Hole Repairing Algorithm for 3D Point Cloud Model of Symmetrical Objects Grasped by the Manipulator

Area of Science:

• Robotics and Mechanical Engineering
• Computer Vision and 3D Reconstruction

Background:

• Manipulator grasping tasks often result in incomplete 3D point cloud models due to mechanical arm occlusion and limited imaging angles.
• Complete 3D point cloud models are crucial for subsequent manipulator task planning and execution.

Purpose of the Study:

• To propose an automated method for detecting and repairing holes in 3D point cloud models of symmetrical objects grasped by manipulators.
• To classify and repair different types of holes, specifically those caused by mechanical arm occlusion and limited imaging angles.

Main Methods:

• Establishment of a virtual camera coordinate system for hole detection.
• Boundary detection, classification, and repair of nested holes.
• Surface reconstruction and utilization of object symmetry for hole repair.

Main Results:

• Successful detection and classification of two types of holes: mechanical claw holes and missing surfaces.
• Effective repair of detected holes using surface reconstruction and symmetry-based methods.
• Demonstrated superior performance of the proposed approach compared to state-of-the-art algorithms on simulated and real data.

Conclusions:

• The proposed method automatically and accurately repairs holes in 3D point clouds of grasped objects.
• This technique enhances the completeness of 3D models, vital for advanced robotic manipulation and task planning.
• The approach offers a significant improvement over existing 3D point cloud hole repair algorithms.