instead of the center of gravity.

The robot 105 is arranged on one side of the conveyor 100, and the camera 110 is arranged at a fixed position on the other side of the conveyor 100 and has a field of view 111 facing the one side. This is clear from FIG.

EFFECTS OF THE INVENTION According to the present invention, a work is conveyed by a conveyor in a horizontal conveyance direction, and a means for determining a posture of the work on the conveyor in a predetermined rough posture is provided in the middle of the conveyance path. Further, a detection means for detecting the work is provided further downstream thereof, and when the work is detected by the detection means, the work on the conveyor which is roughly in the posture by the means for determining the posture is moved to the robot. Since the work is gripped by the hand, the excellent effect that the work can be easily and reliably gripped by the robot and the correction calculation can be further easily performed is achieved.

Furthermore, according to the present invention, a two-dimensional camera is arranged on the other side of the conveyor, and the field of view of this camera faces the one side on which the robot is arranged, and therefore, the hand of the robot. There is also no risk of contact with the camera during movement of the robot and interference, and the effect of being able to reliably grasp and move the work by the hand of the robot and bring it to the desired position accurately is also achieved.

Further, according to the present invention, even if the work is carried by the carrying means to the predetermined rough position, the excellent effect that the work can be positioned with high accuracy as described above is achieved. Of. In general, a robot has very high repeatability (repeatability), but it cannot be said that NC-like accuracy (that is, absolute accuracy) is very high. However, in the present invention, such repeatability of robot repeatability is high. This is a method that skillfully utilizes the fact that the workability is high, and by using this, it is possible to position the work by using the robot with absolute accuracy as high accuracy. Even if the position of the work conveyed by the conveying means is not inaccurately high precision, the work can be accurately and accurately positioned at the set position by using the robot.

Therefore, it is not necessary to convey the work with high precision by the conveying means, and the constitution of the conveying means can be simplified. In addition, even if the work is shaken at the moment when the work is gripped by the robot, as described above,
According to the present invention, it is possible to position the work with high accuracy as described above even if the work is carried to a predetermined approximate position by the carrying means. The excellent effect is achieved. In general, a robot has very high repeatability (repeatability), but it cannot be said that NC-like accuracy (that is, absolute accuracy) is very high. However, in the present invention, such repeatability of robot repeatability is high. This is a method that skillfully utilizes the fact that the workability is high, and by using this, it is possible to position the work by using the robot with absolute accuracy as high accuracy.

[Brief description of drawings]

FIG. 1 is an overall perspective view of an embodiment of the present invention, FIG. 2 is a flow chart for explaining an operation for storing and setting a reference position and posture before work, and FIG. 3 is a robot coordinate. System, camera coordinate system, and vision coordinate system, FIG. 4 is a diagram showing the relationship between the camera coordinate system and the vision coordinate system, and FIG. 5 is a flow chart for explaining the operation during work. FIG. 6 is a flow chart for explaining the operation for removing the noise component of the image of the work 102, and FIG. 7 is a view showing the image of the work 102 processed in the operation shown in FIG. 6, FIG. FIG. 6 is a diagram for explaining the one-pixel enlargement process of step d5 in FIG. 6, and FIG. 9 is a diagram for explaining the procedure for removing the noise component from the contour line image according to another embodiment of the present invention. 100 …… Transport means, 102 …… Work, 105 …… Robot, 1
07 …… Set position, 110 …… TV camera, 111 …… Field of view,
115 …… Minions

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication location G05D 3/12 K (72) Inventor Katsuya Miura 1-1 Kawasaki-cho, Akashi-shi, Hyogo Kawasaki Heavy Industries Ltd. Company Akashi factory (56) Reference JP-A-63-115205 (JP, A)

Claims (1)

[Claims] 1. A means for determining a posture of a work on a conveyor in a predetermined rough posture is provided in the middle of a convey path of a conveyer which has a horizontal conveying direction and conveys the work. The detection means for detecting the work is provided on the downstream side of the conveyor in the conveying direction from the means for determining the posture, and the work is detected by the detection means, whereby the hand of the robot arranged on one side of the conveyor. A two-dimensional camera is arranged at a fixed position on the other side of the conveyor downstream of the detecting means, and the camera has a field of view facing the one side. A method of positioning a work, in which the work is gripped by a robot's hand in a reference gripping state, and the hand is set so that the work is within the visual field of the camera. The work that is brought to the position RT and stops at the predetermined first reference posture T of the hand and stopped, and the work held by the stopped hand in the reference holding state is imaged by the camera, and the first work of the work is performed. 1
The first detection result, which is the reference position Gs and the first reference posture θs of the workpiece, is stored in the memory, and the workpiece stopped in this reference gripping state is at the desired setting position of the workpiece and the desired posture of the workpiece. The hand is moved and positioned so as to be installed in the memory, and the predetermined second reference position R0 of the hand and the second predetermined reference posture 0 of the hand at that time are stored in the memory, and the workpiece is conveyed by the conveyor. The workpiece is conveyed, and the workpiece is placed on the conveyor in a rough posture predetermined by the means for determining the posture, the conveyed workpiece is detected, and the detected workpiece is grasped by the robot's fingertips to grasp the fingertips. The hand is brought to the first reference position RT and is stopped at the first reference posture T of the hand and stopped, and the work in the field of view grasped by the stopped hand is imaged by the camera, Wa The second detection result, which is the position G and the posture θ of the workpiece, is obtained, and based on the first and second detection results, the hand is placed so as to be installed at the desired setting position and in the desired posture. Position RD after correction by correcting the second reference position R0 of
Is calculated and the second reference posture 0 of the hand is corrected to calculate the corrected posture D, and the hand is positioned at the corrected position RD and the corrected posture D. Characteristic workpiece positioning method.