Multi-Region Alignment

## Multi-Region Alignment
Alignment is performed within the physical coordinate system of the robot, using points in mechanical coordinates.
The input for this module is an array of points extracted from multiple cameras.
Additionally, a weighting function has been added, allowing users to set the calculation proportion of each point during alignment. Points with higher weights have a greater influence on the result.

## Usage Scenarios
Machine vision alignment refers to the process of using a machine vision system (cameras, lenses, lighting, image processing software) to precisely identify and measure the position and angle (pose) of a target object (workpiece), and then guide, adjust, or align it to a preset desired position and angle. The core goal is to eliminate positional and angular deviations, achieving high-precision positioning and assembly.
Simply put, the vision system tells the control system: The target is currently here (coordinates X, Y, angle θ). It needs to be at the target position (coordinates X0, Y0, angle θ0). The difference is (ΔX, ΔY, Δθ). Please move by this amount to align.
Alignment is generally used when aligning objects on two planes, commonly in film application and labeling. It is needed in situations where information captured by two cameras needs to be transformed relative to each other.

![](/media/202601/2026-01-16_103750_4119460.3507559082787539.png)

## Module Principle
In previous modules, whether correction or alignment modules, the inputs were image points, along with calibration and mapping data, finally outputting mechanical compensation. In the Mechanical Point Alignment module, both inputs and outputs are mechanical points. It performs alignment calculations directly without calibration conversion. This allows it to handle more complex mechanical models and requires users to handle the conversion between image points and mechanical points within their workflow, facilitating error tracing.
Multi-module alignment is similar to mechanical point alignment, meaning it performs alignment directly on the input data without calibration conversion. This requires that the input data be in the same coordinate system.
![](/media/202603/2026-03-13_115853_8234540.06856568599736468.png)
Mechanical models are divided into types that include angles (XYR, XR, YR) and types that do not include angles (XY). The calculation methods for these two types of alignment are different, and the results will also differ.
Clockwise Coordinate System:
![](/media/202601/2026-01-16_113933_0972430.8700678454763793.png)
Counterclockwise Coordinate System:

![](/media/202601/2026-01-16_113959_6188830.08372255033215326.png)

The mechanical coordinate system type corresponds to the current robot's coordinate type. When the type includes angle R, it affects the calculation results and the direction of the angle.
There are three alignment methods: Point-to-Point, Point + Angle, and Point + Line. The Point-to-Point method uses the SVD decomposition algorithm. The Point + Angle and Point + Line methods use the same algorithm. For the Point + Line method, the angle of the input line is extracted and used in the Point + Angle algorithm to calculate the result.
### Compensation Amount
For mechanical models that include angle R, compensating R will affect the XY values. (i.e., if only R is compensated, the XY position where the robot grasps the workpiece will not change, but the actual mechanical coordinates XY will change slightly to accommodate the rotation.)
The compensation amount is added to the result after the final alignment. If the alignment is not yet complete and needs to continue, the result output will not include the compensation. The compensation amount does not affect the judgment of alignment completion accuracy.
Since inputs are all mechanical points and do not require image data, the alignment can only be completed in one step.

### Advanced Parameters
Departure Pose: The mechanical coordinate position corresponding to the starting point, i.e., the position of the rotation center. This is typically the拍照 position of the bottom camera. When the mechanical type includes angle R, the拍照 pose significantly impacts the results.
Alignment Accuracy refers to the mechanical precision. When the alignment output is less than the Alignment Accuracy, the alignment is successful. Otherwise, if the number of alignments performed is less than the set number of correction attempts, error code 1 is output, and alignment continues.
Mechanical Range defines the robot's permissible motion range. It only validates the actual pose and does not restrict the offset pose. If the output actual coordinates fall outside this range, error code 5 is output.

![](/media/202603/2026-03-13_115838_7761190.08075385735245832.png)

### Weight Parameters

![](/media/202601/2026-01-16_115729_3472490.8188645567092419.png)
Records the weight for each point. The default value is 1. Any input value greater than 0 is acceptable; the calculation will automatically normalize the weights.

## Usage Method

### Pre and Post Modules
Image capture from camera, line extraction, etc., are used to obtain point arrays for alignment.
When the input is a group of image points, the input photography pose requires image pixel points, and the output will also be pixel values.
When the input is a group of mechanical points, the extracted image point data must all be converted to mechanical point data. In this case, the output will be the actual physical offset.

## Parameter Configuration Description
【Mechanical Type】Select the corresponding mechanical model based on the coordinate system type of the actual robot.
【Calibration Group】Select the coordinate system group for the current equipment. The coordinate system group must be set in the Coordinate System Calibration under Global Data Management.
【Device Selection】Select the mechanical device used for mechanical alignment within the chosen coordinate system group. (Used to obtain the rotation direction and coordinate system of the robot.)
【Number of Point Pairs】The number of input regions; changing this value will affect the number of regions that need to be input.
【Compensation Amount】Mechanical coordinates for compensation within the mechanical coordinate system. When subscribing, the compensation amount can be introduced via a port.
【Reference Image】The image used as a reference during processing.
#### Advanced Parameters
【Teaching Pose for Photography】The mechanical end coordinates when the camera can capture the material during the teaching process.
【Alignment Accuracy】Mechanical coordinates defining the accuracy range that the alignment needs to achieve.
【Number of Corrections】The maximum number of alignment attempts allowed. If the alignment result does not meet the Alignment Accuracy but is still within the Mechanical Range, alignment continues until the number of corrections is reached.
【Mechanical Range】The permissible motion range of the robot.

## Output Description

![](/media/202603/2026-03-13_115822_2429570.8065222481590418.png)
When all inputs are mechanical points, the output is the mechanical offset.
When all inputs are image points, the output is in pixel values.
The output includes two sets: Grasp Pose and Offset Pose. The Grasp Pose is the actual pose for the robot to grasp, and it is also the range checked against the 【Mechanical Range】. The Offset Pose is the offset of the robot relative to the taught position and is not checked against the mechanical range.
| Feature   | Grasp Pose       | Offset Pose              |
|------|------------|-------------------|
| Coordinate Type | Absolute Coordinates (GPS-like) | Relative Coordinates (Reference-based) |
| Safety Check | Full check against mechanical range | Range not checked |
| Primary Function | Final execution position | Position fine-tuning |
| Analogy | Detailed delivery address | "To your front left" |
| Dependency | Absolute Coordinates | Includes taught position + offset based only on taught position |
## Notes
The type of **Mechanical Coordinate System** corresponds to the current robot's coordinate type. When the type includes angle R, it affects the calculation results and the direction of the angle.
## Troubleshooting
1.  When alignment is inaccurate, check if the feature matching module is accurately acquiring the image.
2.  If stable correction still cannot be achieved, test the accuracy of the calibration and mapping.
    (Method for testing calibration and mapping issues:
    Change the pick-up拍照 position by 2mm and observe if the result output changes by 2mm. If the result is incorrect in this case, it indicates a calibration error.
    Change the placement拍照 position by 2mm and observe if the result output changes by 2mm. If the result is incorrect in this case, it indicates a mapping error.)
3.  If the offset angle increases, leading to XY inaccuracies, it may be due to an incorrect calculation of the rotation center during calibration. Try re-calibrating, modifying the calibration method, or improving the rotation center calculation (add calibration verification or use the Long Radius Calibration module).
4.  If correction is accurate at the center of the field of view but shows XY deviations at the edges, consider image distortion (use the Image Distortion Correction module) or an incorrect scale calculation in the calibration (use the Calibration Check module).

---

Click here to jump to the corresponding Chinese manual page: [Multi-Region Alignment](/doc/303)