Skip to main content

COMPUTER VISION

AI Smart Stores: The Evolution of Retail

smart store

Labor shortages and ecommerce competition have created a real crisis for retailers today. Customers now expect quick and frictionless interactions. And they no longer have the patience for understaffed stores, empty shelves, or long checkout lines. Retailers need to start rethinking their operations and experiences, otherwise they risk being put out of business.

The Achilles heel of brick-and-mortar stores is the complete dependence on staffing—a weakness that has become painfully clear to so many retail businesses over the past two years.

“Staffing in retail is now a worldwide problem,” says Mark Perry, Head of Global Business Development for Cloudpick, a provider of autonomous store solutions. “In Japan and South Korea, where you have aging populations, businesses have historically struggled with it. But now in places like Hungary, companies have to offer thousands a month just to hire a cashier. Even in the United States, some physical stores are no longer able to serve their customers due to the lack of workers.”

To address this growing problem, many retailers are adopting AI capabilities with an eye toward a fully automated retail future.

“The future of retail is AI-driven,” says Ria Cheruvu, AI Software Architect at Intel. “Today, we are starting to see transformations in customer experiences via automated self-checkout and intelligent queue management, automation of inventory monitoring procedures, and more, with the long-term future of sustainable restocking procedures.”

AI Smart Store Takes Over French Campus

When the French supermarket chain Auchan became overwhelmed by the influx of foot traffic at its EDHEC Business School campus location, they knew they needed to step up their game. They transformed their pick-and-go Auchan Go convenience store into a fully automated smart store.

To do so, the supermarket chain partnered with Cloudpick. The goal: to offer busy students a faster and more convenient shopping experience, without lines or long waits.

Many retailers are adopting #AI capabilities with an eye toward a fully #automated #retail future. @CloudpickTech via @insightdottech

One of the issues Auchan wanted to solve was its inability to rapidly serve students around the clock. For instance, students would often get up early in the morning and rush to classes, which meant they might not have enough time to eat. Or they forgot something and needed to grab supplies quickly before class. But with a lack of staff, the store wasn’t always able to fulfill these needs at certain hours.

By partnering with Cloudpick, it was able to design a completely autonomous AI retail solution capable of running 24 hours a day without cashiers or customer service staff. Because of this effort, Auchan was recognized as a 2021 retail innovator by LSA.

How an AI Smart Store Operates

To access the store, customers must first download a mobile application and provide their payment details. When they arrive at Auchan Go, they’re provided with a QR code that assigns them to a virtual shopping cart. With this in place, customers can simply pick out the items they want, and walk out of the store.

Payment is automatically processed and handled in the cloud. The store also uses edge and computer vision technology to ensure customers are charged only for the items they left the store with (Video 1).

Video 1. Cloudpick smart stores automatically detect and charge for items customers leave the store with. (Source: Cloudpick)

Cloudpick’s solution can process up to more than 800 shoppers per hour, so there’s little chance customers will ever have to wait in lines. And Auchan store managers are relieved that they no longer have to be on-site all the time. If they’re needed to handle a problem or help restock shelves, they’re notified via the smartphone app.

The overall results included improved operational efficiency, reduced labor costs, and faster automated checkout.

“Cloudpick’s integration of AI demonstrates the exciting combination of the benefits of AI in this space,” says Cheruvu.

Powering an AI Smart Store

To achieve this level of autonomy and accuracy, Cloudpick combines several different types of AI technology.

The solution uses a computer vision and deep learning system built on the OpenVINO AI toolkit to create a virtual inventory of all the products. This is how the smart store differentiates one item from another.

With OpenVINO’s CNN-based deep learning inference, the company can optimize AI model performance and compress model sizes to run adequately on low-power devices, according to Perry. Additionally, Cloudpick can minimize its time to market with OpenVINO’s library of computer vision functions and pre-optimized kernels.

Inside the store, Cloudpick uses AIoT motion and weight sensors to keep track of what’s going on with products, and Intel® RealSense 3D cameras track shopper behavior in real time by performing gesture recognition at the edge. The computational heavy lifting is done with on-premises Intel® edge servers.

“It is incredibly exciting to me to see the breadth of AI capabilities that Cloudpick demonstrates leveraging Intel and OpenVINO,” says Cheruvu. “The combination of AI pose estimation, object detection, personalized ads, and more. Working with computer vision and sensor fusion algorithms to create a comprehensive shopping experience for consumers that also enables convenient inventory management is to me, the future of retail.”

Retail’s Missing Link

While the retail space is no stranger to digital transformation, Perry believes AI has been the missing link in the evolution of retail and gathering better retail data.

“When you add AI to the equation, suddenly everything changes, because every process can be automated: from checkout and payment to inventory and restocking,” he explains.

AI not only enables retailers to collect data points they could never access before, it helps stakeholders understand exactly what the data means to make more informed decisions. From understanding customer behaviors better to providing relevant personalized offers in real time, the opportunities are endless.

“Retailers may consider experimenting with different types of models and capabilities to find the optimal scheme that fits their use case,” Cheruvu explains. “To scale AI integration, retailers may also consider upgrading the quality, and types of sensors and capture devices placed throughout the store to capture additional data.”

A Smarter, More Sustainable Future

In the years to come, Perry expects to see in-store inventory systems capable of integrating directly with manufacturers and distributors upstream. This will help goods flow through the supply chain more smoothly, eliminating delays and stock shortages.

Longer term, he sees AI completely automating retail supply chain with self-driving vehicles and stocking robots to be commonplace. The smart stores of the future will know when they need more product, order it automatically from the supplier, and have it delivered and stocked by machines.

“If you automate restocking,” says Perry, “you can optimize the loading of the delivery vehicles, and the routes that they drive. It’s a very effective way to reduce carbon emissions, because you’re not wasting fuel by making excessive deliveries or sending stock where it isn’t needed.”
 

This article was originally published on June 8, 2022.

This article was edited by Christina Cardoza, Editorial Director for insight.tech.