Online shopping and tap-to-pay methods have slowly introduced shoppers to the future of technology-driven retail, where convenience and contactless efficiency are paramount.
Studies show that global retailer spending on AI solutions will reach $12 billion by 2023, due to the increased use of AI to provide personalized services and predict consumer behavior.
The pandemic has only accelerated the deployment of AI in retail with contactless shopping, such as curbside pickup and delivery, due to safety concerns, and these new methods have remained ever since.
Now standalone retail is becoming mainstream, from takeout style grocery stores to cashierless kiosks in sports stadiums and frictionless convenience stores on college campuses.
Grocery technology trial and error
AI-powered smart carts have been getting a lot of attention lately, as have contactless self-checkout systems, which typically rely on depth cameras to identify items from any angle and instantly recall them in a single transaction.
As it grows in popularity and produces higher accuracy than smart carts, no no matter how fast the checkout process itself is, customers still don’t want to wait in line. Autonomous stores have predominated so far and contain two core technologies components: computer vision and shelf sensors. Most implementations use both technologies together, but it can be advantageous to ignore shelf sensors.
Weigh the pros and cons
Most autonomous stores have shelf sensors on their shelves that interact with their computer vision systems. Shop shelves are equipped with a weighing system that provides information about removing or replacing the product from a shelf.
However, merging both weight and visual components can overcomplicate a store’s operation, cost the retailer more, and limit the store’s shape, size, and layout. Camera-only solutions can reduce installation costs by up to 60% and thus reduce the time it takes to set up a shop.
Computer vision with standard security cameras can do the job alone and offers flexibility, affordability and scalability for autonomous stores.
Computer vision predominates
Computer vision can be implemented in a wide variety of store types and products, including large supermarket chains, mini stores and even liquor lanes. Instead of undergoing reconstruction, cameras can adapt to a store’s existing format and work with whatever layout the operators want to create.
Cameras can also easily track hard-to-weigh items such as draft beer, hot and ready-to-eat foods, baked goods, fruits and vegetables, and merchandise.
It should come as no great surprise that one technology system is cheaper than two, and when one system can do the work of both, it is also cost effective. The more sensors you have, be it cameras or weight sensors, the more machines are needed for calculations.
When it comes to scale, given that computer vision can work with a store of any size or format, it’s usually the more viable option for a small or unique store, allowing the technology to do that spread far and wide.
The not so distant future
In the coming years, more and more technological innovations will emerge and change the way we shop. One of the innovations so far is that autonomous retail is picking up steam. Retailers and technology both providers are figuring out exactly what kind of equipment, technology and setups are needed successfully automate the shopping experience.
The industry already sees customers excited about the expansion of autonomous retail and expects widespread adoption.
João Diogo Falcao is CTO at AiFi