# Problem context

• Project type: Regular rectangular workpieces, using edge matching; slightly large workpieces, and minimal position variation of incoming workpieces.
• Project problem: During edge matching, if the shape of the product is a standard rectangle (Figure 1), the extracted point cloud edge appears as a symmetrical rectangle (Figure 2). This symmetry can lead to reverse matching issues, providing incorrect pose results (Figure 3).
• Solution approach: Disrupt the symmetry of the product’s point cloud to ensure the matching process can determine the correct pose.
Complete workpiece point cloud - subtracted point cloud (obtained by extracting point cloud within the ROI, selecting a portion of the complete point cloud) = Point cloud after breaking symmetry

Figure 1

Figure 2

Figure 3

# Case Explanation

This article takes a certain product-picking shelf project as an example. See the actual object in Figure 1.

1. Obtain the complete workpiece point cloud data: Via Step Extract 3D Points in 3D ROI, get point cloud the entire products.
2. Obtain the subtracted point cloud: By using the ROI to highlight a corner of the workpiece within the complete workpiece point cloud, get the subtracted point cloud

Figure 4
3. Disrupt the integrity of the workpiece’s rectangular point cloud: Using Step Remove Points from Point Cloud, remove the subtracted point cloud from the complete product point cloud, disrupting the symmetry of the workpiece point cloud.
Note:
4. The point cloud in Figure 5 represents the processed workpiece edge.
5. Firstly, align the point cloud to the robot’s reference frame (typically static). Process as outlined. Afterward, transform it to the camera’s reference frame for the matching steps that follow.

Figure 5
6. Force lock pose direction after coarse matching: After coarse matching, lock the direction of the derived pose, and then perform fine matching and eliminate any inversely matched poses (Figure 6).

Figure 6
See the docs for the usage of Step “Rotate Poses’ Axes to Specified Directions under Symmetry Constraints”.
7. Optimize the fine matching step parameters:
To enhance the accuracy of the matched pose and prevent potential reverse matching, perform fine matching three times to sequentially decrease the standard deviation and narrow down output poses, ultimately producing the one most accurate pose (Figure 9).

Figure 8

Figure 9