JPH0299802A - Setting method of coordinate system in visual sensor using hand eye - Google Patents

Abstract

PURPOSE:To reduce an error between a robot coordinate system and a coordinate system of a visual sensor and to improve the precision in correction of position by a method wherein the two coordinate systems used for a robot control are set on the basis of one point on a jig fitted to a robot. CONSTITUTION:A jig 9 showing a point P of which the coordinate position in a base coordinate system is known is fitted to a robot 1, a tool center point is positioned at the point P and taught, the origin of the base coordinate system is shifted in parallel to the taught point, with the taught point used as the origin, and a coordinate system thus obtained is set as a robot coordinate system. Next, a camera 8 fitted to a hand 7 is positioned at an image intake position and made to recognize the point P, and this is set as the origin of a visual sensor coordinate system. Subsequently, the hand 7 (camera 8) is moved in parallel to one coordinate axis being vertical to the optical axis of the camera 8 and passing through the point P, and a visual sensor is made to recognize the point P. Then the visual sensor coordinate system and the length for one picture element are set on the basis of a distance of shift between the recognized point and the origin of the visual sensor coordinate system, the direction of shift thereof and the direction of the hand 7.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a method for setting a coordinate system for a visual sensor, and in particular, a method for setting a coordinate system for a visual sensor in which a camera is attached to a robot hand. Regarding the method.

BACKGROUND OF THE INVENTION To set a rectangular coordinate system in a three-dimensional space, it is necessary to determine at least three points on the three-dimensional spaceman. Normally, in order to set the /jth portrait limb reference system (referred to as the ml-coordinate system) for controlling robot 1, for example,
Origin of the orthogonal coordinate system ([Kobot] - position relative to the base coordinate system set in the mechanism), ■ [1 point in the positive direction on the X axis of the orthogonal coordinate system, ■ XY plane of the robot orthogonal coordinate system By teaching three points, such as an arbitrary point in the Y-axis end direction in 1, a rectangular coordinate system has been set. Also, the distance between the visual sensor camera and the detected object,
That is, it is also necessary to set the image capture position.

Similarly, when setting the visual Ron 1 no coordinate system,
For example, ■ the origin of the visual sense coordinate system, ■ 11 points in one direction of the visual sense coordinate system (X 1lIl), ■ 6 points on the XY plane of the visual sense 1 coordinate system + any point in the normal direction. It is necessary to teach three points, and one frame memory that stores images captured by the visual camera.
The length per pixel must be determined (). Then, the robot coordinate system set in this way is made to match the coordinate system of the visual sensor 1, and the position recognized by the visual sensor in the coordinate system of the visual sensor is
) must be defined as the coordinate system 61.9 (J).

Conventionally, such a robot coordinate system and visual position +J [
When setting the reference system, as described above, [the origin position of the 1-bot Cartesian coordinate system, the point in the normal direction on one axis, for example, the X-axis, and YIl! of the XY plane! l14T: Prepare a jig that indicates three arbitrary points in the direction, position the robot's tool center point (TCP) at each of these three points, set the robot coordinate system by teaching these three points,
After that, move the hand to the above image capture position, and
The points are captured with a camera, and a visual coordinate system is established from the three recognized points. 1 In this way, the coordinate system J: is set at the same three points, so the robot coordinate system and the visual sensory coordinate system match.

Problems to be Solved by the Invention As mentioned above, in the conventional method, a jig that teaches three points is used to set the coordinate system of the robot (I) that controls the robot, and the coordinate system used by the visual sensor. The setting of the coordinate system, that is, the operation of the 4-brasion X term [1], which was not efficient, and the However, there was a drawback in that the accuracy of position correction was greatly affected.

Therefore, it is an object of the present invention to set the coordinate system of the D-box and the visual sensor in the cylinder. The object of the present invention is to provide a method for setting a coordinate system on a visual sensor using a hand-eye, which can be moved around.

Means for Solving the Problems In the present invention, in order to solve the above problems,
Attach a jig to the robot that hits one point in one row at the predetermined coordinate position in the base 8 reference system set in [ ), and teach the 11th point of the jig to the 11th box 1-. Then, an orthogonal coordinate system obtained by shifting the base coordinate system with the teaching point as the origin is set as the coordinate system for the robot 1. Next, move the robot to the image capture position and visually detect one point on the jig.
(set it as the origin of the visual coordinate system, and move the robot hand parallel to one coordinate axis of the base coordinate system that is perpendicular to the optical axis of the camera and passes through one point on the jig) One point on the jig is recognized by the visual sensor. The coordinate system and length per pixel of the pixel can now be set.

Position a point on the jig whose coordinate position is known in the operation base coordinate system using the numerical value as the base position in J5 in the base coordinate system, or manually position the tool center point [. By teaching, the teaching point is set as the origin,
Parallel to each axis of the base coordinate system<'j, that is, the coordinate system obtained by moving the origin of the base coordinate system parallel to the teaching point is set as the pot coordinate system to control [1 pot 1~].

Then, while avoiding moving the robot, place the camera held in the hand at the set image capturing position; Set as the origin of the system, then move the n-bot hand parallel to one coordinate axis of the base coordinate system (robot coordinate system) that is perpendicular to the optical axis of the camera and passes through the above one point of 1"- Therefore, when the camera is moved and the above point on the jig is recognized from the visual sensor to
A visual coordinate system is set based on the orientation of the bottom hand.

In other words, from one point on the jig at the visual center recognized as the origin of the visual center coordinate system, the number of pixels in the frame memory between the points on the jig recognized by moving the camera is Then, the length per image system is determined by the moving distance of the robot hand, and the line connecting the two points recognized by the visual center is determined as one coordinate axis of the visual center coordinate system, and this coordinate axis and The coordinate axes of the other two visual centers can be determined by the moving direction of the I robot hand (camera), and the direction of the coordinate axes is determined by the orientation of the tool.

For example, after determining the origin of the visual center, if the robot hand (camera) is moved in the X-axis direction of the base coordinate system (robot coordinate system) and the above point on the jig is recognized, the recognition accuracy is It is a point on the plus X axis of the visual sensor coordinate system, and the Y and Z axes of other vision center coordinate systems are lines parallel to the Y and Z axes of the C base coordinate system (robot coordinate system) perpendicular to this X axis. In addition, the positive directions of the Y-axis and Z-axis can also be determined based on the positive direction of the X-axis and the orientation of the tool.

Embodiment FIG. 2 is a schematic diagram of a robot 1- that implements the coordinate system setting method d1 of the present invention, in which a boss 2 rotatable within a predetermined range is supported on an embossing table 1, and the bosses 1 to 2 A manually operated support shaft 3 is inserted through the support shaft 3, and a first arm 4. is supported horizontally on the support shaft 3. At the tip of the first arm, a second arm t\5 that rotates in the horizontal direction is pivotally supported, and at the tip of the second arm 5, a wrist 6 that rotates in the water XP force direction is pivotally supported. A visual sensor camera 8 is taken from the wrist 6 → the taken hand 7
is taken f・]().

When setting the 11-point coordinate system and center coordinate system for controlling the robot with respect to the base coordinate system (X-Y-, 7) set for the robot mechanism of the robot, immediately When adjusting the origin and performing calibration, the base coordinate system < x −, y −
Indicates a point P whose coordinate position at z> is known;
1. Attach it to the 1-bottom 1-, and perform the 1-shot rebrasion using the jig 9.

Madi, 1" Bot 1~ tool center point 1~ ("CP
) is positioned at one point P of the γh tool 9, [is taught at the origin of the one-bot coordinate system, and stored within the memory of the one-bot control device. If the seat position of the point P on this jig in the base coordinate system is (x O, y O, z O), then the coordinate position (x O, y O, 70) is the origin, and the base A coordinate system whose coordinate axes are parallel to each axis of the coordinate system and in the same direction is set as a robot coordinate system.

That is, the coordinate position (xO, yO,
zO) l, m shift L, ri Cartesian coordinate system (X'Y'-Z
') is set as the 1bot coordinate system, 1. Furthermore, the outer coordinate of the base coordinate system of the point P on the Jukijiman is ``+;'i (x
O,y 0.70H;1. Since there is a knowledge T, the coordinate position may be input numerically into the control device to set the robot FP reference system with the input seat position as the origin. Next, move the robot hand to the set image capture position 1. In this example, move the robot hand in the Z-axis plus h direction of the base coordinate system, position it at the image capture position, and use the camera 8 with a visual axis of 4. The point P on the jig 9 is layered, and the point in the frame memory 10 of the visual center of the nuclear heat P is stored in the f-tameter of the visual center as the origin of the visual axis coordinate system.

Figure 1 shows the base coordinate system (X-Y-7), n box]~
This is a diagram explaining the relationship between the coordinate system (X'-Y'-Z') and the visual center coordinate system. point) is taught, the robot coordinate system (X'-Y'Z') is determined, the nuclear heat P is imaged by the camera 8'cIIIz, and the nuclear heat is located at the point (sxo, 5yo) on the frame. , the position (sxo, 5yO) is stored as the origin of the visual center coordinate system.

Then, the robot hand 7 (camera 8) is moved parallel to one axis of the base coordinate system ([1 point 1 ~ coordinate system)] perpendicular to the optical axis of the camera. In this example, the movement is made parallel to the X-axis of the base coordinate system in the direction of the Meibus force, and the point P of the above-mentioned Yoshikai is moved as much as possible to a position that can be captured in the field of view of the visual sensor (J: will be described later). The viscosity of the quality per pixel in the sensor coordinate system is 1.! (reducing the difference).In the base reference system, the tool center point 1 ~ (camera) is set to P (x O ,y O,z
The field of view of the visual sensor moves from S to S' as shown in Fig. s' xo
, s' yO), then the amount of movement (XiXO) of the 11 tool center points from point P to Pl and the point P on the above jig above the frame
, that is, from the point (sxO, 5yo) to the point (s'x
O.

The length of one plain wood is determined by the number of pixels between s' yO). Then, Sada (sxo, 5yO) and Dot (s'
xo , s' yO) /6 The connecting line is recognized as the X-axis of the visual center coordinate system, and the point (SXOsyo)
The h direction from the point (S' XO, S' yO) to the point (S' Direction of Tli to axis <-i Y-axis direction, Z
The direction of the axis is also determined, and the visual sensor system is determined.

FIG.
``This is the eastern 11th diagram of the f20 and the control unit 40 of the visual sensor υ.
P LJ) 21 contains a ROM 22 in which a control program is stored, and a ROM 21 in which a control program is stored.
4 RΔM23 used for storage etc. Non-volatile memory 2/I, c+bot 1- in which teaching data and various setting values are stored
An axis controller 25 connected to the one-no-one-voice circuit of the one-nobodies driving each axis of the one-nobodies. The teaching operation 5f82G and the communication interface of the operation panel 271 are connected, and the CPU of the visual center 40 is connected. 1.1 by t, L lotus/19 -
) A room memory 4 that stores images captured by the camera 8
2. Controller 1-Roll software method 43. l Program memory 44. '7-Tameshiri 45. Camera interface 469 connected to camera 8 Image processing block 1/4
7 and communication (5 InternoAce 48 are connected, and communication interface 28.48 connects
0 and the visual sensor→) The control unit 40 sends and receives information.

Note that the robot control device 20. Visual sensor control unit 4
The configuration of 0 is well known and detailed explanation is omitted.IJ
Ru.

Therefore, by operating the teaching operation f1 board 26, [position the J hot tool center point 1- at the point P of the jig 90,
The CPU 21 stores the teaching point in the non-volatile memory 24 as the origin of the 1bot coordinate system, and sets the origin of the base coordinate system (X-Y-Z) to this teaching point. Set the coordinate system shifted to the point as 1" as the bot's value. Next, by inputting a visual sensor coordinate system setting command from the teaching operation panel 26, the IZ1 button 1-control device 20. Vision sensor 4 control unit 40 CI) U21, 41
is shown in Figure 4(a).

([)) to open the acceptance process kfi.

Robot control 'The CPU 21 of the JA lock 20 moves [1-bot 1-] to the set image capture command position (Stellab 1
00). That is, the robot 1 is moved in the positive direction of the /axis from the position i5 where the tool control point 1~ is positioned at Jj, (''), and the image is taken in l'/W (in 1). '/'II' and outputs an image capture command to the visual sensor control unit 40 via the communication interface 28.'18 (step 101).C of the visual sensor 1 control unit 40
PtJ41 receives this signal (step 200),
A stab command is output to the C camera 8 via the camera interface 4G, and the projected image is processed by the image processing processor 4.
7 to perform grayscale gradation processing and store frame memory 4.
Then, using the image processing block 1J47, the jig-[
Frames of point P = bottom position @ (sxo, 5yo)
(step 201), this position (sxo, 5y
data memory 4 with o) as the origin of the visual sensor 1 coordinate system.
5. Step 202), work completion No. 411 is output (Step 1" 203). Robot control device 2
0's CPU 21 receives this signal (Step 102)
, set the register R provided in the RAM 23 to +'0.1, and set the +-1 pot to the predetermined - set in the minus force direction of the X-axis of the base coordinate system.
) l is moved (step 104). Then, register R receives the above predetermined amount j! Then, an image capture command is output (step 106). The CPU 41 of the visual sensor control unit 40 receives this image capture command,
As described above, the image from the camera 8 is stored in the frame memory 42 (steps 204 and 205), and the image processing processor 47 judges whether or not the point P on the jig has been recognized. If so, its recognized coordinate position (s' x
o, s' yO) in the coordinate value storage register in the pig memory 45, and the stored coordinate position is newly updated in step 20.
6, 207), output a recognition signal (step 208)
. Note that if the point P on the jig cannot be recognized in step 206, a non-recognition signal is output (step 209).

The CPU 21 of the control device 20 determines whether a recognition or non-recognition signal has been received (step 107), determines whether the received signal is a recognition signal (step 108), and determines whether the recognition signal is a recognition signal. If so, return to step 104 and return to the predetermined +ji 1 again (move the robot in the minus h direction on the X-axis of the Noves coordinate system, and set the predetermined ris in the register R.
1 and output the image capture command in step 105.
, 106), the CPU 41 of the visual sensor control unit 40 performs the processes of steps 204 to 208 described above. Hereinafter, while the point P on the jig is in the field of view of the camera and is recognized in the frame memory 42, the robot control head 20
The CPU 21 performs the processing of steps 10/l to 108,
The CPU 41 of the visual sensor control unit 40 performs steps 204-
208 is repeated] As a result, the coordinate value storage register in the data memory 45 contains the coordinate value () of the point P on the jig captured by the camera. s'xO1s' yO) will be sequentially updated (step 207).

In this way, as shown in FIG. When P is no longer recognized, the CPU/11 outputs a non-recognition signal to the robot control unit 20.
CPU 21 receives this signal (step 107).
, 108>, subtract a predetermined Q'+ 1 from register R)
From this, the amount of movement from the origin (origin of the robot coordinate system) that can capture the point P of the jig 1- in the field of view of the camera 8 is determined (step 109), and the amount of movement stored in this register R and the robot The direction of the vine, that is, whether the tool is facing downwards or upwards (whether the tool is facing toward the vertical axis h or the positive direction of the Z axis) is transmitted to the visual sensor control unit 40.

When the CPU 41 of the visual sensor control unit 40 receives the amount of movement and the direction of the tool (step 210), it executes step 2.
The origin (sxo, 5yo) of the visual sensor υ coordinate system stored in the data memory in step 02 and the movement of the robot in the Coordinate position of point P on the tool (s
The distance between 'xo, s' yO1, and the coordinate positions of the two (sxo, 5yo).

The amount of robot movement between (s' xo , s' yO),
That is, the length per pixel of the room meter 42 is determined from the amount of bot movement received in step 210, and is stored in the data memory 45. Then, the origin (Sxo, 5yO
), set the h direction of the coordinate position (s' xO, s' yO) to the X-axis plus 1J of the visual sensor coordinate system, and set the direction of the Y-axis that intersects 0 with this X-axis to the 7" las of the X-axis. The Z axis of the visual sensor coordinate system is determined from the direction and the orientation of the tool, and the visual axis system is set by making the Z axis of the visual sensor coordinate system the same as the Z axis and the direction of the base coordinate system. (Step 2
11).

No. [1-B [?] in Fig. 4 (a) and (b). The movements shown in -1~ are the 16 methods in which the robot manually teaches the origin of the coordinate system, but since the coordinate position of the base coordinate system of t3'') on the jig is known, the robot controller 20 It is decided to set (or program) the coordinate value F1 of this point P in advance.If a seat FA system setting command is input to the robot controller V♂20, the result shown in Fig. 4(a) will be obtained. Step 10
0, the seat C of the point P on the jig that is automatically set
It is also possible to move the tool inputs 1 to 1 to NV() and then perform the processing below 100.

Effects of the Invention The present invention allows the robot's coordinate system and visual sensor coordinate system used for robot control to be effectively transferred into the cylinder in a series of operations using one point on the jig. Since we set a coordinate system based on one point on the shell, we need 3 points like before.
Compared to the case of setting by points, the error between the robot coordinate system and the coordinate system of the visual sensor 1 is reduced, and the position can be corrected on the surface of Kasumaro.

[Brief explanation of the drawing]

Figure 1 4. L base coordinate system according to the present invention, [1
An explanatory diagram illustrating the relationship between the bot coordinate system and the visual sensor coordinate system. Figure 2 is a diagram showing an example of applying the present invention to 11 robots 1. Figure 3 is a diagram illustrating the robot d3 (") in the same example. Control device. The main parts of the visual sensor control section are shown in Figure 4 (a).
(b) is for setting the coordinate system.
51 Kakurenri control unit's imperial order) fil-'f-1 No 1 ~ 1゜7...Tool, 8...Camera, 9
. . . Jig, P . . . Jig soil point, 20 . . . I Kobot 1 to control device, /IO .

Claims (3)

[Claims] (1) In a visual sensor with a camera attached to the robot hand, attach a jig to the robot that gives one point at a predetermined coordinate position in the base coordinate system set on the robot, and point one point on the jig to the robot. The robot coordinate system is set as an orthogonal coordinate system obtained by shifting the base coordinate system with the teaching point as the origin, and then the robot is moved to the image capture position and one point on the jig is set as the robot coordinate system. is recognized by the visual sensor and set as the origin of the visual sensor coordinate system, and the robot hand is moved parallel to one coordinate axis of the base coordinate system that is perpendicular to the optical axis of the camera and passes through one point on the jig. The visual sensor is made to recognize one point on the jig, and the coordinate system of the visual sensor and the number of points per pixel are calculated based on the moving distance between the recognized point and the origin of the visual sensor coordinate system, the moving direction, and the orientation of the robot hand. A method for setting a coordinate system in a visual sensor using a hand eye whose length is set. (2) The origin of the robot coordinate system is taught at 1 on the jig.
2. A method for setting a coordinate system in a visual sensor using a hand-eye according to claim 1, wherein the method is performed by manually moving a tool center point to a point and teaching the coordinate system. (3) In the visual sensor using a hand-eye according to claim 1, wherein the teaching of the origin of the robot coordinate system is performed by numerically teaching the coordinate position of one point in the base coordinate system on the jig. How to set the coordinate system.