In Norse mythology, the most feared god of them all is Odin. Legend has it that he is not only the mightiest warrior, but also the wisest man. If that name rings a bell – you’re absolutely right. Due to distinctive similarities, Inwatec’s X-ray scanner ODIN is named after precisely this Norse god. As one of Odin’s powers is prophecy, it resembles our ODIN, who, with the help of X-ray vision, always has the insights about what is hidden in the laundry items he is facing in his daily battles.
Taking all the soiled items into account, only around 1-10% of it include potentially harmful foreign objects. Even though this number might seem small, the trouble even one single object can cause, might be extensive. Not only machines are at risk of being damaged (and having to be replaced), also the garments itself could be, e.g., torn or discolored. So what can a laundry do to make sure foreign objects don’t end up in the sorting process in the first place?
Of course employees could be checking every single incoming item by hand. Not only would this be a time consuming and, quite frankly, a rather boring task – it also puts employees in a vulnerable position. Especially in the healthcare sector, sharp and contaminated objects often make their way into the laundry. Tudor Morar, software engineer at Inwatec, recalls the time leading up to one of Inwatec’s first X-ray system in a Danish laundry: “They had the problem that the people who were sorting their garments were getting stung by needles used by hospitals, which wasn’t very pleasant. Employees needed to constantly get tested for a year to see if they infected themselves with a disease or not. So you can imagine how important it was for them to avoid that in the first place, and why they wanted an X-ray scanner”.
With the help of market-leading Artificial Intelligence (AI) ODIN is able to scan and reject items containing harmful foreign objects from the sorting process. This way employees will still have to remove foreign objects by hand, however, thanks to ODINs smart X-ray vision, they know exactly what they have to watch out for. When it comes to sorting linen also other items, besides the “usual” suspects (e.g. pens, needles or scissors), can be found: “it is anything you can think of, from TV remotes, all the way to iPads – that’s the kind of things you tend to forget on a bed”, reports Tudor further. “I guess the nicest thing we found was a credit card. The owner had already contacted the laundry, and we found it”, he continues with a smile.
To achieve the best possible outcome, the machine is being trained and its parameters being optimized in order to serve each laundry’s specific needs. Unlike in the early stages of ODIN’s development, tasks once reliant on image processing methods are now effortlessly handled using deep learning. “We take all these pictures from the laundry and we need to mark them: this should be in the garments and this shouldn’t be. We have millions of samples of ‘accepts’ and ‘rejects’ – that’s what we use to train our neural networks”. But what happens if the system encounters a novel object? “If it sees something new, it will interpret it along the lines if it looks more like an ‘accept’ or ‘reject’”, Tudor further explains. Ultimately it depends on how similar the object is to something the system has seen before. A completely different object, the system probably won’t find it, if it’s similar, however, it will most likely recognize it: “if it’s a lipstick, it looks kind of like a marker pen – so there is a good chance it gets rejected.”
Ultimately the detection quality depends on the specific type of garment being scanned, as well as the items that ODIN is looking for. ODIN excels at finding objects such as pens, scissors, lighters or paper clips. When it comes to items made out of thin plastic or paper, however, the task gets a little bit more complicated: “for the X-ray it is pretty hard to find thin plastic and paper, it is basically transparent. It does show up on the screen, but it looks like the linen itself – it has the same density.” Multispectral X-rays, which are already in use at some airports, work with different bandwidths, and are therefore able to recognize the density of an object. This means the X-ray can tell the operator more about a certain object, besides the fact that it is “just there”. In terms of laundry sorting Tudor adds “be aware that thin plastic might still be a problem, because some clothes are made out of polyester – which essentially is thin plastic.”
When asked about the future of automated laundry sorting, Tudor points out that, compared to the advancements achieved in merely ten years, the future is already here: “if someone now loads a trolley into the system, the next time someone needs to touch the items again is when they are already dry on the clean side. Before that an employee would have had to touch every single incoming piece.” He goes on to say that not only will automated sorting systems become more common, but also much easier to set up: “with all the advancements in deep learning, it will become even easier. And we’ll definitely work on making the systems faster and more efficient”.
In conclusion, ODIN stands as a beacon of innovation in the world of laundry sorting, mirroring the wisdom and foresight attributed to its mythical namesake. By seamlessly integrating market-leading Artificial Intelligence with advanced X-ray technology, ODIN not only enhances safety and efficiency, but also alleviates the task of manual sorting. As ODIN continues to evolve and adapt, driven by the principles of deep learning, it promises a future where sorting processes aren’t just secure, but also streamlined and optimized to meet the unique needs of each laundry. With ODIN leading the charge, the laundry industry embarks on a transformative journey towards a safer, smarter, and more productive tomorrow.