FANUC R-30iB Plus CONTROLLER, iRVision 2D Camera Application. OPERATOR'S MANUAL

Know-How

2. CAMERA DATA SETTING

MEMO

Even in the case of a 6-axis robot, the target position may not always be set

automatically in cases where the motion range is small or the camera's field of

view is at a corner of the motion range. In this case, you can set the target

position by the same steps as a 4- or 5-axis robot.

2.3.8.2

4- or 5-axis robot

For a 4- or 5-axis robot, the target position cannot be set automatically. Set the target position using the

following procedure.

Place the cursor over [Target Position] on the robot-generated grid calibration main screen and press

F3 [POSITION].

A detailed screen of target position will appear.

Press F2 [DISP IMG] and display Vision Runtime.

From the drawing, calculate the frame values as seen from the robot's wrist flange, and enter it as the

target position.

Enter '0' for WPR.

CAUTION

The target position should correspond to the model origin trained in section 2.3.5.

If the positions are different, the camera cannot be calibrated properly.

B-83914EN-2/01

193

Know-How

2. CAMERA DATA SETTING

Refer to Know-how Edition Subsection 2.3.8, [Target Position Measurement] and check that the

status of the [Target Position] of the selected robot-generated grid calibration data is [Measured].

Place the cursor over [Program Generation].

Press F2 [DISP IMG] and display Vision Runtime Page.

Enable the teach pendant and clear the alarm.

While holding down the [SHIFT] key, press F5 [RUN].

Auto generation of a program will be executed.

Keep the [SHIFT] key held down during measurement.

When the measurement finishes, the robot will stop, and a message saying 'Measurement is

successfully' will be displayed on the screen.

Release the [SHIFT] key and click F4 [OK].

If the previous program generation was interrupted midway, a message saying [Are you sure to resume?]

will be displayed if you try to execute program generation. To resume from where it was interrupted,

press F4 [RESUME] while holding down the [SHIFT] key. If you want to start it over from the

beginning, press F5 [RESTART] while holding down the [SHIFT] key.

B-83914EN-2/01

195

2. CAMERA DATA SETTING

Know-How

CAUTION

Measurement cannot be executed if the target robot-generated grid calibration

setup screen is being displayed. Close the teaching screen. How

measurement is going can be checked using Vision Runtime.

The way to limit the target displacement range

In order to avoid the interference with peripheral equipment, you can limit the target displacement range.

Open the camera data edit screen, and click [Train Model] in the navigation.

Click the pattern match tool in the tree view.

Make the search window smaller to exclude sections where you do not want the robot to operate.

Click [SAVE] to save the changes.

Click [END EDIT] to close the edit screen.

Regenerate the program.

Calibration Program

The generated calibration program is like the one shown below.

All the robot positions in the calibration

program are taught in the joint format.

UFRAME_NUM=2

UTOOL_NUM=2

3:L P[1] 1000mm/sec FINE

VISION CAMREA_CALIB 'CALIB1' REQUEST=1

5:L P[1001] 1000mm/sec FINE

CALL IRVBKLSH(1)

VISION CAMERA_CALIB 'CALIB1' REQUEST=1001

8:L P[1002] 1000mm/sec FINE

CALL IRVBKLSH(1)

10: VISION CAMERA_CALIB 'CALIB1' REQUEST=1002

(Repeat as many times as the number of points)

293:L P[2048] 1000mm/sec FINE

294: CALL IRVBKLSH(1)

295: VISION CAMERA_CALIB 'CALIB1' REQUEST=2048

296:L P[2049] 1000mm/sec FINE

297: CALL IRVBKLSH(1)

298: VISION CAMERA_CALIB 'CALIB1' REQUEST=2049

299:L P[2] 1000mm/sec FINE

300: VISION CAMERA_CALIB 'CALIB1' REQUEST=2

The section of the program that finds an individual calibration program consists of the three lines shown

below. This set of three lines is repeated in the middle of the calibration program above.

5:L P[1001] 1000mm/sec FINE

CALL IRVBKLSH(1)

VISION CAMERA_CALIB 'CALIB1' REQUEST=1001

Each command in the program is briefly explained below.

VISION CAMREA_CALIB 'CALIB1' REQUEST=1

196

B-83914EN-2/01

Know-How

2. CAMERA DATA SETTING

If you specify 1 in the request code of the CAMERA_CALIB command, all the calibration points in the

specified camera calibration are deleted. This is the first command to be executed in the calibration

program.

300: VISION CAMERA_CALIB 'CALIB1' REQUEST=2

If you specify 2 in the request code of the CAMERA_CALIB command, camera calibration data is

calculated using the found calibration points. This is the last command to be executed in the calibration

program.

VISION CAMERA_CALIB 'CALIB1' REQUEST=1001

If you specify 1000 or a larger value in the request code of the CAMERA_CALIB command, the program

attempts to find a calibration point. The value specified in the request code is recorded as the index of

the calibration point, along with the found position.

In an automatically generated calibration program, 1000 to 1999 represent the calibration points on

calibration plane 1, and 2000 to 2999 the calibration points on calibration plane 2. Note also that the

index of the position data of the preceding motion statement is the same as the request code that is passed

to the CAMERA_CALIB command.

Calibration points do not necessarily need to be found in the order of request codes. If a calibration

point is found twice with the same request code, the data of the calibration point that is found first is

overwritten by the data of the calibration point found later.

CALL IRVBKLSH(1)

If the KAREL program IRVBKLSH.PC is called, the robot performs an operation intended to remove the

backlash effect at its current position. As the argument, specify the motion group number of the robot

that performs the backlash removal operation.

B-83914EN-2/01

197

2. CAMERA DATA SETTING

Know-How

2.3.10 Executing Calibration Program

Select the generated calibration program in the SELECT menu, and play it back from the first line to

calibrate the camera. A robot moves the target as shown below.

Moving the target in a grid pattern

CAUTION

If running the program as is can cause interference, use lower override values.

In this case, execute the program while making sure that no interference occurs

during operation.

Each calibration point in the generated calibration program can be re-taught or deleted as necessary.

If there is any calibration point that causes the robot to interfere with peripheral equipment, re-teach that

point to move it to a position where it does not cause interference, or delete the calibration point. When

deleting a calibration point, delete not only the motion statement but also the lines of IRVBKLSH and the

CAMERA_CALIB command that are executed after the motion statement.

If there is any calibration point that hinders the robot operation because it is near singularity, re-teach that

point to move it to a position where it can avoid singularity, or delete the calibration point. When

deleting a calibration point, delete not only the motion statement but also the lines of IRVBKLSH and the

CAMERA_CALIB command that are executed after the motion statement.

198

B-83914EN-2/01

Know-How

2. CAMERA DATA SETTING

When re-teaching a calibration point, you may place the target closer to or further away from the camera

within a range in which the camera lens remains in focus.

CAUTION

The calibration program does not stop even if the target fails to be found or it is

found incorrectly during the program execution. After the program ends, open

the Robot-generated Grid Calibration setup page in the Vision Setup screen and

check to see if there is any point incorrectly found.

MEMO

As long as the position where the target is mounted remains unchanged, you can

re-calibrate the camera simply by executing the generated calibration program.

Once the program for calibration has been executed all the way to the end, the camera calibration is

complete. Check the calibration data and calibration points.

2.3.11 Checking Calibration Point

Check the calibration point found in Know-how Edition Subsection 2.3.5, "Calibration Setup".

If you click [Calibration Points] in the navigation area, the calibration points screen will appear.

If there are any found points that have been found by mistake, delete the irrelevant found points using the

following procedure.

Select an irrelevant found point from the found point list, or enter a found point number in the [Point

Number] text box.

Click the [DELETE] button.

The irrelevant found point will be deleted.

B-83914EN-2/01

199

Know-How

3. SETUP OF SNAP IN MOTION

3 SETUP OF SNAP IN MOTION

This function enables iRVision to snap an image without stopping robot motion, and is effective when

measuring a fixed frame offset with a robot-mounted camera and/or measuring a tool offset of a part held

by a robot with a fixed camera or a camera mounted on another robot. Using this function will reduce

robot cycle time compared to the legacy way.

3.1

OVERVIEW OF SNAP IN MOTION

This section gives an overview of the function snap-in-motion.

3.1.1

Features

This function enables iRVision to snap an image without stopping robot motion, and is effective when

measuring a fixed frame offset with a robot-mounted camera and/or measuring a tool offset of a part held

by a robot with a fixed camera or a camera mounted on another robot. The following vision processes

support this function.

•

2D Single-view Vision Process

•

2D Multi-view Vision Process

•

Depalletizing Vision Process

•

3D Tri-View Vision Process

You can calibrate cameras and teach vision processes as is conventionally.

Calibration of a camera and teaching vision processes are performed in the state where it stopped. Even

if it is SNAP IN MOTION, that is the same method with other applications.

Obtaining an accurate robot position at a snapping moment is a key technology to measure a fixed frame

offset of part with a robot-mounted camera or a tool offset of a part held by a robot with a fixed camera or

a camera mounted on another robot.

Conventionally an accurate robot position can be obtained only while a robot remains stationary. The

function snap-in-motion enables measuring a fixed frame offset and a tool offset by getting an accurate

robot position at a snapping moment.

This function has the following restrictions.

• The robot that holds a camera or a part should be controlled by the controller on which iRVision

resides.

• Only one robot can move at a snapping moment.

When performing tool offset with a robot-mounted camera, two robots are used (a robot holding a camera

and a robot holding a part), but the position during motion can be obtained for only one of them. Stop

the other robot when snapping.

iRVision assumes a robot that can move at a snapping moment in the following manner:

• If [This Controller] is selected for Robot Holding the Part, the robot that held a part can move.

• If a robot other than [This Controller] is selected for Robot Holding the Part and [This Controller] is

selected for Robot Holding the camera, the robot that holds a camera can move.

Above conditions are used in both cases that two robots are controlled by one controller and two

controllers.

B-83914EN-2/01

201

3. SETUP OF SNAP IN MOTION

Know-How

3.1.2

Using Snap in Motion

By default, the function snap-in-motion is disabled. To use this function, change the system variable

$VSMO_CFG.$ENABLE to TRUE.

$VSMO_CFG.$ENABLE = TRUE

By setting this system variable to true, the function snap-in-motion itself is enabled. But, to snap an

image actually without stopping robot motion, you need to modify your robot program. Refer to the

Know-how Edition Subsection “3.1.4 Robot Program for Snap in Motion” about how to modify robot

programs.

By enabling the function snap-in-motion, the method to obtain a robot position at the snapping moment is

changed internally. Basically, you can use camera calibrations, vision processes and robot programs that

you taught previously even after this function is enabled. However, potentially a quantity of errors may

be observed if you continue to use camera calibrations and vision processes that you have taught before

the function is enabled. In such a case, calibrate cameras and teach the reference positions again.

3.1.3

Checking Position and Speed at Snap

When the function snap-in-motion is enabled, the actual position and speed of the robot at the last

snapping moment are recorded in the following system variables. Refer to these variables to determine

the exposure time and so on in study for application described later.

$VSMO_VAL.$POSITION

It is the actual robot position at the snapping moment. It is in the Cartesian format. X, Y and Z are in

millimeters, and W, P and R are in degrees. For a robot-mounted camera, the position of the mechanical

interface relative to the

[Application UFrame] is recorded. For tool offset, the position of the

[Application UFrame] that you selected in the camera calibration setup page is recorded.

$VSMO_VAL.$SPEED

It is the actual robot speed at the snapping moment. X, Y and Z are in mm/sec, and W, P and R are in

degrees/sec.

3.1.4

Robot Program for Snap in Motion

For snap-in-motion, you need to execute the VISION RUN_FIND instruction in a different manner from

usual.

Snap after stopping

This is an example of a program for when snapping after stopping the robot, as in the conventional

procedure. The operation mode in line 2 is ‘FINE’. The robot's motion stops while snapping under the

‘VISION RUN_FIND’ command in line 3, and the motion to move to P[3] starts when snapping is

complete.

1: L P[1:start] 500mm/sec FINE

2: L P[2:snap] 500mm/sec FINE

VISION RUN_FIND 'A'

4: L P[3:stop] 500mm/sec FINE

202

B-83914EN-2/01

Know-How

3. SETUP OF SNAP IN MOTION

Snap in motion

To snap an image without stopping the robot, the VISION RUN_FIND instruction should be executed by

using the TIME BEFORE instruction. In the robot program below, ‘CNT100’ is specified for P[2] in the

second line, and the subprogram FIND.TP is called at the moment that the robot passes through P[2] by

using the TIME BEFORE instruction.

1: L P[1:start] 500mm/sec FINE

2: L P[2:snap] 500mm/sec CNT100 TB 0.00sec,CALL 'FIND'

3: L P[3:stop] 500mm/sec FINE

FIND.TP that is called by the TIME_BEFORE instruction is as below. The VISION RUN_FIND

function is executed in this subprogram.

VISION RUN_FIND 'A'

You can tweak the start time of the TIME BEFORE instruction, or you can use the TIME AFTER

instruction or the DISTANCE BEFORE instruction instead of the TIME BEFORE instruction. For

these instructions, refer to “OPERATOR’S MANUAL(Basic Operation) B-83284EN”.

If positions P[1] ~ P[3] in the above program are not on a line, the robot moves inside of the original path

and does not pass through P[2] because of CNT100. In such a case, modify P[2] so that the robot

actually passes through the expected snap position.

3.1.5

Notes

Vibration of robot arm while robot moving is not considered. The bigger the vibration is, the larger the

error is.

3.2

STUDY FOR APPLICATION

This section describes issues to consider when the function snap-in-motion is used.

3.2.1

Light and Exposure Time

If the function snap-in-motion is used, the camera or the target is moving during the exposure, so the

snapped image will be blurred to that extent. Because this blur can cause detection error, the exposure

time should be set a smaller value than usual to mitigate the blur.

Assuming that the robot moves in the direction at a right angle to the optical axis of the camera, the

amount of blur is calculated by the expression V × T × N ÷ S in pixels, where V is the velocity at the

snapping moment in mm/sec, T is the exposure time in sec, S is the size of the camera field of view in

millimeters and N is the number of effective pixels of the camera.

(Pixels) = V × T × N ÷ S

Determine the exposure time so that this amount of blur is smaller than one pixel and prepare a good light

source to get a fully bright image with the selected exposure time.

B-83914EN-2/01

203

3. SETUP OF SNAP IN MOTION

Know-How

3.2.2

Image Processing Time and Motion Time

When using the function to capture without stopping, it is necessary to pay attention to the relationship

between the time required for image processing and the time required for robot operation.

GET_OFFSET

The following program calls the subprogram FIND.TP, which include the VISION RUN_FIND

instruction, at the moment that the robot passes through P[2] in the second line, and tries to get the

resulting vision offset in the line 4. If P[2] and P[3] were too close and the robot reached P[3] in a short

time, the robot would wait for the completion of vision detection in the fourth line, resulting the robot

stopping. To avoid such a thing, consider a good layout of the camera so that the image processing

finishes while the robot is moving from P[2] to P[3].

1: L P[1:start] 500mm/sec CNT100

2: L P[2:snap] 500mm/sec CNT100 TB 0.00sec,CALL FIND

3: L P[3:stop] 500mm/sec CNT100

VISION GET_OFFSET 'A' VR[1] JMP,LBL[1]

5: L P[4:approach] 500mm/sec FINE VOFFSET,VR[1]

Continuous RUN_FINDs

In iRVision, vision detection that has been executed first has to be completed before the next vision

detection can be started. Therefore, if you start up vision detection consecutively in a short interval, the

vision detection that comes after will be kept waiting, and as a result, it will not be able to snap an image

at the expected position.

For example, in the following program, if the time to move from P[2] to P[3] is shorter than the time for

the image processing for FIND1.TP, snapping an image for FIND2.TP will be delayed.

1: L P[1:start] 500mm/sec CNT100

2: L P[2:snap1] 500mm/sec CNT100 TB 0.00sec,CALL FIND1

3: L P[3:snap2] 500mm/sec CNT100 TB 0.00sec,CALL FIND2

4: L P[4:stop] 500mm/sec CNT100

Time a vision process takes depends on the shape of the trained model pattern, the detection parameters,

the condition of the snapped image, and the load of the controller that performs the vision process.

Check your vision process time by actually executing the vision process.

In addition, if the vision process executed previously is configured to log images, the next vision process

can be kept waiting until the previous vision process completes image logging and snapping an image of

the next vision process can be delayed. In such a case, configure the previous vision process not to log

images. The delay of snapping an image can be shortened slightly if the ‘Enable Logging’ is checked

off in the iRVision configuration setup page.

3.2.3

Shift of Snap Position

Depending on the condition of the controller, timing of snap may be fluctuated slightly. The shift of the

snap timing is basically no problem, because iRVision can get an accurate robot position at snapping.

But if the size of the camera field of view is adjusted so that an image is filled with a part, the shift of the

snap timing lead the part being out of the camera field of view and that causes detection failure. Set the

size of the camera field of view to have enough margins to accept a quantity of the shift of the snap

timing. The size of margin depends on the speed of the robot at snapping. The faster the robot is

moving, the bigger the shift of the snapping position is.

Moreover, after turning on a controller, in the first vision detection, detection time may take a few

seconds longer. About the detection time of the second times, it becomes the normal detection speed.

At the detection of first time, when the shift of the snap timing occurs, perform the detection of first time

at somewhere else before the performing snap in motion.

204

B-83914EN-2/01

Know-How

3. SETUP OF SNAP IN MOTION

NOTE

Vision Processes are saved in the FROM or memory card. In addition, If Vision

Process is performed, it will remain in the cache in DRAM. Since access to the

file in cache is quick, the vision detection of second time becomes early rather

than the first time. Since Vision processes in the cache are deleted at the

turning off a control, the detection speed at first time after turning on a control

may become long. In addition, there is a limitation in the cache domain of

DRAM. So other Vision Processes are run, old Vision Processes in the cache

will be deleted from cache. In the system that runs many kinds of Vision

Processes in one cycle, some of the executed Vision Processes may not remain

in cache. The capacity of cache is 2 MB as an initial value.

3.3

SAMPLE APPLICATIONS

This section demonstrates three sample applications using the function snap-in-motion.

• Tool offset with a fixed-mounted camera (2D Single-view Vision Process)

• Tool offset with a fixed-mounted camera (2D Multi-view Vision Process)

• Fixed frame offset with a robot-mounted camera (3D Tri-view Vision Process)

In either application, generally the setup procedure and items to consider are the same as those of each

application without the function snap-in-motion. The descriptions about the area of overlap with those

in each vision processes are left out.

3.3.1

Tool Offset with a Fixed Camera (2D single-view vision

process)

Tool offset with a fixed camera (2-D Single-view vision process) measures the amount of tool offset by

snapping the workpiece that the robot is gripping, using a fixed camera. Using the measured tool offset,

the robot positions are compensated to put the part in a correct position. The tool offset is usually

measured with stopping the robot. Using the function snap-in-motion, the vision detection can be

executed while the robot is transporting a part without stopping robot motion.

The subsection “1.4.4 Robot Program Creation and Teaching” in the setup procedure described in the

section “1.4 SETUP FOR TOOL OFFSET WITH FIXED CAMERA” is different from those using the

function snap-in-motion.

‘Robot Program Creation and Teaching’ is explained below.

The figure below is an example of layout in the case of tool offset using 2D Single-view Vision Process

with a fixed-mounted camera.

B-83914EN-2/01

205

3. SETUP OF SNAP IN MOTION

Know-How

Camera

Compensation plane

Tool offset with a fixed camera (2-D Single-View Vision Process)

3.3.1.1 Robot program creation and teaching

This subsection demonstrates robot programs for tool offset in motion. Based on sample robot programs

below, create appropriate robot programs for your application.

The following two programs are created.

• Main robot program (MAIN.TP)

• Robot program to detect a target (FIND.TP)

MAIN.TP

This is the main program. In this program, the robot moves from the start position P[1] to the stop

position P[3] through the snap position P[2], and call the subprogram FIND.TP at the moment that the

robot passes through the snap position. After that, the VISION GET_OFFSET instruction is called to

get the resulting vision offset. The robot positions P[4] and P[5] are compensated with the vision offset

so that the part is put at the correct location.

UFRAME_NUM=1

UTOOL_NUM=1

4: L P[1:start] 500mm/sec CNT100

5: L P[2:snap] 500mm/sec CNT100 TB 0.00sec,CALL FIND

6: L P[3:stop] 500mm/sec CNT100

VISION GET_OFFSET 'A’ VR[1] JMP LBL[99]

8: L P[4:approach] 500mm/sec FINE VOFFSET,VR[1]

9: L P[5:set] 100mm/sec FINE VOFFSET,VR[1]

10: L P[4:approach] 100mm/sec FINE VOFFSET,VR[1]

11:

END

12:

LBL[99]

13:

UALM[1]

206

B-83914EN-2/01

Know-How

3. SETUP OF SNAP IN MOTION

FIND.TP

This is a program called by the TIME BEFORE instruction. This program executes the vision process

“A” using the VISION RUN_FIND instruction.

VISION RUN_FIND 'A'

3.3.2

Tool Offset with a Fixed Camera (2D multi-view vision

process)

Tool offset with a fixed camera (2-D Multi-View Vision Process) measures the amount of tool offset of a

workpiece by snapping multiple features on a large workpiece that the robot is gripping, using a fixed

camera. Using the measured tool offset, robot positions are compensated to put the part at a correct

position.

The tool offset is usually measured with stopping the robot. Using the function

snap-in-motion, the vision detection can be executed while the robot is transporting a part without

stopping robot motion.

The subsection “2.4.4 Robot Program Creation and Teaching” in the setup procedure described in the

section “2.4 SETUP FOR TOOL OFFSET WITH FIXED CAMERA” is different from those using the

function snap-in-motion.

‘Robot Program Creation and Teaching’ is explained below.

The figure below is an example of layout in the case of tool offset using 2D Multi-view vision process

with a fixed-mounted camera.

Camera

1st view position

2nd view position

Tool offset with a fixed camera (2-D Multi-View Vision Process)

3.3.2.1 Robot program creation and teaching

This subsection demonstrates robot programs for tool offset in motion. Based on sample robot programs

below, create appropriate robot programs for you application.

Following two programs are created.

• Main robot program (MAIN.TP)

• Robot program to detect a target (FIND.TP)

B-83914EN-2/01

207

3. SETUP OF SNAP IN MOTION

Know-How

MAIN.TP

This is the main program. In this program, the robot moves from the start position P[1] to the stop

position P[4] through the two snap positions P[2] and P[3], and calls the subprogram FIND.TP with a

camera view number as an argument at the moment that the robot passes through the snap positions. In

order to confirm that the vision processes are completed in every camera view, it waits until R[1]

becomes 2. After that, the VISION GET_OFFSET instruction is called to get the resulting vision offset.

The WAIT instruction secures that the VISION GET_OFFSET instruction is executed after the VISION

RUN_FIND completion. The robot positions P[5] and P[6] are compensated with the vision offset so

that the part is put at the correct location.

UFRAME_NUM=1

UTOOL_NUM=1

R[1]=0

5: L P[1:start] 500mm/sec CNT100

6: L P[2:snap1] 500mm/sec CNT100 TB 0.00sec,CALL FIND(1)

7: L P[3:snap2] 500mm/sec CNT100 TB 0.00sec,CALL FIND(2)

8: L P[4:stop] 500mm/sec CNT100

WAIT R[1]>=2

10:

VISION GET_OFFSET 'A' VR[1] JMP LBL[99]

11: L P[5:approach] 500mm/sec FINE VOFFSET,VR[1]

12: L P[6:set] 100mm/sec FINE VOFFSET,VR[1]

13: L P[5:approach] 500mm/sec FINE VOFFSET,VR[1]

14:

END

15:

LBL[99]

16:

UALM[1]

FIND.TP

This is a program called by the TIME BEFORE instruction. This program executes detection of the

camera view specified by the argument using the VISION RUN_FIND instruction. And R[1] is

incremented after the image acquisition completes.

VISION RUN_FIND 'A' CAMERA_VIEW[AR[1]]

R[1]=R[1]+1

208

B-83914EN-2/01

Know-How

3. SETUP OF SNAP IN MOTION

3.3.3

Fixed Frame Offset with a Robot Mounted Camera (3D

Tri-view vision process)

In a fixed frame offset application with a robot-mounted camera of 3D Tri-view Vision Process, three

features on a part are detected by the robot mounted-camera and the 3D position of the part is measured.

Using the measured position, robot positions are compensated so that the tool of the robot can reach the

part. The fixed frame offset is usually measured with stopping the robot.

Using the function snap-in-motion, the measurements can be executed continuously without stopping the

robot holding the camera at each snap position.

The subsection “4.3.3 Robot Program Creation and Teaching” in the setup procedure described in the

section “4.2 SETUP FOR FIXED FRAME OFFSET WITH FIXED CAMERA” is different from those

using the function snap-in-motion.

“ROBOT PROGRAM CREATION AND TEACHING” is explained

below.

The figure below is an example of layout in the case of fixed frame offset using 3D Tri-view vision

process with a robot-mounted camera.

Camera

Tool offset with a fixed camera (3-D Tri-View Vision Process)

3.3.3.1 Robot program creation and teaching

Here, examples of robot programs for performing snap-in-motion and fixed frame offset are described.

Change the programs to suit each system, referring the robot programs described here.

The following two robot programs will be created.

• Main robot program (MAIN.TP)

• Detection robot program (FIND.TP)

MAIN.TP

This is a main program. In this program, the robot moves from the start position P[1] to the stop

position P[5] through the three snap positions P[2] ~ P[4], and calls the subprogram FIND.TP with a

camera view number as an argument at the moment that the robot passes through each snap position. To

confirm that the vision processes are completed in every camera view, it waits until R[1] become 3.

After that, the VISION GET_OFFSET instruction is called to get the vision offset. The WAIT

instruction can prevent that the VISION GET_OFFSET instruction executes previously rather than

VISION RUN_FIND is completed. The robot positions P[6] and P[7] are compensated with the vision

offset so that the robot holds the part.

B-83914EN-2/01

209

3. SETUP OF SNAP IN MOTION

Know-How

UFRAME_NUM=1

UTOOL_NUM=1

R[1]=0

5: L P[1:start] 500mm/sec CNT100

6: L P[2:snap1] 500mm/sec CNT100 TB 0.00sec,CALL FIND(1)

7: L P[3:snap2] 500mm/sec CNT100 TB 0.00sec,CALL FIND(2)

8: L P[4:snap3] 500mm/sec CNT100 TB 0.00sec,CALL FIND(3)

9: L P[5:stop] 500mm/sec CNT100

10:

WAIT R[1]>=3

11:

VISION GET_OFFSET 'A' VR[1] JMP LBL[99]

12: L P[6:approach] 500mm/sec FINE VOFFSET,VR[1]

13: L P[7:hold] 100mm/sec FINE VOFFSET,VR[1]

14: L P[6:approach] 500mm/sec FINE VOFFSET,VR[1]

15:

END

16:

LBL[99]

17:

UALM[1]

FIND.TP

This is a program called by the TIME BEFORE instruction. This program executes detection in the

camera view specified by the argument using the VISION RUN_FIND instruction. And R[1] is

incremented after the image acquisition completes.

VISION RUN_FIND 'A' CAMERA_VIEW[AR[1]]

R[1]=R[1]+1

210

B-83914EN-2/01

Know-How

4. FAQS FOR TROUBLESHOOTING

4 FAQS FOR TROUBLESHOOTING

4.1

METHOD FOR ADJUSTMENTS AFTER CAMERA

REPLACEMENT

I don't know how to make adjustments after camera replacement

Measures:

If the camera breaks for some reason, replace it and readjust using the following

procedure.

Cut off the power supply to the robot controller. By cutting off the power supply

to the robot controller, the power supply to the camera will be cut off.

Remove the camera. When doing this, be careful not to apply any pressure to

the diaphragm and the focus ring of the lens.

Remove the lens from the camera.

Attach the lens that was removed in step 3 to the new camera.

Install the camera and fix it in place.

Perform camera calibration.

The work is now complete.

Before removing a camera that is in operation, double-check that the diaphragm of the

lens and the focus ring are fixed securely in place and will not move.

Furthermore, the lens is reusable. If you replace the lens without moving the focus or

the lens ring, readjustment of the locator tools will be unnecessary, which will make

the work easier.

4.2

VISION DATA RECOVERY

I don't know how to recover vision data

Measures:

Recover vision data using the following procedure.

Prepare a memory card on which you have made a backup in accordance with the

procedure in Introduction Edition Section

2.7, "MEMORY CARD

PREPARATION".

The extension for vision data files is VD. If you load all the *.VD files, the

vision data will be recovered.

Be careful with the version ID if the robot controller to copy from is different from the

robot controller to copy to. The version ID of the robot controller to copy to needs to

be the same or higher than that of the robot controller to be copied from.

B-83914EN-2/01

211

4. FAQS FOR TROUBLESHOOTING

Know-How

4.3

DETECTION PROBLEMS

The expected workpiece is not being detected (not found)

Measures:

Please check the following.

• Set it so that logged images are saved, and save the not-found image. Adjust the

locator tool parameters using several images for which detection failed. If

saving logged images, refer to Setup Edition Section 1.7, "VISION CONFIG" in

"R-30iB Plus CONTROLLER iRVision OPERATOR’S MANUAL (Reference)

B-83914EN". After finishing adjustment, change the settings so that images will

not be saved. The detection time may be longer if found images are being saved.

• Adjust the parameters while referring to the logged images. By placing a check

by [Show Almost Found], you will be able to tell which parameter settings are

causing the not-founds.

• If part of the image is causing halation so that the workpiece is difficult to see, the

effect may be reduced to some extent by using the multi exposure function.

However the detection time may be longer if the multi exposure function is used.

If it is necessary to reduce the detection time, do not using multi exposure to

prevent halation, but rather, change the position of the lighting, etc.

• If the contrast between the workpiece and the background is low, the workpiece's

features may not be seen clearly. When detecting the outline of a workpiece

with a pattern match tool, if the workpiece is light-colored, the workpiece will be

seen clearly if you set the background to a dark color.

• When applying the

[Depalletizing Vision Process], if detection becomes

impossible when the number of layers is changed, check whether the [Scale] for

the DOF has been set to something valid.

212

B-83914EN-2/01

Know-How

4. FAQS FOR TROUBLESHOOTING

A different workpiece from the expected one is being detected (false

detection)

Measures:

Please check the following.

•

Set it so that logged images are saved, and save the false-detection image.

Adjust the locator tool parameters using several images for which detection failed.

If saving logged images, refer to Setup Edition Section 1.7, "VISION CONFIG"

in "R-30iB Plus CONTROLLER iRVision OPERATOR’S MANUAL

(Reference) B-83914EN". After finishing adjustment, change the settings so that

images will not be saved. The detection time may be longer if found images are

being saved.

•

Adjust the parameters while referring to the logged images. You can make

adjustments using methods like the following:

Reduce the search area to until it is the smallest one that is necessary.

Set the DOF for [Orientation], [Scale], and [Aspect] to the smallest that are

necessary.

Reduce [Elasticity].

Reduce the setting value for [Area Overlap].

Set the [Emphasis Area].

•

If the workpiece has only a few model features, matching will be likely to occur at

various places within the field of view, which makes it easy for false detection to

occur. By increasing the number of model features, you will be able to perform

detection that is less likely to be affected by noise. If the number of model

features is increased, the detection score may decrease compared to when the

model had only few features. In such cases, set the score to something low.

Even if the score is set to something low, it will still be effective as a

false-detection prevention measure if the number of model features is increased.

•

Enabling [Ignore Polarity] in the setting items for things like pattern match will

make it easy for false detection to occur. It should basically be set to disabled.

B-83914EN-2/01

213

4. FAQS FOR TROUBLESHOOTING

Know-How

I don't know how to redetect

Measures:

If detection has failed owing to a not-found, etc., the detection can be retried using

vision override while changing the detection parameters. For details on vision

override, refer to Setup Edition Section 5.1, "VISION OVERRIDE" in the "R-30iB

Plus CONTROLLER iRVision OPERATOR’S MANUAL(Reference) B-83914EN".

An example of retry while changing the exposure time in a 2-D single-view vision

process is shown below.

The following figure is the setup screen for vision override. Suppose that the vision

process name is 'A', and the vision override name is 'EXPO1'.

The following is a sample program. Vision detection is performed in line 15. As an

example, the exposure time here is set to 20 ms. If the vision detection has failed, a

vision variable setting instruction is called in line 20, and the exposure time is changed

to the value stored in register [5] (for example, 25 ms). If detection is performed

again in this state by executing vision process 'A' in line 15, an image will be snapped

with an exposure time of 25 ms.

12: R[20:retry]=0

;

13: R[15:notfound]=0

;

14: LBL[100] ;

15: VISION RUN_FIND 'A'

;

16: VISION GET_OFFSET 'A' VR[1] JMP LBL[10] ;

17: JMP LBL[20] ;

18: LBL[10] ;

19: IF R[20:retry]=1,JMP LBL[900] ;

20: VISION OVERRIDE 'EXPO1' R[5] ;

21: R[20:retry]=1

;

22: JMP LBL[100] ;

23:

;

24: LBL[20] ;

;

34: LBL[900] ;

35: R[15:notfound]=1

;

Vision override does not overwrite the content of the vision process itself. The values

that are overwritten by override are only enabled for the vision detection instruction

immediately after the vision variable setting instruction is executed. Once the vision

detection instruction is executed, all values set by the vision variable setting instruction

(and all vision variables related to vision processes other than the one that executed

detection) will be cleared.

214

B-83914EN-2/01

Know-How

4. FAQS FOR TROUBLESHOOTING

The lighting environment is not stable

Cause:

In an environment where the intensity in the field of view is not constant between day

and night, detection may not be stable.

Measure 1:

By using the auto exposure function, detection that is resistant to intensity changes can

be performed. Auto exposure is a function that automatically selects the exposure

time in response to the ambient environment so that the intensity of a snapped image

will be the same as that of a reference image. This is done by registering in advance

an image with an intensity to use as the reference.

How to specify the auto exposure area

Specify the auto exposure area. The image that was being displayed when you set the

auto exposure area will be the reference image for auto exposure. Set up the auto

exposure area using the following procedure.

From the [Exposure Mode] drop-down box, select [No].

Adjust the exposure time so that the image has an appropriate intensity.

In the [Exposure Mode] drop-down box, change the selection to [Auto].

Click [OK] on the displayed pop-up that asks for the auto exposure area to be

taught.

Set the auto exposure area.

•

If the auto exposure area has not been taught yet, a window for teaching the

auto exposure area will appear, so set the auto exposure area using the

window.

•

If the auto exposure area has already been taught, click the [Train] button for

[Auto Exposure Area] to change the position and size of the auto exposure

area.

B-83914EN-2/01

215

4. FAQS FOR TROUBLESHOOTING

Know-How

If there is a section you want to ignore in the auto exposure area, click the [Mask]

button and mask the section you want to ignore.

MEMO

1 A section in the image that is completely white or completely black

cannot be specified as the auto exposure area. Set a section that

has intermediate colors to be the auto exposure area.

2 A section in which the image changes significantly is not suitable as

an auto exposure area. For example, in a place where there is

sometimes a workpiece and sometimes not, stable photometry

cannot be performed, as the intensity of its appearance will change

significantly depending on the presence or absence of a workpiece.

How to make fine adjustments for auto exposure

Fine adjustments can be made in auto exposure so that an image will be snapped that

has an intensity that is a little brighter or darker than that of the set reference image.

A value from -5 to +5 can be selected. The larger the value gets in the + direction, the

brighter the image will be, and the larger the value gets in the - direction, the darker the

image will be.

CAUTION

As auto exposure and multi exposures create a single image by

snapping multiple images at a time, the snap-in-motion function

cannot be used.

216

B-83914EN-2/01

Know-How

4. FAQS FOR TROUBLESHOOTING

Measure 2:

Multi exposure snaps multiple images using different exposure times, and creates an

image with a wide dynamic range by combining them. The same as with auto

exposure, detection that is resistant to intensity changes can be performed. Specify

the number of images to snap. A number between 1 and 6 can be set. The more

images are taken, the wider the dynamic range will be, but the time it takes to snap the

images will also be longer. Also the outlines in the image may be blurred overall.

How to specify the multi exposure area

Specify the exposure area used for multi exposure. Images are combined on the basis

of the intensity inside the multi exposure area. The whole of the screen is set by

default, and normally, this does not need to be changed. To set the multi exposure

area, click the [Train] button and set the window. If there is a section you want to

ignore in the auto exposure area, click the [Mask] button and mask the section you

want to ignore.

B-83914EN-2/01

217

4. FAQS FOR TROUBLESHOOTING

Know-How

Multi Exposure Mode

Select the method for combining the images in multi exposure.

• Deviation

For the images in a multi exposure area, the deviation of the luminance is

calculated, and the images are combined in such a way that only a few pixels that

cause halation will remain. This is the default setting.

• Maximum

Images are combined by suppressing the maximum luminance so that the images

in the multi exposure area do not cause halation. If there is even a part of the

multi exposure area where halation is occurring at even a single point in, all other

parts will become relatively darker.

• Average

A combination method that simply calculates the average intensity of each image.

Although it is the method for which the dynamic range is widest, the image will

be darker overall.

CAUTION

As auto exposure and multi exposures create a single image by

snapping multiple images at a time, the snap-in-motion function

cannot be used.

Detection is taking a long time

Cause 1:

A loop using an upward jump label that includes an operation statement with a

movement distance that is 0 has been programmed in.

Depending on how the TP program is created, it may not be possible to process vision

data because almost all of the CPU's capacity will be used for program execution.

Specifically, this applies in cases where a program with a loop that uses an upward

jump has been created, and in that loop, there is an operation statement for which the

movement distance is zero. In such cases, if there is an 'operation statement for

moving to a waiting position', then the movement distance of the operation statement

will be zero starting from the second run of the loop.

Measures:

You can check whether the TP program has the above logic or not by changing the

value of the system variable $PG_CFG.$JMPWAIT_UPR to '16' (the default is '-16').

CAUTION

Changing the above system variable will put the system into the

same state as if a WAIT of 16 ms had been entered in all the

upward jumps. If there is a problem with application as a result of

this change, analyze the TP program to identify the loop that is

causing the problem, and enter a WAIT instruction in that loop (set

the system variable back to the default).

Cause 2:

The execution history is being saved.

Measures:

If the execution history is being saved using a USB memory, it may take a long time to

perform vision detection.

218

B-83914EN-2/01

FANUC R-30iB Plus CONTROLLER, iRVision 2D Camera Application. OPERATOR'S MANUAL - page 8