Reference Points and Camera Points
Reference points are important points to be used to calibrate the relation of the image coordinates and the Camera or Robot coordinate systems. Camera points are points used to capture an image for detecting a target. Calibration associates the coordinates with the reference points. The reference points and camera points are configured by jogging the robot in the point teaching mode.
Required reference points and camera points vary depending on camera mount types.
The reference points specified by using a robot have the following three types.
- TaughtPoints (Teach point)
- EndEffector (End effector)
- UpwardCamera (Upward camera)
The reference points and camera points can be selected according to the camera mount type as shown in the table below. For standalone camera calibration which does not use a camera, you manually enter the coordinate values (horizontal and vertical distance between reference points) of the reference points into the system.
Types of reference points by camera mount type
| Camera mount type | Selectable reference point type | Reference points teaching method | Target detecting method | Required vision sequence |
|---|---|---|---|---|
| Mobile Camera | TaughtPoints | One-point teaching | Nine-camera-point teaching * | Detect one target |
| UpwardCamera | Unnecessary | Nine-camera-point teaching * | Detect one target | |
| Fixed downward Camera | TaughtPoints | Nine-point teaching | One-shot nine-point detection | Detect nine targets |
| EndEffector | Unnecessary | Nine-camera-point teaching * | Detect one target | |
| Fixed upward Camera | EndEffector | Unnecessary | Nine-camera-point teaching * | Detect one target |
| Standalone Camera | - | Nine-point coordinate specification | One-shot nine-point detection | Detect nine targets |
*Automatic camera point generation is available. (Refer to the description below)
Reference points teaching method
When “TaughtPoints (teach point)” is set as a reference point type, jog the robot so that positions of the tool and the target match, and teach one or nine points.
For how to set the reference point for a mobile camera, see the following.
- Mobile Camera Reference Points
For how to set the fixed downward and standalone camera reference points, see the following. - Fixed Downward and Standalone Camera Reference Points
Nine-camera-point teaching
When “nine-camera-point teaching” is used for detecting the target, a total of nine positions are detected with one point for each image capturing.
In the calibration, move the target to the specified nine areas in the field of view of the camera and capture images to detect the target. In the point teaching mode, jog the robot to teach points while checking the images so that the nine camera points can be detected at a proper position in each image.
You need to create a vision sequence that detects one target with the target camera beforehand.
For how to create a sequence, see the following.
Creating Vision Sequences for Calibration
Automatic Camera point generation
When “nine-camera-point teaching” is used for detecting the target, the automatic camera point generation is available. Instead of teaching nine camera points, eight camera points can be generated automatically by teaching only one point at the center of the field of view.
KEY POINTS
When using the automatic camera point generation, the robot moves automatically to generate camera points during the calibration. Be careful of interference between the robot and peripherals. Also, use with avoiding singularity nearby posture that each axis extends to prevent an error during automatic camera point generation.
One-shot nine-point detection
When “one-shot nine-point detection” is used for the calibration, positions of nine targets are detected in one shot.
It is necessary to create a vision sequence for detecting nine targets by a camera in advance.
For how to create a sequence, see the following.
Creating Vision Sequences for Calibration
Mobile Camera Reference Points
This scheme requires one reference point.
You can specify a taught point (reference point type=“TaughtPoints”) that you manually teach by jogging the robot, or you can specify a point (reference point type= “UpwardCamera”) that is found with a fixed upward camera. The latter is preferred for greatest accuracy and automation.
Here are some examples of taught reference points:
- A part or calibration target in the robot work envelope.
- A hole somewhere in the work envelope that a tool mounted on the robot end effector can be slipped into.
When using a fixed upward camera to find the reference point, there needs to be “a thin plate with hole through it that can be seen by both the mobile camera and the fixed upward camera” and a calibration plate.
During calibration, the fixed upward camera which has been previously calibrated locates the reference hole in the plate. Then the mobile camera is calibrated by searching for the reference hole in nine positions.
Using a fixed upward camera to find the reference point for mobile camera calibration is more accurate because during calibration of the fixed upward camera, the robot tool is rotated 180o for each camera point so that the precise center of the U axis of the robot can be determined. This allows the calibrated fixed upward camera to more accurately find the reference hole that the mobile camera will locate during calibration.
Mobile Camera Calibration using Fixed Upward Camera
| Symbol | Description |
|---|---|
| a | Mobile camera |
| b | Calibration plate |
| c | Fixed upward camera |
| d | Hole |
Fixed Downward and Standalone Camera Reference Points
The following calibration schemes for which the reference point type is set to “TaughtPoints (teach point)” require a calibration target plate or sheet that contains nine targets. The figure below shows the pattern for the nine calibration targets.
- Fixed downward Camera
- Standalone Camera
Fixed Downward and Standalone Camera Calibration Targets
For fixed downward camera calibrations, the targets could be holes in a plate that a rod on the robot end effector can be slipped into. The distances between the targets do not have to be exact. When the points are taught, the calibration software will read the robot positions.
For standalone cameras calibrations, a pattern on paper, mylar, etc. can be used. The horizontal and vertical distances between the targets must be known. These are entered when you teach the points from the [Calibration] dialog box.
Teaching by TwoRefPoints
When teaching reference points for mobile camera calibrations and fixed downward camera calibrations, you will be prompted to teach the point and then optionally teach it again at 180° from the first position. This allows the system to determine more accurate location of the reference point in the robot coordinate system. However, if you are using a robot tool that has been accurately defined and specified in the calibration setup, you can skip teaching the point at 180°.
To skip the 180° steps, ensure that the TwoRefPoints calibration property is set to False.
For details on defining tools, refer to the following section.
Using Vision Guide 8.0 with SPEL+