The first use of computer vision on the shop floor is almost always inspection: a camera observes the product and separates conforming parts from defective ones. That is useful, but it leaves the most important question open: why did the defect form?
From defect to cause
When images are linked to machining parameters, material batches and machine conditions, the vision system becomes an analytical tool. It becomes possible to see which parameter combinations precede a rise in scrap and to intervene before the problem repeats.
Vision and simulation
In some processes, observation alone is not enough. Combining high-resolution imaging, process data and simulation models makes it possible to compare what is observed with what the model predicts, validate a new configuration and reduce the number of physical trials.
What it takes to work
- stable, repeatable acquisition conditions;
- process data synchronised with images;
- acceptance criteria shared by quality and production;
- a clear way to handle uncertain cases.
Vision technology is mature. The value depends on how it is connected to the process and to the decisions of the people who run it.
