Welcome & Project Overview

SUMMARY

Manufacturing defect inspection still relies heavily on humans, but visual checks are demanding, hard to train, and can vary with shifting specs. This project develops a human-centric XR training and assistance system that uses data from an existing inspection process to create immersive practice with many defect cases and keep training content continuously updated.

Main points

• Human inspectors remain essential: they decide on rework, catch missed defects, and prevent errors.
• Visual inspection is challenging: high cognitive load, inconsistency across shifts, long training, rare defects, and angle-dependent detection.
• Defect definitions/classifications can change with customer requirements, especially in small-batch production.
• Current training mostly uses defect catalogues (paper or screens) and doesn’t sufficiently build visual acuity.
• Goal: build a human-centric XR learning app to let workers practice many realistic defect cases efficiently (aligned with XR2LEARN).
• Approach: leverage an existing inspection system’s data/processes to create an end-to-end XR training + assistance setup.
• Pilot aim: create an immersive learning environment with a data pipeline that digitizes processes/defects, converts them into XR-ready modules, and uses SAR to tag new defect cases and feed them back into training for ongoing updates.