Machine Learning: the Genius Behind Our Smart Machines

Machine Learning has become crucial in the ever-evolving landscape of technology. In order for our machines to become smart, however, they need to learn and adapt – much like we humans learn from experiences. This is where our cutting-edge AI software steps in as the brain behind the operation. Its main job is to keep teaching and improving our systems so they can become geniuses at sorting laundry. 

“Even though machine learning is a task our customers can easily do themselves after a little introduction, of course we have a great service team that will be on site to help and advise them during the process. This way the highest level of sorting accuracy for our systems can be ensured”,  says Nicolai Lund Christoffersen, manager of Inwatec’s growing service team. 


For example: think of our HEIMDAL.Camera module as the eyes of the operation. While in operation, it takes thousands of pictures of all different kinds of laundry items. Why, you might ask? Because these pictures are like school lessons (in this case machine learning) for our AI system. The AI uses them to train a so-called neural network. A neural network can be explained as a way for the system to learn and process information in a similar way to how a human brain would do it. Simply put, the HEIMDAL.Camera acts like a teacher, showing our AI system lots of pictures to help it understand and learn. This way the HEIMDAL.Camera is able to sort products based on their visual appearance, including characteristics such as color, pattern, texture and even size.

HEIMDAL.Camera utilizes Artificial Intelligence to categorize products based on visual characteristics. 

“On the one hand our service team is there to help new customers with the initial installation of their sorting systems – meaning they help set up the HEIMDAL.Camera module, so it has a sufficient amount of pictures that enable flawless sorting. On the other hand, however, the service team also pays visits to existing customers to enhance their already installed systems. This way sorting categories can be optimized, double picks (note: when the robot separator picks more than one garment at a time) can be rejected, and the overall sorting accuracy can be raised to the highest level possible, reports Christoffersen further.


To scan and reject garments that contain possibly harmful foreign objects, also our X-ray scanner ODIN makes use of smart AI software. The detection quality ultimately depends on the types of garments being scanned, as well as the items being searched for. ODIN excels at recognizing pens, needles, paper clips and scissors, as well as lipsticks, lighters, sunglasses and the like. A conventional metal detector, in comparison, would for instance only reject a knife, but not a pen made out of plastic (even though it can cause great damage to a batch of laundry!).

The ultimate goal is to provide the neural network with as many samples it needs, to have seen almost all items that might possibly end up in a laundry. As neural networks shine at identifying items they have encountered before, the goal is to provide them with as many examples of accepted (e.g., buttons or zippers) and not-accepted and possibly dangerous (e.g., knives or needles) items as possible.


The better the AI is trained, the higher the sorting accuracy ultimately becomes. The secret ingredient here is machine learning, which enables our systems to not just understand, but also become experts when it comes to sorting laundry items – all without requiring additional external assistance. What sets us apart is our commitment to customization – we meticulously optimize the parameters of each system to align them seamlessly with the unique needs of the individual laundries. This ensures a tailored and highly efficient sorting process, making our technology adaptable and effective across diverse settings. A customized sorting system that comes with impeccable service is the best way to get any laundry ready for the future. Learn more about our automated solutions here.