Amazon Enhances Just Walk Out Technology with a New AI System
Amazon is elevating its Just Walk Out technology, the system that enables checkout-free shopping experiences, with a significant upgrade. The company unveiled a new multi-modal foundation model, powered by the same transformer-based machine learning models used in generative AI applications. This innovation promises to enhance the accuracy and efficiency of Just Walk Out systems for retailers and shoppers alike.
Just Walk Out technology is expanding into hospitals, offering a ‘badge pay’ option for healthcare staff.
The new AI system analyzes data from cameras and sensors throughout the store simultaneously, rather than sequentially, which was the approach of the previous version. This shift promises better performance and reduced need for manual adjustment.
For retailers, the new AI system means faster deployment, increased efficiency, and a more streamlined experience. For shoppers, it translates to worry-free shopping at an expanding number of third-party checkout-free stores worldwide.
The multi-modal AI system uses a sophisticated architecture to analyze data streams from various sources.
How the New System Works
Previously, Just Walk Out’s AI analyzed shopper behavior step by step, processing movement, item pickup, and quantities in a linear sequence. However, in complex or unusual situations—like obstructed camera views or crowding—this sequential approach could require time to accurately determine purchases. The new system addresses these issues by examining all sensor data at the same time.
The system’s architecture takes inputs from multiple sources, including cameras and weight sensors, and prioritizes the most crucial information to identify the correct items and quantities selected by each shopper. It uses continuous self-learning and transformer technology—a type of neural network architecture that transforms inputs into outputs, such as receipts for checkout-free shopping.
For instance, a shopper might consider several varieties of yogurt, and as they do so, another customer might reach for the same item, or the freezer door could fog up, obscuring the cameras’ view. The new AI model is designed to determine accurately the items taken by each shopper, minimizing receipt delays and accelerating technology deployment for retailers.
The new AI can accurately determine what items each shopper selected.
The new AI model also reduces a need for frequent model retraining, as the system continues to learn independently. The system is trained using a 3D map of the store, which understands the arrangement of fixtures and merchandise, as well as an image catalog of the store’s products.
Seamless Checkout-Free Shopping
The enhancements to the AI system are designed to be seamless. Customers should continue to enjoy the same contactless shopping experience. Upon entry to a Just Walk Out store, customers check in via a payment method at the entry gate, which the system uses to track them as they browse.
Customers can pay by various methods upon entering a Just Walk Out store.
Just Walk Out technology is currently available in over 170 third-party locations. Amazon plans to launch more Just Walk Out stores in 2024 than in any previous year, more than doubling the number of third-party stores with the technology this year. The company says that the system will continue to learn from everyday shopping scenarios and raise the bar for accuracy and convenience, benefiting both retailers and customers worldwide.
Just Walk Out is expanding to provide checkout-free shopping for a wider range of merchandise.
