Funded by
Within the industrial production of the automotive sector, quality control of the wheels represents a critical phase to ensure safety, reliability, and performance. In the specific case of Honda, the quality control line for the wheels was not yet operational at the start of the project. This posed the challenge of developing an effective quality control system in the absence of a real dataset, typically derived from data acquired in-line.
To address this need, the project involved the development of a vision system based on artificial intelligence systems capable of performing advanced quality control on the wheels: specifically on the presence/absence of bolts on the disc and the orientation of the disc and tire. In the absence of real data, a synthetic dataset was constructed by generating images and defect scenarios through simulations and digital models, thus automatically obtaining the labels of the components to be checked. AI algorithms based on deep learning were trained on this data to detect typical production anomalies and defects. The system was designed to integrate natively with the future production line, ensuring its operational readiness from the first start-up.
The project demonstrated that the use of synthetic datasets can represent a concrete and scalable alternative for training AI models when a production line is not yet available. The solution allows Honda to implement a quality control system that is already mature and adaptable from the start of the plant’s operation, reducing development times and optimizing performance from the beginning of production. Furthermore, the method can be reused for future lines or new models, reducing costs and increasing efficiency in production deployment.
Funded by
Don’t miss any updates on tenders, job openings, and CIM news directly in your feed.
"*" indicates required fields