Pose Correction (Assembly)

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

■ If you're unsure about the version of the product you are using, please contact Mech-Mind Technical Support for assistance.

This section describes the Pose Correction (Assembly) function in the Advanced Pose Transformation step and its workflows.

Function Introduction

This function automatically corrects tool pose or flange pose before assembly, based on secondary vision capture after robot has picked the workpiece. It compensates picking error and placement offset to improve assembly precision.

It supports secondary capture with either 3D or 2D camera, and supports both fixed and non-fixed placement scenarios.

Prerequisites and Constraints

Before using this function, make sure the following conditions are met:

  • Coordinate-system constraint

    All poses input to Advanced Pose Transformation must be in robot coordinate system.

    • 3D poses from steps such as 3D Matching are usually in camera frame and should be converted by Transform Poses.

    • 2D poses from steps such as 2D Matching should be converted to 3D robot-frame poses by relevant 2D-to-3D conversion steps.

  • TCP consistency requirement

    Teaching and runtime must use the same TCP configuration. If gripper or TCP changes, teaching must be redone.

  • Secondary vision capture requirement

    Pose correction depends on secondary image capture at a fixed photo point after picking.

    • In 3D-camera scenarios, provide tool pose at secondary photo moment from teaching.

    • In 2D-camera scenarios, provide flange pose at secondary photo moment from teaching.

  • Placement-type definition

    • Fixed placement: placement position and orientation remain unchanged during runtime.

    • Non-fixed placement: placement position/orientation may vary and must be recognized in real time before assembly.

Workflow

The following four scenarios are supported:

3D Camera + Fixed Placement

After picking, robot moves to fixed photo point. 3D camera captures workpiece point cloud and calculates real-time workpiece pose. Corrected assembly tool pose is computed together with taught fixed placement pose.

3d camera fixed placement pose correction

Configure Step Ports

  1. Open configuration wizard and select Pose Correction (Assembly).

  2. Set camera type to 3D Camera and placement type to Fixed Placement.

  3. Click OK.

Configured ports:

Input

Input Port Data Type Description

Tool Pose at Capture

Pose []

Tool pose (robot frame) recorded at secondary photo point during teaching.

Workpiece Pose at Capture

Pose []

Real-time workpiece pose (robot frame) at secondary capture.

Fixed Placement Pose

Pose []

Taught tool pose for placement.

Output

Output Port Data Type Description

Assembly Pose

Pose []

Corrected assembly pose for direct downstream output.

Teaching Phase

  1. Record fixed placement pose in ideal assembly condition.

  2. Record tool pose at secondary photo point and create workpiece point-cloud template.

  3. Connect taught poses to corresponding input ports.

Runtime Phase

  1. Capture workpiece point cloud at secondary photo point.

  2. Match with taught template via 3D Matching.

  3. Convert matched pose to robot frame and input it as Workpiece Pose at Capture.

Output Corrected Assembly Tool Pose

Run project and output corrected assembly pose through Output step.

3D Camera + Non-Fixed Placement

In addition to workpiece real-time pose, obtain real-time placement pose from placement-recognition project and compute corrected assembly tool pose dynamically.

3d camera non fixed placement pose correction

Configure Step Ports

  1. Select Pose Correction (Assembly).

  2. Set camera type to 3D Camera and placement type to Non-Fixed Placement.

  3. Click OK.

Key inputs: Tool Pose at Capture, Workpiece Pose at Capture, Real-time Placement Pose. Output: Assembly Pose.

Teaching Phase

  1. Create placement point-cloud template under ideal setup.

  2. Record tool pose at secondary photo point and create workpiece point-cloud template.

  3. Connect taught pose data.

Runtime Phase

  1. Obtain workpiece real-time pose at secondary photo point.

  2. Obtain real-time placement pose from placement-recognition project.

  3. Input both poses to this step for correction calculation.

To avoid mutual interference in recognition, it is recommended to run workpiece capture and placement capture in separate Mech-Vision projects and pass data via global variables or files.

Output Corrected Assembly Tool Pose

Run project and output corrected assembly pose through Output step.

2D Camera + Fixed Placement

After picking, robot moves to fixed photo point. 2D camera captures image, recognition output is converted to 3D pose, and corrected assembly flange pose is computed with taught fixed placement pose.

2d camera fixed placement pose correction

Configure Step Ports

  1. Select Pose Correction (Assembly).

  2. Set camera type to 2D Camera and placement type to Fixed Placement.

  3. Click OK.

Key inputs: Flange Pose at Capture, Workpiece Pose at Capture, Fixed Placement Pose. Output: Assembly Pose.

Teaching Phase

  1. Record taught fixed placement flange pose.

  2. Record flange pose at secondary photo point.

  3. Create workpiece 2D template at secondary photo point.

  4. Connect taught poses to step inputs.

Runtime Phase

  1. Capture image at secondary photo point.

  2. Match workpiece with taught 2D template.

  3. Convert 2D pose to 3D robot-frame pose and input as Workpiece Pose at Capture.

Output Corrected Assembly Flange Pose

Run project and output corrected assembly pose through Output step.

2D Camera + Non-Fixed Placement

After picking, robot captures secondary image for workpiece pose and also obtains real-time placement pose from placement-recognition project. Corrected assembly flange pose is then computed dynamically.

2d camera non fixed placement pose correction

Configure Step Ports

  1. Select Pose Correction (Assembly).

  2. Set camera type to 2D Camera and placement type to Non-Fixed Placement.

  3. Click OK.

Key inputs:

  • Flange Pose at Capture

  • Workpiece Pose at Capture

  • Taught Placement Pose

  • Taught Placement-position Pose

  • Real-time Placement Pose

Output: Assembly Pose.

Teaching Phase

  1. Record taught assembly flange pose in ideal condition.

  2. Create placement 2D template and obtain taught placement-position pose.

  3. Record flange pose at secondary photo point.

  4. Create workpiece 2D template.

  5. Connect taught data to corresponding input ports.

Runtime Phase

  1. Obtain workpiece real-time pose from secondary image (2D match + 2D-to-3D conversion).

  2. Obtain real-time placement pose from placement recognition (2D match + 2D-to-3D conversion).

  3. Input both real-time poses to this step for correction.

To reduce interference, run workpiece-recognition and placement-recognition in separate Mech-Vision projects and exchange data through global variables or files.

Output Corrected Assembly Flange Pose

Run project and output corrected assembly pose through Output step.

Is this page helpful?

You can give a feedback in any of the following ways:

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.

  • Preference
  • I accept
  • I don't accept