It sounds simple, but it has a clever purpose: Nestlé produces nutritional drinks and supplements at its Osthofen plant in the Rhineland-Palatinate region. The powder is automatically dispensed into the can. A measuring scoop is also included. A Sick system with a snapshot camera detects whether it has actually been added. This makes sense on the assembly line.
“The human eye can easily detect whether a spoon has been added”says Marcus Kauf, automation engineer at Nestlé, “but at a filling speed of over 80 cans per minute, this is no longer possible without errors.”
Until recently, a camera monitored the application process by counting the colored pixels of the plastic spoon. However, Nestlé has recently switched to colorless portioners to increase recycling rates. So now, grayish, shimmering, transparent spoons rest on aluminum with a similar hue on the lid... or do they? Difficult to identify on the corrugated, embossed, and reflective metal – the classic solution using sensor-based image processing reached its limits.
Photos: Sick/Nestlé
Now, an "artificial intelligence" quickly and reliably learn new patterns continuously by collecting a vast amount of data, which is then directly incorporated into algorithms. Using the integrated deep learning software "dStudio" from the Sick AppSpace, a 2D snapshot camera (picoCam) practices "thinking" The web service allows neural networks "trained" in this case with images of the included spoons in a wide variety of positions. Just as humans ideally solve problems and make decisions, so too does deep learning—only many times faster.
“The learned decision-making algorithm”says expert Klaus Keitel, “is transferred to the camera system. This enables it to independently recognize the essential differences between images.” The camera can also be easily adapted to each new product by training a new neural network for the specific requirements. The intuitive user interface requires no special AI knowledge or image processing expertise from the user.
If the camera "No spoon" , the system stops automatically. Once the missing spoon is added, the camera software recognizes this and allows the system to continue running – without the need for a manual restart.
