Nowadays, almost every major supermarket chain offers the opportunity to pay at self-service checkouts. Self-service checkouts use less space than registers, have reduced wait times compared to cashier lanes, and need fewer staff members to be controlled. A downside is an increase in shoplifting (Schmidt-Jacobsen, n.d.). Unfortunately, there is no alternative that offers at least the same convenience with almost zero shoplifting.. or is there?
You may have heard of it already, but Amazon has already opened 28 stores with so-called ‘Just walk out’ technology (Statista, 2021). They are called Amazon Go or Amazon Go Grocery stores. By scanning a QR code in the Amazon app on your phone, you get access to the store. Whenever you take an item from the shelf, the app automatically adds the item to your virtual shopping cart. When the customer changes their mind and puts the item back, the app will update the virtual shopping cart automatically. After you have gathered all the desired products, you can just leave the store and you will be billed for the products that you have taken via the app (Cheng, 2019).
So, how does this ‘Just walk out’ technology work? Amazon is using a combination of deep learning, computer vision, and sensor fusion. To get a better understanding of these technologies, I will explain each technology briefly. Deep learning is the creation of machines that use techniques inspired by the human brain’s ability to learn (Mandal, Mehta & Mishra, 2017). It is used to recognize objects, in Amazon’s occasion customers and products. However, just deep learning alone will not recognize anything that happens in the store. That is when computer vision comes into play. It mimics the human vision to look, discover, and method images (Mandal et al., 2017). This is done with the help of hundreds of sensors in the store that register every movement. Sensor fusion is the technology that tracks the movement of customers when cameras cannot register the movement, for example, because the visual input is blocked by another customer. This is done by the combination of accelerometers and gyroscopes in the customer’s mobile phone that together registers the location (Mandal et al., 2017). This technology is so accurate that shoplifting is almost impossible. Apart from that, even when a customer puts an item back on the wrong shelf, the technology will register that action and the customer will not be charged for that product (Cheng, 2019).
However, as disruptive as this concept might look, Amazon has not yet reached the 3000 stores they intended to have by 2021 (Statt, 2018). One possible explanation is the costs of the technology. According to Focal Systems (2019), the system will not prove a break-even in a large-format grocery store compared to operating the front-end with cashiers until after 2040. For now, Amazon should therefore stick with smaller stores.
What do you think about the technology and do you think we will see this much more often in the near future? Let me know in the comments.
References: https://www.researchgate.net/publication/315380585_JUST_WALK_OUT_TECHNOLOGY
https://www.theverge.com/2018/9/19/17880530/amazon-go-cashier-less-stores-new-locations-2021
https://sprintingretail.com/blog/retail-self-checkout-systems/
https://www.statista.com/statistics/1155873/amazon-store-openings-number/