Mech-Vision 1.7 Release Notes

You are viewing an old version of the documentation. You can switch to the documentation of the latest version by clicking the top-right corner of the page.

This document introduces the new features, improvements, and resolved issues of Mech-Vision 1.7.

Mech-Vision 1.7.5 Release Notes

This document introduces the improvement, resolved issues, and known issues of Mech-Vision 1.7.5.

Improvements

The following improvements have been made to Mech-Vision 1.7.5:

  • From Mech-Vision 1.7.5, links in the software will redirect to the new Documentation Center.

  • In the menu of the Help option in the menu bar, a new entry for the Online Community has been added. You can access Forum, enter the Download Center, view video tutorials, and provide feedback through this entry.

  • The “Pick Packages from Assembly Line” solution is removed from the Solution Library.

  • The “Re-calculate the Compensation Parameters” feature is no longer supported during calibration. Please contact Mech-Mind engineers if you want to use this feature.

  • The “Measurement Mode” is no longer supported in Mech-Vision 1.7.5. Please contact Mech-Mind engineers if you want to use this feature.

  • The Camera Config Group parameter in the “Capture Images from Camera” Step will be displayed as frequently used parameters. You do not need to click Show all parameters to view this parameter.

  • When the “Predict Pick Points (Any Objects)” Step is run, if the bin is empty, the Mech-Vision project will not be interrupted.

Resolved Issues

The following issues have been resolved in Mech-Vision 1.7.5:

  • When the camera failed to be connected in the “Capture Images from Camera” Step, the error code was missing.

  • In the Configuration before Calibration window, where the movement of the gantry robot that affects the camera position should be selected, the configuration did not take effect after Translation along the Z-axis was selected.

  • In the Configuration before Calibration window, where the movement of the gantry robot that affects the camera position should be selected, after selecting No movement affecting the camera position, the robot flange pose still needed to be input when running the project.

  • When using Matching Model and Pick Point Editor to open a corrupted point cloud file, the software crashed.

  • After creating and saving a point cloud model using the Matching Model and Pick Point Editor, there was an extremely rare chance of the point cloud model file being lost.

  • The first boolean value in the filtering result output by the “Path Planning” Step was consistently “true.”

  • If the path planning tool was not opened, and the robot model was switched in the “Robot Communication Configuration” window in the toolbar, the soft limits of the robot were not reset properly.

  • Other software crashes with small possibility.

Known Issues

  • The text recognition, text detection, and unsupervised segmentation models trained with Mech-DLK 2.5.0 or above cannot be used in Mech-Vision 1.7.5. If you want to use these models, please use Mech-Vision 1.8.0 or above.

  • When using the fast positioning model trained with Mech-DLK 2.5.0 or above in Mech-Vision 1.8.0 or below, the visualization of inference results will be abnormal.

Mech-Vision 1.7.4 Release Notes

This document introduces the improvements, and resolved issues of Mech-Vision 1.7.4.

Improvements

Added Region-Specific Software License Control

The software license is updated to add the restriction of using the software in regions where the software is purchased.

In Mech-Vision 1.7.4, you can go to Help  About to check the software license and region restrictions.

Removed “System Language” from the Software “Language” Option

From Mech-Vision 1.7.4, “System Language” has been removed from the “Language” option (Settings  Options  Common  Language) of the software. If the operating system language of the computer is not a language supported by the software (English, Simplified Chinese, Japanese, or Korean), the default language when the software is installed for the first time is English.

This improvement has resolved the following two issues:

  • The link redirection failed and Step names were displayed abnormally in the software when the operating system language of the computer was not a language supported by the software.

  • The language pack could not be loaded successfully according to the operating system language of the computer.

Scene Point Cloud Supported Point Cloud List

From Mech-Vision 1.7.4, when setting the scene point cloud for reference on the Project Assistant tab, you can select an output port that provides a list of point clouds.

Remove Some Projects/Solutions from the Solution Library

To ensure the quality and performance of the projects and solutions in the solution library, some projects and solutions have been temporarily removed from the solution library in Mech-Vision 1.7.4. These projects and solutions will be improved and re-uploaded in subsequent versions. The removed projects and solutions are listed in the table below.

Category

Name (Built-in projects or solutions are marked in bold)

Workpiece Loading

Crankshaft (Large), Full Partition, Shallow Bin, Aluminum Ingot Bricks, Crankshafts (Small), Condensers, CV Joints, Copper Wires, Connecting Rods, Chain Links, Compressors, Four-Wall Iron Bin (Good Point Cloud), Gaskets (Small), Gear Shafts, Hubs, Iron Balls, Iron Bin (Four Columns), Incompletely Captured Bin, Iron Bin (Double Side Walls), Leaf Springs, Large Rings, Multiple Partitions, Neatly Arranged Small Workpieces, PVC Panels, Rotors, Ring Gears, Rotating Shafts, Randomly Stacked Small Workpieces (Highly Reflective), Randomly Stacked Small Workpieces (Bolts), Randomly Stacked Small Workpieces (Slightly Reflective), Sleeves, Steel Spacers, Steel Bars, Square Bricks, Valve Connectors

Palletizing & Depalletizing

Barrels

Locating & Assembly

Wheel Hub Locating, Bolt Locating, Car Frame Locating, Car Door Frame Locating, Screw Hole Locating, Wheel Hub Valve Core Locating

Piece picking

Cables, Medicine Boxes

  • If you have applied any one of the above projects or solutions, you can keep working with them.

  • If you have updated the online solution library resources with the software of previous versions and then upgraded the software to Mech-Vision 1.7.4, the tabs of the projects or solutions may remain. However, the removed projects or solutions cannot be used in the software of the new version.

Resolved Issues

Mech-Vision 1.7.4 has fixed the following issues:

  • There was a small possibility that the “Cluster Point Clouds and Output Eligible Point Clouds” Step outputs a wrong number of point clouds.

  • When single-channel images were inputted into the neural network and “Draw Defect Mask on Image” was enabled for the “Deep Learning Model Package Inference” Step, an OpenCV-related error occurred after the project was run.

  • The model package inference time was too long when the “Deep Learning Model Package Inference” Step was loaded with an object detection model package exported by Mech-DLK 2.4.1 with the export parameter “Max num of inference objects” set to 1 and “Hardware type” set to CPU.

  • There was a small possibility that an OpenCV-related error occurred when the “Deep Learning Model Package Inference” Step was used to perform instance segmentation for the input images.

  • The truss robot’s extrinsic parameter files generated after the hand-eye calibration could not be used together with the Steps “Transform Poses for Truss” and “Transform Point Clouds for Truss”.

  • The “Fixed-Point Move” and “Relative Move” Steps in the “Path Planning” Step still sent waypoints to the robot even if the “Whether to send to robot” parameter was set to “Plan and not send”.

Mech-Vision 1.7.2 Release Notes

This document introduces the new features, improvements, and resolved issues of Mech-Vision 1.7.2.

New Features

Medicine Boxes Scenario Is Added to the “Predict Pick Points V2” Step

A new “Medicine Boxes” scenario is added to the “Predict Pick Points V2” Step, which enables to sort medicine boxes that are stacked randomly.

You can find the corresponding “Medicine Boxes” project with the deep learning model in the Solution Library.

New Feature in “Deep Learning Model Package Management Tool”

In the new “Deep Learning Model Package Management Tool”, the model package’s inference efficiency can be configured by adjusting the “Batch size” and “Precision” parameters. Only models exported by Mech-DLK 2.4.1 or above support this feature.

It is recommended to use deep learning model packages exported by Mech-Vision 2.4.1 or above with Mech-DLK 1.7.2 or above.

Some deep learning model packages exported by Mech-DLK 2.4.1 can be used in Mech-Vision 1.7.1 or later versions. However, please pay attention to the compatibility issue.

New Step “Deep Learning Model Package Inference”

From Mech-Vision 1.7.2, “Deep Learning Model Package CPU Inference” and “Deep Learning Model Package Inference (Mech-DLK 2.2.0+)” are merged into one “Deep Learning Model Package Inference” Step.

If projects that are created with the software of previous versions are opened in Mech-Vision 1.7.2, the “Deep Learning Model Package CPU Inference” and “Deep Learning Model Package Inference (Mech-DLK 2.2.0+)” Steps will be automatically replaced with the “Deep Learning Model Package Inference” Step.

This Step performs inference with single model packages or cascaded model packages exported by Mech-DLK and outputs the inference result. This Step only supports model packages exported by Mech-DLK 2.2.0 or above.

From Mech-DLK 2.4.1, model packages can be divided into single model packages and cascaded model packages.

  • Single model package: There is only one deep learning model in the model package, such as an “Instance Segmentation” model.

  • Cascaded model package: Multiple models are cascaded in the model package, and the output result of the previous model is input to the next model. For example, there are two models, Object Detection and Instance Segmentation, in the model package, and the inference sequence is Object Detection  Instance Segmentation. The output of the Object Detection model is input to the Instance Segmentation model.

When a cascaded model package is used for inference in this Step, the “Deep Learning Result Parser” Step can be used to parse the inference result of the cascaded model package.

New Step “Deep Learning Result Parser”

Mech-Vision 1.7.2 has introduced the “Deep Learning Result Parser” Step, which can parse the cascaded model package’s inference result exported by the “Deep Learning Model Package Inference” Step.

New Solution/Projects Added to the Solution Library

The following solution and projects have been added to the Solution Library in Mech-Vision 1.7.2.

  • Solution: Brake Discs (Single Station).

  • Project: Medicine Boxes, General Workpiece Recognition, Randomly Stacked Small Workpieces (Bolts).

If you need to use newly added built-in projects or solutions, please use the latest software.

Improvements

Algorithm Improvements

The algorithms of 3D matching Steps (“3D Coarse Matching”, “3D Fine Matching”, etc) in Mech-Vision 1.7.2 were improved, and the Step running speeds were enhanced.

Updated Step and Parameter Names

The following Step and parameter names have been updated in Mech-Vision 1.7.2.

Mech-Vision 1.7.2 and earlier versions

Mech-Vision 1.7.2

Step Name

Adjust 3D Poses by 2D Poses

Convert 2D Poses to 3D Poses

Parameter Name

Filter Candidate Poses by Specified Axis Angles (in “3D Coarse Matching V2”, “3D Fine Matching”, etc)

Filter Poses by Model Rotation Angle

Mech-Vision 1.7.1 Release Notes

This topic introduces the resolved issues in Mech-Vision 1.7.1.

Resolved Issues

The following issues are resolved in Mech-Vision 1.7.1:

  • The “Interface Service” failed to be started when a Mech-Vision project was opened by double-clicking the .vis project file.

  • The Mech-Vision software failed to jump to the online user manual due to network latency.

  • When the “Capture Images from Camera” Step was used to connect a camera of the LSR or DEEP series, and the checkbox of the “Rectify to Depth Map” parameter was not selected, an error occurred in the visualization of the colored point cloud.

  • After the “Capture Images from Camera” Step was renamed, the selected “Data Path” in the virtual mode failed to take effect.

  • In the visualized configurator interface of the “3D Workpiece Recognition” Step, if a workpiece was selected and then unselected in the Workpiece Library, the workpiece was still selected after the Workpiece Library was opened again.

  • After you opened the visualized configurator of the “3D Workpiece Recognition” Step and exited the configurator when the Step was still running, the software would crash.

  • In the “3D Fine Matching Lite” Step, the surface matching result might be incorrect.

  • After the manipulator type was switched in the Matching Model and Pick Point Editor tool, the axes of the manipulator became thicker.

  • The Set ROI window could not be closed successfully when Mech-Vision was closed.

  • In the calibration process, if there was a loss of partial image information or the camera was disconnected, the software might crash.

  • When the hand-eye calibration was performed when the robot was already connected, the calibration method and robot control method could only be selected after the robot was disconnected.

Mech-Vision 1.7.0 Release Notes

This document introduces the new features, improvements, and resolved issues of Mech-Vision 1.7.0.

New Features

New “Solution” Feature, Supporting Deploying Vision Applications Only with Mech-Vision

Mech-Vision 1.7.0 has introduced the solution function. A solution is a collection of functional configurations and data for robot, communication, vision processing, and path planning required for a vision application.

A solution consists of one or more projects. It is not recommended to continue using projects alone. Projects need to belong to the solution.

With the solution feature, you can finish the vision application deployment in one solution, including selecting the robot, configuring the communication with the robot, building the vision project, and planning the robot path.

  • For the project requirements of obtaining vision results and simple path planning, you can only use Mech-Vision software to complete all the application deployment.

  • When the Standard Interface or Adapter communication option is used, the Mech-Vision project ID called by the robot program will no longer be obtained from Mech-Center, but from the Mech-Vision project list.

  • For earlier Mech-Vision projects, you need to complete project migration after upgrading. For details, please refer to 1.7.0 Project Migration Guide.

Mech-Center’s Communication Service Being Integrated with Mech-Vision

Mech-Vision 1.7.0 has introduced the robot and interface communication function. With this function, you can select the robot, configure the communication without opening Mech-Center and finally allow Mech-Vision to communicate with external devices such as robots successfully.

As the interface communication service has been integrated into Mech-Vision 1.7.0, please note that:

  • If a Mech-Vision solution is used and Standard Interface or Adapter is used for communication, you do not need to open Mech-Center anymore.

    • The interface service will be automatically started when you click the Robot and Interface Configuration button on Mech-Vision’s toolbar, select the desired robot, configure the communication option, and apply the configuration.

    • For ABB, FANUC, YASKAWA, KAWASAKI, KUKA, UR, TM, ELITE, JAKA robots, the software has configured the default communication settings same as those specified in the example programs provided by Mech-Mind for these robots. For robots of other brands, TCP/IP is selected by default.

    • When interface configuration is set, you can select the “Auto enable interface service when opening solution” option. When the solution is opened next time, interface service is started and the project is ready for the communication with the robot side.

    • Communication-related logs will be displayed on the Console tab of the Mech-Vision Log panel.

    • When the Adapter communication is configured and the Adapter project folder is selected, the Adapter project folder will be copied into the solution folder.

    • If the Adapter program in the solution has been modified, you just need to reboot the Adapter program and the interface service for the modification to take effect.

    • You need to disable the “Run Mech-Center at PC startup” option in Deployment Settings of Mech-Center.

  • If you do not use the Mech-Vision solution, or Master-Control communication is used, you still need to use the Mech-Center software.

    • You need to open Mech-Center first, and then open Mech-Vision from Mech-Center. You can configure the communication in the same way as in earlier versions.

    • It is recommended to enable the “Run Mech-Center at PC startup” option for Mech-Center. Mech-Vision should be opened by Mech-Center.

Newly Added “Path Planning” Advanced Component

Mech-Vision 1.7.0 has introduced the Path Planning feature (Step), which plans and outputs the collision-free robot path based on the input vision points. This Step suits for metal workpiece loading and unloading scenarios, and can meet project requirements for simple path planning.

Newly Added “3D Workpiece Recognition” Advanced Component

Mech-Vision 1.7.0 has introduced the “3D Workpiece Recognition” Step, which has integrated the vision processing functions such as point cloud preprocessing, 3D matching, and removing overlapped objects. It can significantly accelerate workpiece recognition. This Step suits for metal workpiece loading and unloading scenarios, and can recognize workpieces in different shapes and different stacking manners.

New “Welcome” Interface

Mech-Vision 1.7.0 has added the “Welcome” interface. It provides the software version information, links to help documents, and quick accesses to common operations.

Solution Library

Mech-Vision 1.7.0 has introduces the solution library feature. The solution library provides solution or project cases for five major application scenarios, namely workpiece loading, depalletizing and palletizing, locating and assembly, piece picking, and quality inspection. Also, example data is provided for learning. Beginners can select the desired project or solution according to the case introduction and pictures, and use it for deployment after simple modification.

Mech-Mind will expand the online solution library from time to time. You can always obtain the latest solution library data by one-click refreshing operation. You can download the project or solution as required.

New “General Workpiece Picking” Solution Template

Mech-Vision 1.7.0 has added the “General Workpiece Picking” solution template to the solution library. It can be used to recognize workpieces in different models and different stacking manners and guide the robot to pick workpieces without collision. You can build a 3D vision solution by using only four Steps. It is applicable to scenarios of workpiece loading and unloading for machine tool processing, and workpiece handling.

Algorithm Improvements

Mech-Vision 1.7.0 has added the following Steps to improve algorithms.

New Step Description

3D Fine Matching Lite

This Step provides a lite version for Step “3D Fine Matching”. It eases the parameter adjustment, matches the point cloud models with scene point cloud accurately, and outputs poses of target objects.

Remove Overlapped Objects V2

This Step removes the vision recognition result of overlapped objects according to user-defined rules. Compared with Step “Remove Overlapped Objects”, this Step accelerates the processing of the projection method.

Large Workpiece Measurement Procedures

For the large workpiece measurement industry, several Procedures are provided to facilitate the rapid construction of simple measurement projects on site.

Predict Pick Points V2

This Step recognizes the pickable objects based on the 2D images and depth maps and outputs the corresponding pick points.

Calc Normals and Estimate Edges of Point Cloud

This Step is used to calculate the normals, estimate the edge of the object point cloud, and output the edge point cloud.

Transform Point Clouds for Truss

This Step is used to transform the input point clouds to the camera reference frame or the truss system reference frame and then output the transformed point clouds.

Transform Poses for Truss

This Step is used to transform the input pose to the camera reference frame or the truss system reference frame and then output the transformed poses.

Fit Circle

This Step is used to find the fit circle for the points in the input 2D image. It is generally used in measurement scenarios.

Fit Line

This Step is used to find the fit line for the points in the input 2D image. It is generally used in measurement scenarios.

Calc Intersection between Two Line Segments

This Step is used to calculate the pixel-wise coordinate of the intersection between two line segments. It is generally used to locate object feature points in measurement scenarios.

Calc Intersections between Line Segment and Circle

This Step is used to calculate the pixel-wise coordinate of the intersections between a circle and a line segment. It is used to calculate the pixel-wise coordinate of the intersections between a circle and a line segment.

Process 2D Shapes

This Step processes the shapes in the input binary images by the specified method. It is usually used to process object contours to facilitate various calculations in measurement scenarios.

Fill Holes

This Step is used to fill the holes in the non-zero pixel regions in the input binary image. It is usually used for image segmentation. It obtains a complete image of the target area and avoid interference caused by missing images in the hole section.

Evaluate Image Clarity

This Step measures the clarity of the input images according to the specified method. It is usually used to assist the parameter and position adjustment of the camera in measurement scenarios.

Count Color Info

After a color or gray image is input and a color space is selected, this Step will calculate the mean value, standard deviation, maximum, and minimum of the pixel values in the specified channel. It is generally used to evaluate the image color or grayscale in measurement scenarios.

Caliper Tool

This Step is used to detect edge points or edge point pairs along the vertical direction of an ROI that is usually elongated, and output the pixel-wise coordinates of edge points and distances of edge point pairs (if detecting edge point pairs).

From Shape2DList to Pose2DList

This Step is used to assemble a new 2D pose list by taking the lists of X values, Y values, and Theta values (Theta represents the tilt angle) from the three input 2D shape information lists respectively.

Adaption to LNX Cameras

“Laser Profiler” Step has added new camera type “LNXCamera” for supporting LNX cameras.

New Calibration Procedure for Truss Robots

Mech-Vision 1.7.0 has improved the calibration tool to add the calibration procedure for truss robots. It has designed a dedicated calibration procedure for truss robots, allowing you to complete hand-eye calibration for truss robots without complex settings.

In addition, Steps “Transform Point Clouds for Truss” and “Transform Poses for Truss” have been added for truss robots. In a Mech-Vision project, you need to use these steps to calculate dynamic extrinsic parameters.

New “Deep Learning Model Package Management Tool”

Mech-Vision 1.7.0 has added the Deep Learning Model Package Management Tool, which can be used to optimize the deep learning model packages used by Deep Learning Model Package Inference (Mech-DLK 2.2.0+) and Deep Learning Model Package CPU Inference Steps, and manage their operation modes, hardware types and status. In addition, this tool can also monitor the GPU usage of the IPC.

Improvements

Algorithm Improvements

Mech-Vision 1.7.0 has improved the following Steps.

Improved Step Description

Deep Learning Model Package Inference (Mech-DLK 2.2.0+)

Model Package Settings and Defect Judgement Rules Settings (for defect detection) parameter groups are added.

Deep Learning Model Package Inference

Model Package Settings parameter group is added.

Check Pose Repeatability by Statistics

New outlier handling option is added. For detected outliers, you can select either Raise Error on Outlier or Record and Mark Outlier. When you select the “Record and Mark Outlier” method, the outlier in the output text file will be marked red.

Convert Lengths Pixel-Wise to Physical

The Calculate by Calibration option is added to calibrate based on input images and then calculate the physical distance based on the camera’s extrinsic and intrinsic parameters obtained during calibration. The calculation result using this method is more accurate.

Image Thresholding

DualThreshold and DynamicThreshold thresholding methods are added.

Sort and Stratify

Parameter Starting Point of Each Stratum is added. It outputs the sorted data and indices according to the input layer interval and post data to sort.

Step Removal

Mech-Vision 1.7.0 has removed the following obsolete Steps.

Category Removed Step

2D Feature - Others

Detect Rectangles Given Corners and Sizes in Pixels

2D General Processing - Others

Place Polygons

Deep Learning - Novel Objects Picking

Calc Poses from Heat Map of Pickability, Predict Pick Points (Single Object Type) (Please use the new version of the step Predict Pick Points V2), Pixel-Wise Pickability Evaluation

Measuring - 3D Length/Distance

Calc Distances from 3D Points to Intersection of Two Planes, Calc Distances from 3D Points to Plane

Legacy

Visualize Information on Image (Please use “Visualize Information on Image” instead), Adjust Targets by Obstacles

Pose - Adjust Translation & Orientation

Adjust Targets by Obstacles V2

Others

Detect Pickable Rectangles

Improved Matching Model and Pick Point Editor

Mech-Vision 1.7.0 has improved the Matching Model and Pick Point Editor tool.

  • Improved the main interface to highlight the main functions and improve ease of use.

  • Improved the toolbar layout and adding the GIF tip.

  • Improved the workflow of “Capture point cloud” to support capturing edge point cloud.

  • Improved the unit selection for importing CAD files.

Removed Model Editor Tools (old)

In Mech-Vision 1.7.0, the Model Editor of old version has been removed. If you want to make a point cloud model and generate pick points, please use Matching Model and Pick Point Editor instead.

Improved Calibration Procedure for 6-axis Robots

Mech-Vision 1.7.0 has improved the calibration procedure for 6-axis robots.

  • For pre-calibration configuration, the functions of selecting the robot, robot integration, and connecting to the robot are added. You can establish the connection to the robot during pre-calibration configuration.

  • Scene Viewer (robot simulation interface) is added. During calibration, you can view the calibration path and the robot position in real time without opening Mech-Viz.

  • The interaction interface is improved to make the layout more reasonable and guide the calibration task more clearly.

Improved Parameter Recipe

Mech-Vision 1.7.0 has improved the parameter recipe function. Now, you can synchronize the current parameter settings of the current to the parameter recipe by just clicking the Update Parameter button.

Improved Project Configuration Pane

Mech-Vision 1.7.0 has improved the interaction design for the Project Configuration Pane. Canvas grid and alignment functions are added. You can choose whether to display the alignment grid according to your preference, adjust the reference lines, and drag aligned Steps.

Supporting Displaying Camera Custom Name

After a camera is connected by Mech-Vision, you can view the custom name and IP address of the camera by hovering the cursor over the camera ID.

Changed Global Default Length and Angle Units

The global default length and angle units have changed from “Use built-in units of Steps” to “mm” and “°”.

Supporting Updating Japanese and Korean Language Packs Online and Offline

Mech-Vision 1.7.0 supports updating the Japanese and Korean UI language packs online. In addition, you can obtain the offline Japanese and Korean UI language packs from Mech-Mind Technical Support and drop them into the software for update.

Resolved Issues

Solved Connection Issue with the “Capture Images from Camera” Step

Mech-Vision 1.7.0 has solved the issue that the Capture Images from Camera Step could not connect to Hikvision 2D cameras.

Solved the Issue with the “Calc Results By Python” Step

Mech-Vision 1.7.0 has solved the following issues of the “Calc Results By Python” Step.

  • This Step may stuck when the project is running.

  • When the data in PoseList format are merged, the quaternion sequence of the output pose is incorrect.

Solved the Issue with the “Adjust Poses” Step

Mech-Vision 1.7.0 has solved the crash issue caused by the Adjust Poses Step when the project was running.


We Value Your Privacy

We use cookies to provide you with the best possible experience on our website. By continuing to use the site, you acknowledge that you agree to the use of cookies. If you decline, a single cookie will be used to ensure you're not tracked or remembered when you visit this website.