← AI Inspection Atlas

Automotive · Production · Vendor case study

European Tier-1 brake supplier: U-Net vision AI on Jetson Orin finds micro-fractures in cast aluminum calipers at 1.2 m/s conveyor speed

Company / plantUnnamed Tier-1 automotive supplier (Europe), braking systems and cast aluminum components
Defect typeMicro-fractures (hairline cracks), casting pores and scratches
PartBrake calipers (cast aluminum)
MaterialPolished cast aluminum
VendorYalantis (custom AOI system)
Camera / lightingMulti-angle dome lighting with polarizing filters; high-resolution monochrome sensors with telecentric lenses; IP67 enclosure
Training dataThousands of labeled images from the client's line, augmented with synthetic (rendered) images of rare defect types
Reported result (quoted)Vendor-reported: "99.7%" defect detection accuracy (against a 99% KPI), "15% increase in line speed" and "8-month ROI." Inference "under 45ms" per frame on NVIDIA Jetson Orin; the detect-to-reject cycle "happens in under 50ms" (OPC UA to a pneumatic divert arm). Before, micro-fractures escaped inspection "at a 0.8% rate."
NotesYalantis case study (page dated Feb 24, 2026), summarized by Yalantis CTO Mykhailo Maidan in Embedded.com (Nov/Dec 2026 ESD eBook): https://www.embedded.com/how-to-make-embedded-computer-vision-work-in-the-real-world/ . A generic off-the-shelf vision system had failed because of glare from polished aluminum and changing foundry light. Partnership since March 2024; the build took 10 months. Customers are German and American OEMs. Vendor-published results, not independently verified.
Defect categoriesCasting & machining, Cracks, Surface scratches/dents
Sourcehttps://yalantis.com/works/automotive-aoi-vision-system-case-study/

Results are quoted from the source. Vendor-published results are vendor claims and were not independently verified. Spot an error? Contact us.