Mech-Vision Release Notes

You are currently viewing the documentation for the latest version (2.2.0). To access a different version, click the "Switch version" button located in the upper-right corner of the page.

■ If you are not sure which version of the product you are currently using, please feel free to contact Mech-Mind Technical Support.

Mech-Vision 2.2.0 Release Notes

This section introduces the new features, improvements, and resolved issues of Mech-Vision 2.2.0.

New Features

Mech-Vision 2.2.0 supports trajectory scene operations with multiple new trajectory-related features as follows.

Added trajectory target object configuration workflow to the target object library, supporting quick completion of trajectory target object configuration by importing STEP files, importing processed point clouds, and collecting point clouds with cameras.
Added 3D Trajectory Recognition Step. This Step integrates point cloud preprocessing, 3D matching, and other vision processing functions, enabling quick target object recognition and trajectory generation.
Improved Output Step functionality to support outputting trajectory-type vision results.

Added multiple trajectory processing Steps as follows.

Step	Description
Generate Trajectory Points from Point Cloud	This Step generates trajectory points from input point clouds for subsequent trajectory processing.
Refine Trajectory with Point Cloud	This Step optimizes the original trajectory based on actual point clouds, making the trajectory better fit the target object surface. It supports multiple operations such as smoothing, sorting, and simplification, improving trajectory accuracy.
Smooth 3D Trajectory	This Step smooths trajectory points to reduce the impact of noise in the trajectory and generates a smoother trajectory.
Simplify Trajectory	This Step simplifies trajectory shape by reducing the number of trajectory points while maintaining the overall trajectory form.
Sort Trajectory Points	This Step sorts trajectory points based on their positions to optimize the trajectory point sequence.
Remove Overlapped Trajectory Points	This Step removes overlapped points with excessive proximity in the trajectory to reduce the number of invalid points and optimize trajectory point distribution.
Interpolate Trajectory Points	This Step inserts more intermediate points between trajectory points to make the trajectory smoother and more continuous.
Acquire Trajectory Information	This Step obtains trajectory information related to trajectory-type target objects from the target object library based on the target object center point and target object name output by the previous Step, for use in subsequent Steps.

Step

Description

Generate Trajectory Points from Point Cloud

This Step generates trajectory points from input point clouds for subsequent trajectory processing.

Refine Trajectory with Point Cloud

This Step optimizes the original trajectory based on actual point clouds, making the trajectory better fit the target object surface. It supports multiple operations such as smoothing, sorting, and simplification, improving trajectory accuracy.

Smooth 3D Trajectory

This Step smooths trajectory points to reduce the impact of noise in the trajectory and generates a smoother trajectory.

Simplify Trajectory

This Step simplifies trajectory shape by reducing the number of trajectory points while maintaining the overall trajectory form.

Sort Trajectory Points

This Step sorts trajectory points based on their positions to optimize the trajectory point sequence.

Remove Overlapped Trajectory Points

This Step removes overlapped points with excessive proximity in the trajectory to reduce the number of invalid points and optimize trajectory point distribution.

Interpolate Trajectory Points

This Step inserts more intermediate points between trajectory points to make the trajectory smoother and more continuous.

Acquire Trajectory Information

This Step obtains trajectory information related to trajectory-type target objects from the target object library based on the target object center point and target object name output by the previous Step, for use in subsequent Steps.

Support for Bin Recognition in Picking Scenarios

Mech-Vision 2.2.0 supports bin recognition in picking scenarios, mainly including:

Added bin target object configuration workflow to the target object library, supporting quick completion of bin target object configuration by importing STL files, collecting point clouds with cameras, and configuring without point cloud models.
Added 3D Bin Recognition Step. This Step integrates point cloud preprocessing, 3D matching, deep learning and other vision processing functions, enabling quick bin recognition.

Support for Pose Transformation in Assembly Scenarios

Added Advanced Pose Transformation Step. This Step can be used to calculate pose relationships between different coordinate systems, including coordinate system transformation, dual-robot relative pose calibration, and pose correction (assembly).
Added Frame Transformation Calculator Step. This Step automatically finds the shortest transformation path based on multiple known pose relationships and calculates transformation relationships between two target coordinate systems.

Extended 2D Vision Features

Mech-Vision 2.2.0 added multiple 2D vision-related features, mainly including:

Added 2D Smart Camera Step to acquire image data through a 2D smart camera as input for subsequent vision processing tasks.
Added 2D Camera Calibration feature, including distortion calibration and hand-eye calibration (external parameter calibration), to establish the mapping relationship between camera imaging and actual space, improving image measurement and positioning accuracy, and serving as the foundation for the vision system to achieve high-precision recognition and positioning.
Added 2D Camera Management feature to centrally manage 2D cameras in a solution, serving as the device configuration and debugging entry point before image acquisition.
Added 2D Template Editor feature to create and manage 2D matching templates.
Added 2D Target Object Recognition Step. This Step can quickly complete target object recognition and detection in positioning picking, placement deviation correction, error prevention inspection, and information reading scenarios.

Added multiple 2D-related Steps covering image preprocessing, image post-processing, and measurement scenarios. The Step library supports grouping Steps by 2D and 3D dimensions.

Category	Step	Description
Preprocessing	Channel Merging or Splitting	This Step splits a three-channel image into three single-channel images, or merges three single-channel images into one three-channel image.
Image Preprocessing	This Step performs preprocessing operations on input images such as enhancement, denoising, morphological transformation, grayscale inversion, and edge extraction.
Compose Images	This Step composes multiple images by placing them at specified positions into a single image.
Crop, Pad, or Resize Image	This Step crops, pads, or resizes input images.
Flip and Rotate Image	This Step flips and rotates images as required.
Rotate Image	This Step rotates an image by a specified angle around the set rotation center.
Extract Target Region by Color	This Step extracts target regions in a specified color space based on color range (upper and lower threshold limits) of three channels to generate a binary image (target region pixels have a value of 255, other pixels have a value of 0).
Post-processing	Mask Logical Operation	This Step performs logical operations on two groups of masks with the same dimensions to merge two groups of masks, extract the same part, or remove the same part.
Count Pixels in Grayscale Range	This Step counts the number of pixels in a specified region of a grayscale image that meet the grayscale threshold range.
Grayscale Histogram Analysis	This Step outputs information about the input grayscale image including grayscale histogram, number of pixels, minimum and maximum grayscale values, grayscale median, grayscale mode (most frequent grayscale value), grayscale mean, grayscale standard deviation, and contrast. A grayscale histogram is a statistical analysis of the grayscale level distribution in a grayscale image, reflecting the frequency of a certain grayscale value appearing in the image.
Image Arithmetic Operation	This Step performs pixel-wise arithmetic operations (addition, subtraction, multiplication, or division) or combined operations (maximum, minimum, inversion) on input grayscale images, and can adjust results through multiplier and addend. The two images must have the same dimensions.
Measurement	Detect and Fit Line	This Step detects line edges from an image and fits a line.
Detect and Fit Circle	This Step detects circular edges from an image and fits a circle.
Detect and Fit Rectangle	This Step detects rectangular edges from an image and fits a rectangle.
Detect Edge Point	This Step detects an edge point that meets the requirements on the vertical center line of a specified region in an image.
Detect and Fit Oblong Hole	This Step detects oblong hole edges from an image and fits an oblong hole.
Measure Edge-to-Edge Width	This Step detects a pair of edge points from an image and measures the distance between them as the edge-to-edge width.
Measure Feature-to-Feature Distance	This Step measures the geometric distance between two specified feature types. Supported feature combinations include: point-to-point, point-to-line, point-to-circle, line-to-line, line-to-circle, and circle-to-circle.
Measure Angle Between Segments	This Step measures the angle between line segments.
Other	2D Matching	This Step searches for and locates features matching the template in a 2D image, calculates object poses, and provides data for subsequent Steps that need to perform 2D pose transformation simultaneously. Supports multi-object positioning and recognition.
2D Alignment	This Step aligns input images with the template through translation and rotation to achieve consistency between the two.
2D Blob Analysis	This Step detects blobs in images and filters them based on their geometric characteristics such as area and roundness.
2D Blob Alignment	This Step detects blobs in images, filters them based on their geometric characteristics, and adjusts image pose so that the blob centroid coincides with the original image center point.
Image Visualization (2D)	This Step overlays features, images, text, poses and other results on 2D images to realize custom visualization.
1D/2D Barcode Recognition	This Step recognizes 1D/2D barcodes in a specified image region and outputs the barcode content and its position in the image.
2D Target Object Counting	This Step detects and counts target objects in an image through 2D template matching.
2D Target Object Deformation Detection	This Step performs deformation status detection for target objects in an image through 2D template matching.
2D Target Object Pose Deviation Detection	This Step performs pose deviation detection for target objects in an image through 2D template matching.
Locate Mask Feature Point	This Step locates the feature points of a mask.
Create Point (2D)	This Step creates 2D points based on geometric features such as points and lines.
Create Line (2D)	This Step creates 2D lines based on geometric features such as points and lines.
Convert 2D Point to 3D Pose	This Step converts the 2D pose or 2D shape of a target object recognized by a 2D camera into a 3D target object pose in the robot coordinate system by combining external parameter calibration data and teaching poses.

Enhanced Deep Learning Model Package Inference Features

Mech-Vision 2.2.0 enhanced the Deep Learning Model Package Inference Step as follows.

Supports inference of Fast Positioning, Text Detection, Text Recognition, and Unsupervised Segmentation model packages, as well as inference of Multi-Model Package.
Enhanced inference configuration functionality, supporting post-processing parameter settings in the Inference Configuration Tool to improve inference result accuracy.

Starting from Mech-Vision 2.2.0, this Step only supports loading model packages exported from Mech-DLK 2.5.4 or later.

Added Lite AI Steps

Mech-Vision 2.2.0 added the following Lite AI Steps.

To use the "Lite AI" related Steps, please contact Mech-Mind Sales to obtain software licenses supporting this feature. After updating the software license, you can use this Step.

Step	Description
AI Classification (Multi-class)	This Step performs multi-class intelligent classification on target objects in an image.
AI Binary Classification	This Step performs binary classification (e.g., front/back, present/absent) on target objects in an image.
AI Optical Character Recognition	This Step performs intelligent recognition on letters, numbers, and other characters in an image.

Step

Description

AI Classification (Multi-class)

This Step performs multi-class intelligent classification on target objects in an image.

AI Binary Classification

This Step performs binary classification (e.g., front/back, present/absent) on target objects in an image.

AI Optical Character Recognition

This Step performs intelligent recognition on letters, numbers, and other characters in an image.

Added Other Deep Learning Steps

To use the "Pick Anything V2" Step, please contact Mech-Mind Sales to obtain software licenses supporting this feature. After updating the software license, you can use this Step.

Step	Description
Pick Anything V2	This Step performs surface segmentation on input depth images and color images based on the Pick Anything V2 model package, identifying each independent graspable surface and overlapped surfaces, and outputs a list of masks sorted by grasping priority.
Object-Bin Segmentation	This Step performs segmentation of target objects and bins on input depth images and color images based on the target object and bin segmentation model package, outputting target object masks and bin masks, and provides visualization results.

Step

Description

Pick Anything V2

This Step performs surface segmentation on input depth images and color images based on the Pick Anything V2 model package, identifying each independent graspable surface and overlapped surfaces, and outputs a list of masks sorted by grasping priority.

Object-Bin Segmentation

This Step performs segmentation of target objects and bins on input depth images and color images based on the target object and bin segmentation model package, outputting target object masks and bin masks, and provides visualization results.

Support for Global Variables

Mech-Vision 2.2.0 supports using Global Variable Steps and Global Variable Viewer, enabling data sharing between different projects, achieving unified data management and transmission.

Added Other Steps and Procedures

Mech-Vision 2.2.0 added several other Steps and Procedures as follows.

Category	Name	Description
Step	Fit Pose from Feature Points	This Step calculates the overall optimal pose of a target object by performing best-fit matching between multiple real-time feature points on the target object and template feature points, improving adaptability in deformation scenarios.
Light Source	This Step is used to connect and configure a light source controller to control the light source operating mode and output brightness.
Save Data to Files	This Step saves data to a local path.
Logical Judgment	This Step can perform logical judgment on input data according to set rules and output boolean results (True or False).
Procedure	Point Cloud Preprocessing	This Procedure performs preprocessing operations such as point filtering, point cloud merging, and image filtering on raw point clouds to delete interfering points, thus speeding up the processing of subsequent Steps.
Get Point Clouds of Target Objects on the Highest Layer	This Procedure extracts point clouds of target objects on the highest layer from multi-layer target object point clouds. By filtering point cloud height in the specified direction, it removes point cloud interference from lower-layer target objects.
Get 2D Image of Highest Layer Target Object	This Procedure filters the original image through point cloud information and extracts 2D images containing only the highest layer target objects, effectively removing background and lower-layer target object interference.
Get Point Cloud of Target Object Flat Surface	This Procedure obtains relatively flat surface point clouds by removing point clouds far from the target object flat surface.
Multidimensional Pose Sorting (Height, Angle)	This Procedure sorts target object poses by multiple dimensions based on target object height and angle, optimizing grasping order to improve grasping stability and success rate.
Validate Target Object Pose Angle and Position Compliance	This Procedure validates whether the pose angle and position of target objects meet requirements.
Count Number of Elements	This Procedure counts the number of elements (such as poses) in input data for use in subsequent logical judgment.
Create Target Object Reference Frame	This Procedure establishes a target object reference coordinate system through three feature points (primary reference point, Y-direction auxiliary reference point, and XY-plane auxiliary reference point), thereby determining the position and orientation of the target object in space.
Transform Point Cloud to Specified Frame	This Procedure transforms point clouds from camera or robot coordinate systems to a user-specified coordinate system.

Support for Solution Switching

Mech-Vision supports Solution Switching Management functionality, enabling automatic switching of solutions based on received solution IDs. Each solution switching rule defines the mapping relationship between a solution ID and the solution path.

Support for Procedure Locking

Mech-Vision 2.2.0 supports Locking and Protecting Procedures. After a Procedure is locked, non-administrator users cannot view or modify the Step logic inside the Procedure. They can only call and adjust the Procedure through the exposed parameters.

Added Mech-Vision Secondary Development Features

Mech-Vision 2.2.0 added Mech-Vision SDK documentation and interface capabilities, supporting integration of solution, project, and Step vision capabilities into client applications, with C++, C#, and Python interfaces available.

Improvements

Enhanced Case Library

Mech-Vision 2.2.0 added the following cases:

Case Category	Case Name
Hands-on Examples - 3D Locating	3D Trajectory Generation, Clothing Bin Picking, 3D Bin Recognition (Standard Bins), 3D Bin Recognition (Other Bins)
Hands-on Examples - 2D Locating	2D Target Object Recognition (Positioning and Picking), 2D Target Object Recognition (Placement Correction)
Hands-on Examples - Deep Learning	2D Target Object Recognition (Information Reading), 2D Target Object Recognition (Error-Proofing Check)
Typical Cases - Randomly-Stacked Part Picking	Automotive Sheet Metal Parts Loading
Typical Cases - Locating and Assembly	Double Hook Loading, Composite Locating and Assembly, Automotive Windshield Assembly

Case Category

Case Name

Hands-on Examples - 3D Locating

3D Trajectory Generation, Clothing Bin Picking, 3D Bin Recognition (Standard Bins), 3D Bin Recognition (Other Bins)

Hands-on Examples - 2D Locating

2D Target Object Recognition (Positioning and Picking), 2D Target Object Recognition (Placement Correction)

Hands-on Examples - Deep Learning

2D Target Object Recognition (Information Reading), 2D Target Object Recognition (Error-Proofing Check)

Typical Cases - Randomly-Stacked Part Picking

Automotive Sheet Metal Parts Loading

Typical Cases - Locating and Assembly

Double Hook Loading, Composite Locating and Assembly, Automotive Windshield Assembly

Enhanced Target Object Library Picking Configuration Workflow

Mech-Vision 2.2.0 optimized the target object library picking configuration workflow as follows.

Target Object Configuration Workflow	Optimization Details
Import STL File	When editing point cloud models, support "Reset point cloud model to original point cloud". When setting pick points, the tool no longer auto-generates default pick points. When setting collision models, support configuring whether to display the collision model.
Get Point Cloud by Camera	When editing point cloud models, support "Reset point cloud model to original point cloud". When setting pick points, the tool no longer auto-generates default pick points. When setting collision models, support configuring whether to display the collision model. When setting collision models, support generating point cloud cubes based on point cloud models.
Create Common 3D Shape	When editing point cloud models, support "Reset point cloud model to original point cloud".
Jog Robot and Get Point Cloud	When teaching pick points, support capturing background. When setting ROI and removing background, support background-removal-only option. When editing point cloud models, support "Reset point cloud model to original point cloud". When editing point cloud models, support replacing point cloud models with STL models. Added manual pose compensation feature. When setting collision models, support configuring whether to display the collision model. When setting collision models, support generating point cloud cubes based on point cloud models.
Import Processed Point Cloud - Obtain Pick Points from Steps	When editing point cloud models, support "Reset point cloud model to original point cloud". When editing point cloud models, support replacing point cloud models with STL models. Added manual pose compensation feature. When setting collision models, support configuring whether to display the collision model. When setting collision models, support generating point cloud cubes based on point cloud models.
Import Processed Point Cloud - Edit Manually to Set Pick Points	When editing point cloud models, support "Reset point cloud model to original point cloud". When setting pick points, the tool no longer auto-generates default pick points. When setting collision models, support configuring whether to display the collision model. When setting collision models, support generating point cloud cubes based on point cloud models.
Import Processed Point Cloud - Jog the Robot to Set Pick Points	When editing point cloud models, support "Reset point cloud model to original point cloud". When setting pick points, the tool no longer auto-generates default pick points. When setting collision models, support configuring whether to display the collision model. When setting collision models, support generating point cloud cubes based on point cloud models.

Target Object Configuration Workflow

Optimization Details

Import STL File

When editing point cloud models, support "Reset point cloud model to original point cloud".
When setting pick points, the tool no longer auto-generates default pick points.
When setting collision models, support configuring whether to display the collision model.

Get Point Cloud by Camera

When editing point cloud models, support "Reset point cloud model to original point cloud".
When setting pick points, the tool no longer auto-generates default pick points.
When setting collision models, support configuring whether to display the collision model.
When setting collision models, support generating point cloud cubes based on point cloud models.

Create Common 3D Shape

When editing point cloud models, support "Reset point cloud model to original point cloud".

Jog Robot and Get Point Cloud

When teaching pick points, support capturing background.
When setting ROI and removing background, support background-removal-only option.
When editing point cloud models, support "Reset point cloud model to original point cloud".
When editing point cloud models, support replacing point cloud models with STL models.
Added manual pose compensation feature.
When setting collision models, support configuring whether to display the collision model.
When setting collision models, support generating point cloud cubes based on point cloud models.

Import Processed Point Cloud - Obtain Pick Points from Steps

When editing point cloud models, support "Reset point cloud model to original point cloud".
When editing point cloud models, support replacing point cloud models with STL models.
Added manual pose compensation feature.
When setting collision models, support configuring whether to display the collision model.
When setting collision models, support generating point cloud cubes based on point cloud models.

Import Processed Point Cloud - Edit Manually to Set Pick Points

When editing point cloud models, support "Reset point cloud model to original point cloud".
When setting pick points, the tool no longer auto-generates default pick points.
When setting collision models, support configuring whether to display the collision model.
When setting collision models, support generating point cloud cubes based on point cloud models.

Import Processed Point Cloud - Jog the Robot to Set Pick Points

When editing point cloud models, support "Reset point cloud model to original point cloud".
When setting pick points, the tool no longer auto-generates default pick points.
When setting collision models, support configuring whether to display the collision model.
When setting collision models, support generating point cloud cubes based on point cloud models.

Enhanced Advanced Component Steps

Mech-Vision 2.2.0 optimized advanced component Steps as follows.

Step	Optimization Details
3D Target Object Recognition	In the point cloud preprocessing workflow, support retrieving the highest-layer point cloud.
Adjust Poses V2	Added "Add Default Strategy" button in step parameters to create new default pose processing strategies. Enhanced configuration wizard: Optimized and consolidated pose adjustment items for improved usability. Support 2D pose sorting. Support pose filtering based on highest-layer poses.

Step

Optimization Details

3D Target Object Recognition

In the point cloud preprocessing workflow, support retrieving the highest-layer point cloud.

Adjust Poses V2

Added "Add Default Strategy" button in step parameters to create new default pose processing strategies.
Enhanced configuration wizard:
- Optimized and consolidated pose adjustment items for improved usability.
- Support 2D pose sorting.
- Support pose filtering based on highest-layer poses.

Enhanced Other Steps

Step	Optimization Details
Output	Support outputting trajectory-type vision results and bin information. When output type is "Custom", collision detection settings are no longer supported. When output type is "Picking-type Vision Result", support customizing port names. Optimized parameter names.
3D Matching	Adjusted step parameter order to improve ease of use.
Save Images	Support asynchronous saving. After enabling the "Asynchronous Save" parameter, saving will execute asynchronously in the background. Project completion status will no longer wait for this step to finish.
Generate Target Object Picking Strategy	Step renamed to "Acquire Target Object Information". No longer supports generating target object picking strategies.

Step

Optimization Details

Output

Support outputting trajectory-type vision results and bin information.
When output type is "Custom", collision detection settings are no longer supported.
When output type is "Picking-type Vision Result", support customizing port names.
Optimized parameter names.

3D Matching

Adjusted step parameter order to improve ease of use.

Save Images

Support asynchronous saving. After enabling the "Asynchronous Save" parameter, saving will execute asynchronously in the background. Project completion status will no longer wait for this step to finish.

Generate Target Object Picking Strategy

Step renamed to "Acquire Target Object Information".
No longer supports generating target object picking strategies.

Enhanced Step Library

Mech-Vision 2.2.0 optimized the step library, including optimizing some step names and removing some steps, as follows.

Optimized Step Names

Mech-Vision 2.2.0 optimized the following step names:

Original Name	New Name
Generate Target Object Picking Strategy	Acquire Target Object Information
Filter	Filter Data Based on Boolean Value
Binary Classification of Numerical Values Based on Threshold	Determine Whether Numerical Value Exceeds Threshold
Read Object Dimensions	Quick Create Target Object Dimensions
Convert Cloud (XYZ-Normal) to Cloud (XYZ-RGB)	Convert Point Cloud with Normal to Colored Point Cloud
Calculate Angle Between Two Vector3D	Calculate Angle Between Two 3D Vectors
Calculate Cross Product of Vector3D	Calculate Cross Product of 3D Vectors
Calculate Dot Product of Vector3D	Calculate Dot Product of 3D Vectors
Calculate Length of Vector3D	Calculate Length of 3D Vector
Calculate Unit Vector of Vector3D	Calculate Unit Vector of 3D Vector
Compose Vector3D from Numerical Values	Compose 3D Vector from Numerical Values
Decompose Vector3D into Numerical Values	Decompose 3D Vector into Numerical Values

Original Name

New Name

Generate Target Object Picking Strategy

Acquire Target Object Information

Filter

Filter Data Based on Boolean Value

Binary Classification of Numerical Values Based on Threshold

Determine Whether Numerical Value Exceeds Threshold

Read Object Dimensions

Quick Create Target Object Dimensions

Convert Cloud (XYZ-Normal) to Cloud (XYZ-RGB)

Convert Point Cloud with Normal to Colored Point Cloud

Calculate Angle Between Two Vector3D

Calculate Angle Between Two 3D Vectors

Calculate Cross Product of Vector3D

Calculate Cross Product of 3D Vectors

Calculate Dot Product of Vector3D

Calculate Dot Product of 3D Vectors

Calculate Length of Vector3D

Calculate Length of 3D Vector

Calculate Unit Vector of Vector3D

Calculate Unit Vector of 3D Vector

Compose Vector3D from Numerical Values

Compose 3D Vector from Numerical Values

Decompose Vector3D into Numerical Values

Decompose 3D Vector into Numerical Values

Removed Steps

Mech-Vision 2.2.0 removed the following steps:

Removed Step	Alternative Step
Predict Grasp Point	Pick Anything V2
Detect and Measure Line	Detect and Fit Line
Fit Line	Detect and Fit Line
Detect and Measure Circle	Detect and Fit Circle
Fit Circle	Detect and Fit Circle
Measure Circle	Detect and Fit Circle
Detect and Measure Oblong Hole	Detect and Fit Oblong Hole
Measure Circle to Circle Distance	Measure Feature-to-Feature Distance
Measure Circle to Line Distance	Measure Feature-to-Feature Distance
Measure Point to Point Distance	Measure Feature-to-Feature Distance
Measure Point to Line Distance	Measure Feature-to-Feature Distance
Measure Point to Circle Distance	Measure Feature-to-Feature Distance
Measure Line to Line Distance	Measure Feature-to-Feature Distance
Measure Angle Between Lines	Measure Angle Between Segments
Calculate Intersection of Line and Circle	Measure Angle Between Segments
Calculate Intersection of Two Lines	Measure Angle Between Segments
2D Camera	N/A
Measure Point to Point Height Difference	N/A
Measure Point to Reference Line Height Difference	N/A
Measure Longest Line	N/A
Caliper Tool	N/A
Detect Vertex	N/A
Convert Circle to 2D Pose	N/A
Screenshot	N/A
Visualize Information on Image	N/A

Removed Step

Alternative Step

Predict Grasp Point

Pick Anything V2

Detect and Measure Line

Detect and Fit Line

Fit Line

Detect and Fit Line

Detect and Measure Circle

Detect and Fit Circle

Fit Circle

Detect and Fit Circle

Measure Circle

Detect and Fit Circle

Detect and Measure Oblong Hole

Detect and Fit Oblong Hole

Measure Circle to Circle Distance

Measure Feature-to-Feature Distance

Measure Circle to Line Distance

Measure Feature-to-Feature Distance

Measure Point to Point Distance

Measure Feature-to-Feature Distance

Measure Point to Line Distance

Measure Feature-to-Feature Distance

Measure Point to Circle Distance

Measure Feature-to-Feature Distance

Measure Line to Line Distance

Measure Feature-to-Feature Distance

Measure Angle Between Lines

Measure Angle Between Segments

Calculate Intersection of Line and Circle

Measure Angle Between Segments

Calculate Intersection of Two Lines

Measure Angle Between Segments

2D Camera

N/A

Measure Point to Point Height Difference

N/A

Measure Point to Reference Line Height Difference

N/A

Measure Longest Line

N/A

Caliper Tool

N/A

Detect Vertex

N/A

Convert Circle to 2D Pose

N/A

Screenshot

N/A

Visualize Information on Image

N/A

Enhanced Production Interface Configurator

Mech-Vision 2.2.0 optimized the production interface configurator as follows.

When configuring execution screens, support selecting "2D Image Visualization" screens to display 2D images and measurement results in the production interface.
Support setting data "save mode" and "subfolder creation method" in general settings.

Enhanced Adapter Program Generator

Mech-Vision 2.2.0 optimized the Adapter program generator functionality. When using the Adapter program generator to obtain robot names from Mech-Viz, it is no longer necessary to set the Mech-Viz project to auto-load.

Enhanced 3D Camera Hand-Eye Calibration

Mech-Vision 2.2.0 optimized 3D camera hand-eye calibration functionality, supporting use of calibration plate/camera offset relative to flange in the set motion path workflow to reduce displacement in camera view during rotation and ensure the calibration plate remains within the camera field of view.

Enhanced Auxiliary Tools

Mech-Vision 2.2.0 optimized some auxiliary tools as follows.

Tool	Optimization Details
Deep Learning Model Package Management Tool	Support managing more types of model packages, including fast positioning, text detection, text recognition, unsupervised segmentation, and multi-model packages. Optimized tool interface layout and interaction for improved user experience.
Parameter Recipe	Optimized tool interface and access point for improved user experience.
Data Storage	Optimized tool interface and access point for improved user experience. Support configuring "save mode", "subfolder creation method", and "save rules when disk space alert triggers".
Scene Point Cloud	Optimized tool interface and access point for improved user experience.

Tool

Optimization Details

Deep Learning Model Package Management Tool

Support managing more types of model packages, including fast positioning, text detection, text recognition, unsupervised segmentation, and multi-model packages.
Optimized tool interface layout and interaction for improved user experience.

Parameter Recipe

Optimized tool interface and access point for improved user experience.

Data Storage

Optimized tool interface and access point for improved user experience.
Support configuring "save mode", "subfolder creation method", and "save rules when disk space alert triggers".

Scene Point Cloud

Optimized tool interface and access point for improved user experience.

Adjusted Port Data Type Names and Basic Data Type Dimensions

Mech-Vision 2.2.0 adjusted port data type names and the data dimensions of basic data types themselves.

Resolved Issues

Mech-Vision 2.2.0 resolved the following issues:

Renaming a project or solution could fail to save normally if the name ended with two spaces.
When saving a solution, if the temporary solution directory hierarchy is too deep, Windows flickering may occur.
When opening a solution where multiple projects use different parameter sets for the same camera, there is a small probability that parameter sets will be incorrectly modified.
When triggering project execution using standard interface, there is a small probability of delay of several seconds before execution begins.
After switching and applying a robot model in robot communication configuration, if the solution is not saved, the switched robot model will not be saved.
There is a small probability of software crash during production interface execution.
In the target object library "Import Processed Point Cloud - Robot Teaching" workflow, if the camera lacks internal and external parameters, the target object library may crash.
In the target object library "Import STL File" workflow, after switching model views and then switching point cloud generation modes, generating target object point cloud from STL file may fail.
During continuous execution, the "3D Matching" Step may occasionally crash.
After enabling the "Consider Holes in Surface Matching" feature in the "3D Matching" Step, OpenCV errors may occur with specific data.
In the "Output" Step, abnormalities may occur when outputting target object pick point information that does not require point cloud models.
After customizing port names in the "Output" Step, switching to "Predefined (Vision Result)" and checking "Other Input", the "Output" Step did not add corresponding input ports.
After setting the port type of the "Output" Step to "Predefined (Vision Result)" and checking "Other Input", Mech-Viz may incorrectly generate multiple target object models when outputting array pick points.
When multiple pick points correspond to the same target object center point, the "Output" Step output duplicate target object information.
Before the first execution of the "Read Image" Step, after setting "Read Mode" to "Repeat Single Image", images could not be read normally.
Saving images with the "Save Images" Step took a long time.
Opening the deep learning model package management tool caused the software to be unresponsive.
After deleting a Procedure containing custom alert settings, enabling custom alerts may cause the software to crash.
After enabling or disabling the custom alert feature, the settings did not take effect.