I got some Flexible workpiece the whole workpiece is not in the same shape for maching it was twisting
or deform can how can in do the 3D maching or can I just segment the point cloud then do a multiple 3D maching
For flexible workpieces, it depends on how flexible the workpiece is.
Let’s take rubber components as an example. Generally, rubber components are not particularly soft and only have some slight bending at the edges. In this case, we can select a small feature near the center of the rubber component (a small region with minimal changes) and use 3D matching for recognition.
However, if the workpiece is extremely flexible and even folds, it becomes difficult to recognize, as the pose of the workpiece in the bin becomes more complex. In such cases, we recommend using deep learning to extract and recognize the plane in the identified region for grasping. Alternatively, picking-anything techniques from Mech-Mind can also be employed.
Certainly, for soft workpieces, we recommend using deep learning for preliminary extraction and positioning of the workpieces. There are two reasons for this:
The Z-axis pose of the software artifact is difficult to determine., making it challenging to provide accurate results for overlaping detection. In such cases, deep learning can be used to recognize only the non-overlapped workpieces.
The features of flexible workpieces may vary with their poses. Preliminary extraction and positioning can aid in better recognition and improve efficiency. Using deep learning can reduce recognition errors and enhance stability in recognition.