The carmaker's analysts learn why a part failed in seconds, not hours of expert study.
India's largest automobile manufacturer wanted its failure analysts to find out quickly and consistently why a motor part fractured, including for urgent cases.
What changed
What stood in the way.
- Fracture analysis depended on a few experts and was slow and inconsistent, and urgent requests were hard to handle.
- Telling brittle from ductile fractures and finding the direction a crack travelled was difficult, and the data had to stay inside the client's environment.
- 88%accuracy in failure mode detection (as reported)
- 2.5 to 3 sper prediction (as reported)
- 2 hrssaved per defect, about 60 hours a month (as reported)
What we built.
- A ReactJS and Python portal to upload microscope images, view past reports in PPT and PDF, review drafts and accept or reject pending defects.
- Images labelled in AWS Ground Truth inside the client's environment, sorted by material, magnification from 10x to 200x, grade and fracture type, with new ductile features agreed with domain experts.
- A custom UNet segmentation model on AWS SageMaker plus classification models to tell brittle from ductile fractures.
- Direction marking that starts from the hackles and works back to the origin, so direction is found even when no origin is visible.
- Delivered in 2 months with 43 CMMI Level 5 documents.
- ReactJS
- Python
- AWS Ground Truth
- AWS SageMaker
- UNet segmentation
- Classification models
What this work shows about us.
The labels that worked came from discussions with the client's domain experts, and analysts still accept or reject each result.
Started as a POC and became a full project, with CMMI Level 5 documentation kept throughout.
Want AI image analysis for part failures like this? Tell us what you want to change. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with the right team.
Start your plan ↗If we built it today, with AI.
What we would add now, moment by moment.
An agent would match a failed part returned from the field to similar past cases and draft the failure report for the analyst. The analyst still signs off the cause.
An agent would tell the supplier quality team when one grade of material keeps failing the same way, so they can act before more parts ship. Engineers decide the fix.
Where this work fits.
- Manufacturers & distributorsYour company is meant to get dealers the right stock, on time, without making them chase it.
- Supply chain & assets (inventory, assets, procurement)The right stock in the right place on time, and everyone knows what has been ordered without chasing.
Facts and figures from the project records; nothing added.
Want this for your business?
Tell us what you run or what you're building. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.
Send us a note instead
An AI agent. This prototype uses prepared answers and sends nothing.
We use essential cookies to run this site. With your consent we also use analytics cookies to see which pages help. Privacy policy