Kiosk Brain: How Our Key Copying Machines Learn
You may have used one of our key copying machines before, but did you know that these machines operate in a very similar fashion to the human brain?
When scanning and digitally decoding your house keys, our kiosks use two primary processes that mirror how the human brain functions – computer vision and neural networks.
Our computer vision technology allows the kiosk to scan and recognize your key using multiple cameras in a process very similar to how facial recognition technology identifies a person based on a digital image.
Based on preset algorithms, our key duplication kiosk then generates a 3D image of the key’s teeth. When this image is analyzed, the kiosk’s brain comes to life. Using a collection of simple, trainable mathematical units that collectively learn complex functions, the kiosk imitates the neural networks that connect the various sectors of the human brain. The key scan is then matched to existing information on various key types that the kiosk has collected. With this knowledge, our machine not only reads more keys more accurately, it can even reset your worn out key to what it should look like coming out of the factory!
However, just like the human brain, our kiosks need to be “taught” about different key types. Our engineers are kept busy constantly educating our kiosk about thousands of new key types by inputting that data into learning algorithms – think of flashcards for children, just slightly more advanced.
The most fascinating part about this process is the so-called “network effect” – the impact that one user of a service has on the value of the service to other people. In our case, the more kiosks we have and the more keys we copy, the smarter our neural networks become.
As our key copying machine learns from each scan, it can support more key types more accurately!
Although all of this can sound futuristic and “space age”, the next evolution of this technology is even more exciting.
Some of the most innovative technology companies – Google, Facebook, Amazon – are already using artificial intelligence on a whole new level, employing a concept called deep neural networks learning.
While quite complicated, deep neural networks learning is worth the effort . With this cutting edge technological concept, image features can be learned in an optimal state for detection, adding natural variations to the data, making decisions more quickly and intuitively than ever before.
Our engineers are dedicated to getting it just right — testing theories, recalibrating machine learning — all in the name of giving you a more accurate key in the most convenient way possible.