Intelligent visual detection methods for enhanced quality

Status today
At ArianeGroup several launcher components are manufactured as piece parts and assembled to sub-systems. On the top, strict requirements must be respected in terms of cleanliness and burrs on piece parts as well as cleanliness during integration operations. Imperfections might not be effectively detected by classical visual detection methods. Also, current automated systems need to be adapted with high efforts to the single tasks, environments, and operations.

Opportunity
Visual detection methods based on artificial intelligence (AI) might support the nominal operations by detecting imperfections/deviations on any level of detail and under varying environmental conditions. Feedback-based algorithms and camera systems might be utilized. In addition, neural networks could be trained and adapted to other optical spectrums than visible light.

Potential Application
A typical application could be during equipment integration, especially in the area of optical quality inspection (e.g. incoming inspection, acceptance testing and final inspection before delivery). In this particular area, the applicability and the performance improvement can be studied and tested in detail.

Scalability
The scalability of AI-based assistance is enormous and comprises many fields of activity within any company. Some further examples within production are “automated optical inspection”, “asset detection and inventory assessment”, “pose estimation”, “activity tracking” or “anomaly detection”.