Go to sporting goods brand Decathlon’s swish new store in San Francisco, just down from the Museum of Modern Art, and you may come across something unexpected. A slim, white robot — sort of a cross between Pixar’s Wall-E and Star Wars’ C3PO — may be ambling between the aisles. It takes pictures of the shelves, pausing to let workers stroll past, and recovering after hugs from enthusiastic kids.
This would be Tally, the newest autonomous retail robot from San Francisco-based robotics startup Simbe.
The ‘zero maintenance robot’ — smart enough to charge itself when needed — drives around the store and uses its 12 high resolution cameras to take frequent images of stock on the shelves.
It then uses computer vision and machine learning to identify whether products are in the right place and correctly priced. A single image can capture everything from the floor to the top shelf and can tell between reduced-fat and ordinary products.
The problem
Simbe writes that although stores know when products arrive, and what they sell, there are too many things they don’t know. They don’t know whether a product is still in the store. They don’t know if it’s on the right shelf, if it’s priced correctly, and if the end caps are properly set up.
It says that this means one in ten products are not on shelves when they should be and that this problem can cost stores billions of dollars per year. Simbe quotes an IHL Group report that finds that almost a quarter of Amazon’s current retail revenue comes from customers who first tried to buy a product in-store and were met with obstacles.
The Tally solution
Tally, Simbe says, can take the boring and manual stock checking work away from staff in a way that is more accurate and more successful.
It says that while a manual stock check is usually done once a week, with an estimated 65% accuracy, Tally can audit 10,000 SKUs — Stock Keeping Units, the unique number assigned to a product by a store — 3 times a day with more than 97%+ accuracy.
Overall, the manufacturers say the robot can check 15,000 to 30,000 products an hour. No flash photography is needed, meaning customers are not startled and the machine is not put off by glass doors.
“People are curious, and they’ll ask the associate, ‘what is that’,” Ashley Benson, Senior Product Manager for Decathlon USA, said in a video. “And it makes for a nice rapport between the associate and the customer to have a conversation piece and talk about what Tally is doing, and from there guide them towards the products they’re looking for.”
Benson is able to check a dashboard where it can see the photos and videos that Tally takes. She can look through Tally’s eyes, and see how sales change over time.
“If we didn’t have Tally we’d be promising products to customers that maybe we didn’t have in stock,” she says.
More growth
Techcrunch reports that the San Francisco-based startup has raised $26m in Series A funding in a round led by Venrock, Future Shape, Valo Ventures and Activant Capital.
The money will reportedly go towards funding for headcount, exploring new markets and accelerating the deployment of existing robots.
“Our investors, both previous and new, provide much more than financial support. They are advocates and trusted advisors who bring invaluable institutional knowledge to all facets of our business,” co-founder and CEO Brad Bogolea said in a release, quoted in TechCrunch.
Tally is now used in Schnuck Markets and Groupe Casino, with US supermarket chain Giant Eagle saying that it will use Tally in selected stores.