GermanyManufacturing2021 – ongoingHELLA
Neural-network camera inspection of automotive sensors
HELLA reported more than 99% fewer false rejects in rain-light sensor inspection
Problem
Conventional inspection cameras at HELLA Fahrzeugkomponenten in Bremen sometimes rejected good sensors. False rejects wasted components and could stop production.
Approach
Engineers worked with an external camera supplier to train a neural network on images of good and defective parts. HELLA installed 16 AI cameras at inspection stations with high false-reject rates.
Results
Takeaway
The reported gain concerned false rejects, rather than a measured reduction in defective products reaching customers.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Partner co-development
- Evidence
- Company's own claim
Sources
Source summaries
HELLA engineers described training inspection cameras and installing 16 units. They reported more than 99% fewer false rejects for rain-light sensors and investment payback within less than 6 months.
Full details are in the original sources.