Mech-Vision Release Notes
Mech-Vision 2.2.2 Release Notes
This section introduces the new features, improvements, and resolved issues of Mech-Vision 2.2.2.
New Features
Support for Changing the Runtime Data Path
Mech-Vision 2.2.2 supports changing the runtime data path. By default, runtime data is saved in the root directory of the software installation directory. If you do not have write permission for the installation directory, go to to change where runtime data is saved. This prevents runtime data from failing to save due to insufficient permissions while the software is running.
|
For changes to other resource storage paths related to this feature, see Upgrade Notes. |
Support for Creating Project Copies
Mech-Vision 2.2.2 adds the Create a Copy of a Project feature, which lets you copy project files and Step settings as needed.
Support for Range XL-GL Cameras
Mech-Vision 2.2.2 adds support for Range XL-GL cameras and the Capture Images from Line Scan Camera Step. This Step can acquire images and point clouds with these cameras for subsequent vision processing and analysis.
New Custom Communication Mode
Mech-Vision 2.2.2 supports establishing communication connections using standard or custom industrial communication protocols, and exchanging data according to data formats and parsing rules agreed upon by both parties. This mode is suitable for scenarios with special communication requirements or custom protocols.
-
Supports configuring multiple custom communication services, including TCP Server, TCP Client, Siemens S7 Client, Modbus TCP Master, and Mitsubishi MC Client.
-
Supports combining communication parsing, project triggering, picking execution, and result transmission into a configurable workflow, without requiring separate scripts or adapters for each project.
-
Supports building communication and picking projects. Data processing and result transmission can be completed using Steps such as Receive Data, Parse Data, Trigger Project, Assemble Data, and Send Data.
-
Supports managing custom communication services and configuring communication trigger rules. For TCP Server and TCP Client services, receiving data triggers the project. For PLC-type services, the corresponding project can be triggered based on register variable values. For details, see Quick Guide for Robot and Communication Configuration.
Support for OPC UA Server Communication
Mech-Vision 2.2.2 adds support for the OPC UA Server communication protocol, enabling data exchange between an OPC UA client and an OPC UA server.
Support for Operating the Production Interface Using a Touch Screen
When using the Production Interface, you can operate it using a touch screen.
Support for Creating Surface Trajectories
-
The trajectory-type target object configuration workflow in the target object library supports creating surface trajectories.
-
Added the Generate Trajectory from Surface Point Cloud Step, which generates trajectory points from input point clouds of target object surfaces.
Support for Robot Calibration
Mech-Vision 2.2.2 adds the Robot Calibration feature. This feature uses a camera to measure the pose of a calibration sphere at different robot poses, and calculates the robot’s joint zero positions and link lengths based on the measurements. The parameters are used to compensate for the robot’s structural parameters, thereby improving its absolute accuracy.
Support for AI 1D/2D Barcode Recognition
Mech-Vision 2.2.2 adds the AI 1D/2D Barcode Recognition Step, which recognizes 1D/2D barcodes in a specified image region and retrieves information such as the barcode content, type, center pose, contour, and judgment result. Supported barcode types include Code 128, Code 39, Code 93, EAN, UPC, DM Code, and QR Code.
AI 1D/2D barcode recognition is also supported in the Information Reading scenario of 2D target object recognition. For details, see Information Reading (1D/2D Barcode Recognition).
| This feature requires a "Lite AI" license. Contact Mech-Mind Sales to obtain one. |
Added Several Steps
Mech-Vision 2.2.2 adds several Steps, as described below.
| Step | Description |
|---|---|
This Step precisely localizes a target object by detecting predefined lines or circles in an image, and outputs the object’s precise pose. |
|
This Step transforms input 2D poses between coordinate systems based on the object pose. It can transform 2D poses obtained in a rectified/aligned image coordinate system back to the original image coordinate system, or transform 2D poses in the original image coordinate system to the rectified/aligned image coordinate system. |
|
This Step compares the translational and rotational deviations between input poses and reference poses in a shared coordinate system, and filters valid poses based on the allowed deviation ranges. Before using this Step, make sure that the input poses and reference poses are in the same coordinate system. |
|
This Step extracts valid poses within a specified cuboid from one or more input poses. The cuboid’s position and orientation are determined by the input pose of its center. The cuboid dimensions can be set using Step parameters or dynamically provided through input ports. |
|
This Step extracts points within a specified geometry from an input point cloud. Supported geometries include cuboids, cylinders, rings, and sectors. |
|
This Step converts input Euler angles to the corresponding quaternions, standardizing the representation of rotation in poses for subsequent pose calculations. |
|
This Step splits an input alignment parameter group into X, Y, or Z coordinates and rotation angles to obtain a reference point and reference orientation. |
|
This Step executes preconfigured actions in sequence to complete initialization before measurement or inspection. |
|
This Step reads data from a specified JSON file for use by subsequent Steps. |
Improvements
Improved Usability of Lite AI Features
Mech-Vision 2.2.2 improves the usability of the "Lite AI" features as follows.
| Target Object Configuration Workflow | Improvements |
|---|---|
Common Improvements |
|
Improvements Specific to the AI Binary Classification (Front/Back, Present/Absent) Step |
|
Improved the Target Object Configuration Workflow
Mech-Vision 2.2.2 improves target object configuration workflows in the target object library as follows.
| Target Object Configuration Workflow | Improvements |
|---|---|
Grasp-type Target Objects |
|
Trajectory-type Target Objects |
|
Bin-type Target Objects |
|
Improved the Usability of the Auto-Correct Accuracy Drift in Vision System Feature
Mech-Vision 2.2.2 improves the Auto-Correct Accuracy Drift in Vision System feature as follows.
-
Improved the deployment workflow for Auto-Correct Accuracy Drift in Vision System to make it easier to use.
Deployment Workflow Improvements -
Supports automatic generation of a robot self-correction program. The exported program can be used to automatically acquire calibration sphere poses and generate drift correction data.
-
Supports point cloud correction. Combine the point cloud correction Steps, export them, and load them into a picking project to correct point clouds.
-
Supports an automated deployment workflow. After the reference image-capturing point is validated, all image-capturing points can be generated and validated automatically.
-
Improved the existing manual deployment workflow by adjusting the Step sequence, making the workflow easier to use.
-
Supports standard and simplified workflows. The standard workflow plans nine image-capturing points per layer, while the simplified workflow plans three.
-
-
Improved the Generate Drift Correct Model and Correct Accuracy Drift of Vision System Steps to support the new deployment workflow.
Improved Matching and Alignment
Mech-Vision 2.2.2 improves how the 2D Matching and 2D Alignment Steps handle scaling results.
Improved the 2D Matching Template Editor
Mech-Vision 2.2.2 improves the 2D Matching Template Editor as follows.
-
Some feature parameters have been moved to the parameter configuration panels of the related Steps for easier adjustment.
-
You can set the template coordinate system so that ROIs or feature regions in downstream Steps can continue to be reused after the template is updated.
Enhanced Global Variable Management
Mech-Vision 2.2.2 improves the Global Variables feature as follows.
-
The Global Variable Viewer Step supports setting the queue capacity, that is, the maximum number of data entries to store.
-
The maximum number of global variables that can be added in the Global Variable Viewer in a solution has increased from 32 to 100.
-
The Global Variable Step can read all data stored in the queue without removing it.
Improved Other Steps
Mech-Vision 2.2.2 improves several Steps as described below.
| Step | Description |
|---|---|
Supports retrieving camera parameter configurations from the "2D Camera Management" window. |
|
Improved the ordering of Step output ports. When depth map and color image outputs are selected, image ports are listed first, making them easier to connect to downstream Steps such as "3D Target Object Recognition" and "Adjust Poses V2" and keeping dataflow connections clear and concise. |
|
Added pose filters for filtering poses by reference pose and extracting poses within a cuboid. |
|
Expanded the usage scenarios to include object pose output, positioning and picking, placement correction (fixed placement position), and placement correction (non-fixed placement position). The Step can output poses, pose offsets, and other data according to the selected scenario. |
|
|
|
Supports creating poses using quaternions or Euler angles. |
|
Improved the Step and parameter names. |
|
Renamed the "Gamma Correction" parameter to "Brightness Adjustment". |
|
Improved the Step’s English name. |
Resolved Issues
Mech-Vision 2.2.2 resolves the following issues:
-
Flange pose data might not be saved correctly when saving a project containing data from the "2D Smart Camera" Step.
-
When performing Auto-Correct Accuracy Drift in an EIH Vision System with an ULTRA M-GL camera, selecting and validating a region might incorrectly indicate that the extracted point cloud is invalid.
-
A configured 2D smart camera might fail to connect when a Mech-Vision project is opened by double-clicking it.
-
The software might crash when viewing the extrinsic parameter report after hand-eye calibration is completed with a FANUC robot installed at a 90° angle to the camera and flange.
-
In the "Visualize Point Cloud and Pose" Step, the debugging output window might continue to display the previous pose after the connection to the "Pose" input port is disconnected.
-
Post-processing might incorrectly filter instances when using a long, narrow instance segmentation model, affecting multi-object picking.
-
During EIH hand-eye calibration for a four-axis robot, the software might not correctly explain the error when too few rotational calibration points are generated.
-
Images might not be saved when the "Save Images" Step receives image data through the "Mapping Index" input port.
-
Lite AI-related Steps might incorrectly report a missing software license on Windows 10 version 1809.
-
When the camera is installed in an EIH setup, standard interface communication is used, and the robot does not send flange poses, the software might fail to display an error message.
-
Deep learning inference results might differ between Mech-Vision and Mech-DLK when using the same data and ROI settings.
-
Step ports inside a Step group might not be displayed when configuring 2D smart camera output channels, preventing port binding.
-
A target object might not be recognized correctly when removing its background in the target object library.
-
The error_msg field might be missing from saved data when a custom alarm or system error occurs.
-
RGB channel values might change when a defect segmentation model performs inference on the same image multiple times, resulting in inconsistent inference results.
-
Dataflows between Steps inside a Step group might be disconnected after reopening a project if an "Accept All Parameters" Step is added before a "Logical Judgment" Step in the group.
-
The ABB robot self-correction program for Auto-Correct Accuracy Drift in ETH Vision System contains an undefined variable.
-
In a gantry project, the secondary camera might fail to receive flange poses when robot communication is configured using "Other Robot" and data is transferred between projects.