Maddox AI automates crack inspection of screw heads reliably and in real time. Even fine defects are detected precisely, while false rejects and manual re-sorting effort are significantly reduced.
Product
Industry
Total savings / year
Our customer inspects various screw heads for cracks and other surface-level irregularities. The previously used rule-based camera system led to increased false rejects, additional manual re-sorting effort, and inconsistently stable inspection decisions due to changing surfaces, reflections, and production-related variances.
The goal was to reduce false rejects, significantly ease the burden on re-sorting, and establish a clear OK/NOK logic. At the same time, the inspection had to be reliably integrated into the existing production line process within a cycle time of 250 ms. With Maddox AI, different screw types are now inspected automatically and robustly without affecting production cycle time.
The existing rule-based camera system frequently classified non-critical deviations as defects. This resulted in unnecessary scrap.
With Maddox AI, real cracks are reliably detected, while non-critical deviations are far less likely to be classified as defects.
Suspicious parts had to be regularly re-sorted and checked manually. This tied up quality assurance capacity.
The reduced misclassification eases the burden on re-sorting and significantly lowers inspection effort during operation.
With changing surfaces and reflections, the system could not always reliably distinguish between OK and NOK. This also led to customer complaints.
The AI-based inspection makes stable inspection decisions within the required cycle time and can be flexibly transferred to additional screw variants.
The existing rule-based camera system frequently classified non-critical deviations as defects. This resulted in unnecessary scrap.
Suspicious parts had to be regularly re-sorted and checked manually. This tied up quality assurance capacity.
With changing surfaces and reflections, the system could not always reliably distinguish between OK and NOK. This also led to customer complaints.
With Maddox AI, real cracks are reliably detected, while non-critical deviations are far less likely to be classified as defects.
The reduced misclassification eases the burden on re-sorting and significantly lowers inspection effort during operation.
The AI-based inspection makes stable inspection decisions within the required cycle time and can be flexibly transferred to additional screw variants.
Fewer false rejects
Cycle time per part
Less manual re-sorting effort
Would you like to learn more about our solution? – Test for free now!
We are ISO 27001 certified.
