IntelliGuide projects self-explanatory pictograms into the machine operator's work area. The picture shows this: The marked part receives a mitre cut. The operator must feed it to the mitre cutting unit accordingly.

© Homag

If the machine operator inserts the wrong part, for example, the machine either does not start working (LED bar lights up red, machine stops) or - if this is feasible - the saw still starts cutting by simply retrieving the appropriate information and assigning it to the part. For example, the operator can swap the order of the strips when cross-cutting (by mistake or on purpose) - the saw reacts to this and cuts the dimensions belonging to the inserted strip. If the saw splits two strips simultaneously with the 'Power Concept', the operator does not have to pay attention to the order in which the pair is placed against the fence; the control system automatically adjusts to the selected order, provided the widths allow it. If the alignment of the inserted part is not correct, the saw will not cut until the operator has inserted the part correctly.

Multitasking: Even if two strips are cut to length at the same time, the LED light clearly indicates each work step for each strip and each part.

© Homag

In short, the assistance system reacts intelligently and interactively to the operator's actions from the 'Advanced' configuration level upwards. To make this possible, artificial intelligence and machine learning (ML) came into play at this point in the development of IntelliGuide, because: The enormous variance in different decors, colors as well as workpiece dimensions and their combination with different environmental influences would have pushed classic image recognition and evaluation to its limits during development. Firstly, because it is impossible to record and maintain all material data, and secondly, because influencing factors such as hand movements, soiling, labels, foreign objects or even concealment of the parts to be processed make it difficult to define what constitutes a 'workpiece' using conventional programming.

To enable the machine to teach itself what a workpiece is, the Homag engineers used an open-source AI, which they adapted to their needs, as the basis for their programming. To understand artificial intelligence, an analogy can be drawn with humans: In our childhood, we were taught over a long period of time and with many case studies to recognize, identify and classify objects based on unique features. We also learned to assign these objects spatially. This is exactly what we now need to do with AI. In principle, artificial intelligence is an imitation of human thinking: simple algorithms are used to simulate intelligent behavior. The major advantage of this is that the use of ever-increasing computing power and the generation of artificial training data means that this process can be significantly accelerated compared to human learning.

IntelliGuide Professional

The third and highest expansion stage is IntelliGuide Professional. Here, the image analysis is additionally expanded to include a laser projection system: The camera already included in the 'Advanced' expansion stage first provides the images of the machine table, which IntelliGuide then evaluates. This means that the saw always knows where and how the operator is inserting a workpiece. The additional laser projection system then displays self-explanatory pictograms with clear instructions for the operator directly on the workpiece: For example, directional arrows indicating where to move the part or how to turn it if necessary.

Finally, after cutting, laser icons appear with instructions on where to take each part: to the waste garbage can, to a specific stack, to the manual remnants store or whether it needs to be recut.

What happens next?

If storage locations are predefined, numbers appear on the finished workpieces. The operator simply stacks the part on the corresponding pallet.

© Homag

With IntelliGuide, the saw specialists ventured into the field of machine learning for the first time. They took the first steps with an interdisciplinary team. The aim was to quickly go into practice in order to gain important experience. The group also sought external help from AI experts, which proved to be very helpful, especially in the early days.

Machine learning is now one of many tools at Homag that the developers use as a matter of course when required. In addition to pure parts handling, other tasks that need to be performed on the saw will come into focus in the future - including maintenance, tool changes, parameter settings for materials and much more.

The overriding aim is always to simplify the interaction between man and machine to such an extent that the operator is guided through the cutting process in a pleasant, intelligent and efficient manner.

Author: Manuel Friebolin is the developer and project manager of IntelliGuide at Homag.