A brand new autonomous store in Seoul, South Korea combines load cells with 2D cameras and AI to offer a 24/7 shopping experience with no staff or human-in-the-loop interaction. The autonomous shopping experience was developed in response to staff shortages and the rising minimum wage in South Korea (sound familiar?).
Based on a convenience store in terms of stock and layout, the new GS25 autonomous store addresses some of the limitations of previous autonomous store models.
“Existing autonomous stores can only stock as much as 50% of traditional shelves. This leads to sales losses for offline retail. In contrast, our weight sensor-mounted shelves boast the shelf capacity improved by 50% compared to other autonomous stores.”
As the creators of the FAIM system used in the shop explain:
“Retailers’ profit margins are very low and maximizing item density is of utter importance. We therefore adopted a flexible hardware design that can be easily mass- produced to bring costs down. Our single-size design consists of narrow weight plates laid contiguous to each other… Such small weight plates won’t have as much weight on top, allowing for lower maximum capacity load cells thus higher weight resolution (sub- gram) without requiring expensive ADCs (Analog-to-Digital Converters).”
In addition, deep learning algorithms developed by Fainders.AI have also reduced the computing resources required and associated hardware costs.
Convenience shopping v. supermarket shopping
The new GS25 store also draws on specific behaviours that differ from how people shop in supermarkets. These behavioural assumptions are:
- No more than one person at a shelf at a time
- No outside objects placed on shelves (so the FAIM system “can safely assume any put back event corresponds to items from the inventory”.
- One action at a time (i.e. picking up one item with one hand)
- Customers don’t alter the weight of items before replacing (such as eating half the packet!)
So, unless someone deliberately drops an item of very similar weight just as they pick up an item off the shelf (a la Indiana Jones with his gold idol and bag of sand), “the weight difference on the load sensors is generally enough to detect an event, and trigger the rest of the system.
Who wants autonomous shops?
Quite a lot of people, apparently!
“According to Business Insider Intelligence, the number of global stores with autonomous capabilities rose from only 350 in 2018 to a forecasted 10,000 stores in 2024, with a sales transaction volume of under $70 million to over $20 billion.”
The question is, will it catch on here? According to a survey conducted in 2022,
“70% of UK shoppers want an autonomous, checkout-free shopping experience … and a further 10% would favour a fully autonomous shopping experience.”
However, the reality seems a little different. There were at one stage 20 Amazon Go and Amazon Fresh locations in the UK, 19 of them in London. However, it appears that these “Just Walk Out” stores haven’t caught on as much as the survey might suggest. In August 2023, Amazon closed three stores in London, including their first flagship store. However, Amazon remained buoyant, saying:
“While we decided to close three Amazon Fresh stores, it doesn’t mean we won’t grow – this year, we will open new Amazon Fresh stores to better serve customers in the greater London area.”
A new hybrid alternative has been trialled in Salford Manchester. The In:Five store:
“Features a bank of locker doors from which customers collect their groceries ordered on an app minutes before. Behind the bank of lockers, robotic arms pick items from shelves and place them in totes. The store can stock around 1,500 SKUs covering ambient, chilled and fresh produce.”
So, it doesn’t look like the corner shop has had it’s day quite yet….
Load cells for all your disruptive technology projects
If you’re planning an innovative new project that’s going to the HUGH, and need load cells to make it happen, contact us first. We have extensive experience in the design of load cell systems for a wide range of markets and sectors, from health to automotive, leisure to food.