In a nondescript building in Seattle lies a secret Amazon lab that could “reimagine” the future of retail with checkout-free shopping.
Amazon and Just Walk Out created a mock store setup where next-gen technology meets everyday shopping. It’s a test lab for how consumers navigate and transact in retail environments.
Amazon says its engineers and researchers are “hard at work solving one of retail’s most complex challenges.”
The issue? How to create a seamless and highly accurate checkout-free shopping experience.
The solution? Giving customers a store where they can grab what they need, tap their card or mobile wallet, and leave without worrying about queues or checkout lines.
In the future, we shop on the go
Imagine just entering a store, taking what you need from the shelf, and paying automatically on your way out.
The magic happens in the system that makes all of this possible – charging the payment instantly and sparing shoppers the hassle of queuing during peak times.
This is where Just Walk Out comes in, with its cameras, weight sensors, and a combination of AI tech to correctly identify what was taken, and the quantity of the items as it leaves the store.
How Amazon uses the secret retail lab
Amazon’s team is testing every possible scenario, like multiple people shopping simultaneously, or even shoppers returning items.
The retail giant explains: “To figure out the ‘who’ part of the ‘who took what’ equation, Just Walk Out associates a shopper with their payment method at the entry gate.”
But here’s the kicker: They don’t use biometrics-based tech (like facial recognition sofware) to identify shoppers. Instead, the system “only tracks how their hand interacts with the products and fixtures (such as shelves or fridges).”
The system then “correctly identifies the products and quantities they leave the store with. In fact, if a shopper comes to the store again with another credit card, the system would have no idea it was the same person.”
Amazon’s checkout-free shopping: How it works
Chris Broaddus, senior manager of applied science at Amazon Web Services (AWS) says the system needs to identify what the shopper is doing with their hands.
Are they:
- Picking up a product?
- Returning a product?
- Just rummaging through the shelf?
He says the system also uses “continuous self-learning and transformer technology.” This is a special kind of neural network architecture that translates inputs, tokens and outputs.
The inputs are what the sensors pick up in the store: For example, if a customers picks up an item, or returns it to the shelf.
Then, using the sensor data, the networks converts it into a simplified form known as tokens. These are small, digestible pieces of information that the neural network can easily process.
Finally, after processing the tokens, the neural network will produce an output. Think of this as the shopping receipt.
This output should accurately reflect what items the shopper has with them as they walk out the store, without the need to check manually through a paypoint or cash register.
A better experience for shoppers and retailers
Amazon and Just Walk Out Labs take this one step further by analyzing shopper behaviour based on locations and store types.
For example, at an airport the system would be able to identify people carrying big bags or luggage items that may obstruct cameras and sensors. If a store is located near a sports stadium, the network could identify shoppers who are wearing the same team colors.
To create an error-free shopping experience devoid of any hassles, Just Walk Out researchers test these scenarios first by mapping a store’s layout in the lab.
They do this by using Light Detection and Ranging (LiDAR) technology to create detailed 3D maps of space, using laser light.
Based on this data, the system can be adjust with extra cameras, if needed. Think of the airport setting where customers carrying luggage could block certain cameras.
For shoppers, this means a hassle-free shopping experience, where they literally just walk into a store, pick their items, and walk out.
Retailers, meanwhile, analyze client data and behavior, which will help them prepare for complex shopping scenarios or simply to adjust their system accordingly (the network, cameras, sensors, lights, etc.)
Images: Amazon and Just Walk Out.
About the author
Cheryl has contributed to various international publications, with a fervor for data and technology. She explores the intersection of emerging tech trends with logistics, focusing on how digital innovations are reshaping industries on a global scale. When she's not dissecting the latest developments in AI-driven innovation and digital solutions, Cheryl can be found gaming, kickboxing, or navigating the novel niches of consumer gadgetry.
