Case Study

Installation at the powertrain production line of
a Tier 1 automotive parts manufacturer

Our client’s problem

“With a large volume of production each day, it takes both physical and mental strength to maintain quality without dropping production rates. Inspectors are constantly working under great pressure.”

Company A

Company A is one of the world’s leading manufacturers of automotive powertrain components, and they needed to automate their inspection process. Inspections relied on human senses (sensory inspection), and it was very difficult for skilled workers to pass on their skills to others, particular skills related to visual inspection. Considering the expected labor shortage due to the declining birthrate in Japan, automating visual inspections was an important goal for the company.


Under the above circumstances, Company A made contact with several AI companies and began considering what to do. Adacotech was contacted about two years ago through its website. While other AI companies produced relatively good results during the POC phase, when it came to actual implementation on the production line, no company was able to achieve high detection accuracy within the given tact time, because of various disturbances in the factories. In contrast, Adacotech was able to generate an accurate model that met the standards for implementation. Because Adacotech requires only normal data and the learning process takes less than 30 minutes , Company A was able to iterate through trial and error multiple times per day. This allowed them to reach the level where the solution could be applied to the actual production line.

“With Adacotech, from the beginning we got good numbers, but not good enough for use the actual production line. However, because the processing is fast and the logic is transparent and understandable, we were able to iterate through trial and error in short cycles, and at a certain point, we felt confident about implementing their solution. Thanks to Adacotech’s lightweight AI, we were able to reach mass-production standards in a short period of time.”

About a year has passed since the introduction of the system,
and it continues to operate with 0% false negative and false alarm rates on the production line.

Company A

False Negative 0% / False Alarm Rate 0%