AI Powered Aircraft visual inspection — Transcript

Full transcript

1. 0:00Today, damage assessment is still
2. 0:02largely manual. An engineer scans the
3. 0:05surface, flags suspect regions, marks
4. 0:07their locations, and measures each dent
5. 0:10with depth gauges and instruments.
6. 0:13Intel Vision automates this workflow. It
7. 0:16identifies damage with 90% accuracy from
8. 0:19a single 3D scan and estimates dent
9. 0:22depth. This reduces manual effort,
10. 0:25measurement time, and variability. By
11. 0:29automating detection, Intel Vision
12. 0:31improves operational efficiency for MRO
13. 0:34teams and supports quick turnaround
14. 0:36times in critical aircraft on ground
15. 0:39situations, helping reduce potential
16. 0:42revenue loss. Given a 3D scan of the
17. 0:45aircraft surface, the software
18. 0:47automatically locates dents and
19. 0:49calculates their depth from as little as
20. 0:52100 microns to over 1,000 microns. We
21. 0:56start with a highfidelity 3D mesh of an
22. 1:00aircraft panel captured by a structured
23. 1:02light scanner. Subtle dips and bumps
24. 1:05almost invisible to the naked eye become
25. 1:08clearly visible in the mesh. You can
26. 1:10observe a 3D sheet panel and see that
27. 1:13there are small damages in the panel.
28. 1:16These damages we called as dent. The
29. 1:18mesh is then loaded into the ML
30. 1:20pipeline. Automated checks, alignment
31. 1:24and surface fitting run without manual
32. 1:26intervention. Confirmed dent areas are
33. 1:29outlined and extracted as focused
34. 1:31patches. surrounding context is
35. 1:34preserved.