Skip to main navigation Skip to search Skip to main content

Robotic Fusion: 3D Point Cloud and Ultrasonic Signal Integration for ML-Driven Composite Non-destrcutive Testing and Evaluation

Activity: Talk or presentation › Invited talk

Description

This project developed a multi-stage 3D point cloud segmentation pipeline to enhance automated ultrasonic evaluation of complex composite structures. Back-wall geometric features such as edges, ramps, bends, and thickness transitions can cause signal loss that automated inspection systems misclassify as defects. To address this, the pipeline extracts and classifies geometric primitives from laser-scanned composite surfaces and pairs each ultrasonic inspection point with its corresponding back-wall geometry, producing ML-ready datasets that enable geometry-aware defect detection. The project, a collaboration between Ulster University and the NCC, also delivered a browser-based web application for interactive 3D segmentation analysis and ML dataset export.
Period28 May 2026
Degree of RecognitionNational

Keywords

  • Non-Destructive Testing (NDT)
  • Ultrasonic Inspection
  • 3D Point Cloud Processing
  • Machine Learning
  • Composite Materials
  • Robotic Inspection
  • Primitive Segmentation
  • Aerospace Inspection
  • Automated NDT
  • Structural Health Monitoring
  • Composite Structures
  • Sensor Fusion
  • Point Cloud Segmentation
  • Machine Vision