# Problem context

• Project type: Recognize and pick horizontal plate workpieces using edge matching; workpieces have long horizontal sides.
• Project issues: For horizontal plate workpieces similar to those in Figure 1, with a substantial horizontal surface, surface matching doesn’t perform well. Edge coarse matching presents a likelihood of X-axis deviation (see Figure 3), leading to either no results in fine matching or some discrepancies.

Figure 1

Figure 2

Figure 3

# Solution

1. Extract the complete partition board point cloud: Start by extracting a partition board. Then, through point cloud processing, extract a somewhat complete portion of the flat point cloud from the middle of this horizontal plate. (Note that the extracted point cloud must be complete in the horizontal direction.)

Figure 4
2. Calculate the center pose of the workpiece: Calculate the pose of this flat point cloud to determine the pose of the center of the horizontal plate.

Figure 5
3. Calculate the pose’s horizontal distance deviation: Next, perform coarse matching with the complete edge of the workpiece. Compare the pose obtained from the coarse matching with the pose of the flat point cloud from step 2 to calculate the distance between the poses, obtaining an approximate horizontal distance deviation value.

Figure 6
4. Correct the pose to the center of the workpiece: Move the pose obtained from coarse matching along the X-axis both forward and backward by this deviation value. The aim is to expand to three poses. If there’s a horizontal deviation in coarse matching, this step helps correct it, ensuring that at least one of the poses aligns with the center of the horizontal plate.

Figure 7
5. Output the corrected pose for fine matching: Finally, take the pose obtained from coarse matching and the poses obtained from moving left and right, resulting in a total of three poses, and input them into fine matching. This ensures that the pose derived from fine matching has no horizontal deviation.