Extraction, Segmentation, and 3D Reconstruction of Wire Harnesses from Point Clouds for Robot Motion Planning

Area of Science:

• Robotics
• Computer Vision
• Geometric Modeling

Background:

• Accurate robot simulation requires modeling flexible components like wire harnesses.
• Existing methods lack precise 3D models for wire harnesses, leading to simulation-reality discrepancies.
• Wire harnesses are often omitted in CAD models, hindering collision detection and safety.

Purpose of the Study:

• To develop an automated framework for extracting, segmenting, and reconstructing 3D wire-harness models.
• To enable realistic motion simulation by distinguishing static and dynamic harness parts.
• To improve collision detection and digital-twin accuracy in robot simulation.

Main Methods:

• Automated framework using terrestrial laser scanning point clouds.
• Motion-aware segmentation classifying harnesses into static and dynamic components.
• Dual reconstruction: OBB-tree for unbranched cables and Reeb-graph for branched structures.

Main Results:

• High-fidelity 3D wire-harness models generated from dense, occluded point clouds.
• Successful reconstruction even with severe data occlusions.
• Demonstrated suitability for collision detection and digital-twin simulation on industrial robots.

Conclusions:

• The proposed framework effectively addresses the challenge of modeling flexible wire harnesses in robot simulation.
• This work bridges the gap between geometric sensing and physics-based robot simulation.
• Enables more reliable and safer robot operations in manufacturing environments.