On January 22, 2018, Amazon opened its first Amazon Go store to the public, in Seattle. The convenience store had no checkout lines and no registers. Shoppers scanned an app on the way in, took items off the shelves, and walked out, with the items charged to their Amazon account automatically. Amazon called the underlying system “Just Walk Out” technology.
Amazon describes the system as a blend of computer vision and machine learning that determines, in the company’s words, “who takes what and charge them correctly when they walk out.” Ceiling-mounted cameras give a store-wide view, while load-cell weight sensors on the shelves help detect small items like gum or lipstick. Amazon calls the combination “sensor fusion.” To train the system to recognize shopping actions across different store layouts, lighting and crowds, the company generated synthetic training data, including the use of generative adversarial networks.
The launch was widely seen as a “quake moment” for retail, prompting Walmart and others to pursue their own vision-based checkout systems. From 2020 Amazon began licensing Just Walk Out to outside retailers, putting it in stadiums, airports, hospitals and college campuses.
Amazon later concluded that different store formats wanted different technology. In larger grocery stores it leaned on the Amazon Dash Cart, a smart cart that lets shoppers skip the checkout line, while keeping Just Walk Out for smaller-format stores and third-party venues. By the mid-2020s Amazon said Just Walk Out was running in well over 100 locations across multiple countries.