One of the great paradigm changes brought on by Industry 4.0 has come through the coupled advances of sensing, processing, and machine communication capabilities. As manufacturers transition to smart quality control processes, they should begin to investigate how available technology can make quality control not only faster and independent, but also more holistic.
The Problem with Machine Vision Today
It is undeniable that Machine Vision solutions currently in the market have sped up the visual inspection process. If the algorithm is good, they may even be more efficient at identifying visual range defects and contaminants than human inspectors. But for a field that prides itself on presenting a cutting-edge image, solutions in the market are hard to differentiate. Specifically, they all share the same serious technical limitation: they are only scratching the surface of the product. Most products, be they fast-moving consumer goods or cutting-edge aerospace components, are too complicated to be inspectable only visually. The consequence of this in a manufacturing context is that even if you, the maker, can make the visual inspection process faster, you will still face bottlenecks down the road, when you perform inspections on non-visual parameters, like weight. Information is everything, but speed of information is a key constraint in manufacturing. When you have black boxes in your quality control processes because you lack the technology to perform tests non-destructively, at the required speed, you are playing Russian Roulette with every batch. To be clear, this problem does not result from lack of acumen in the factory, but rather from a lack of solutions that address this problem. Even though machine vision promises to be augmentative, the reality is that players are doing little more than offering a faster, but just as flawed alternative to a human inspector. To truly assess a product’s quality in a fast-moving production line, it is not enough to have a camera that looks at the object fast, but rather, to have a camera that can look through the object, fast.
Redefining Machine Vision
These pain points highlight a problem in the linguistics of the term machine vision, which have pidgeonholed manufacturers into thinking in optical terms. TiHive started in 2018 with the goal of finding a solution for this problem, filling in the holes that currently exist in quality monitoring. To do this, we redefined the industry’s definition of vision to include Terahertz. Terahertz is an electromagnetic frequency that can travel through objects, creating a distinct signature for each of its material components. Like the x-ray machine at an airport, these contrasts create an image that can be interpreted by a deep learning algorithm, or a human operator. But unlike x-rays, TiHive terahertz technology is completely non-ionising and safe. TiHive’s terahertz sensors capture images of the product at a very fast (5000fps) rate, meaning they can meet production lines speed requirements. The best part is that with AI, this all happens automatically, on the fly.
What the Terahertz revolution means for manufacturers
Today, manufacturers can see the surface of what they are making. Soon, with TiHive Terahertz, they will be able to see everything. This means that they will be able to get formerly unknowable data on product material specifications and detect contaminants. Seamlessly, inline, quickly, nondestructive. This will mean a tighter control on process variability, better and more consistent quality reaching the customers, and less waste. Technologically, the shift from optical to terahertz in quality process monitoring will be like shifting from Clark Kent to Superman. Seeing the invisible is a game changer.