- Getty/Rick T. Wilking
- Most retail stores have no way to account for what merchandise is actually in a store and where it is in an efficient way.
- But, Bossa Nova Robotics developed a solution. Its current signature device is an autonomous robot that roams up and down store aisles, checking for pricing issues, out-of-stocks, and shelf irregularities.
- It was no easy task, and the company’s cofounder and chief technology officer, Sarjoun Skaff, admits he was “naive” when he started the project.
- Walmart recently expanded a pilot program involving Bossa Nova’s shelf-scanning robot as part of a larger push into robotics.
- Business Insider named Skaff to its list of the 100 people transforming business.
- See the full list of the 100 people transforming business here.
Customers move stuff around, people steal things, and sometimes product falls through the cracks.
These are just a few of the perils goods face in a big-box store. They help explain why the number of items being shipped to a store does not always equal the number of things that are on store shelves and being purchased by customers.
These stores are large, and it’s hard to keep track of individual items, even when using supply-chain logistics.
“There is a fundamental challenge in knowing what’s in a store and where,” Sarjoun Skaff, Bossa Nova Robotics’ cofounder and chief technology officer, told Business Insider in a recent interview.
Bossa Nova is a robotics startup founded in 2005. Its current signature device is an autonomous robot that roams up and down store aisles, checking for pricing issues, out-of-stocks, and shelf irregularities. It can do this far more efficiently and regularly than a human taking stock by hand can, providing retailers with valuable real-time data.
“I admit to being naive when we first started this,” Skaff said. “As we started to build them, we started to realize the scope, the magnitude of the challenge is enormous.”
- Hollis Johnson/Business Insider
Essentially, Bossa Nova has built a mini self-driving car with an enormous camera on top that uses computer vision to see like a human does.
“From a technical perspective, it turns out that the hardware was difficult because we needed the right imaging and optics in order to simplify the work of the artificial intelligence on the back end,” Skaff said.
He estimated that it took the company one-and-a-half to two years to find a solution that could scale.
Bossa Nova works closely with Carnegie Mellon University, and it is even funding a biometrics lab affiliated with the school.
“That may sound a bit odd,” Skaff said. “But it turns out that the technology is the same at its core. The way you recognize faces with deep learning algorithms is fundamentally the same way you recognize product.”
The next job was making sure the robots were accurate enough for practical use. Working with the biometrics lab, the robots are now able to decipher what is on a shelf with “exquisite” accuracy, Skaff said.
“We are actually working at scale in stores today, which means that we have hit the minimum percentage – which is very high – that our customers imposed on us,” he said.
One of those customers – Walmart – has had Bossa Nova robots roaming its stores since 2017. The retailer announced earlier this month that it is expanding the number of the robots it will have in its stores by 300, indicating at least some success with it. Walmart was previously testing the robots in 50 stores.
“We have worked hand in hand with Walmart to build the product since 2014. They’ve been really incredibly patient,” Skaff said. “This is now bearing fruit. We built something that they want.”
Though Skaff says “customers,” plural, to date the company has only revealed the identity of one of them: Walmart. Skaff says others are currently under a non-disclosure agreement.
These customers aren’t buying Bossa Nova’s robots, however. Instead, Bossa Nova itself installs, maintains, and updates the robots in stores, selling the retailers the data they generate as a service.
“Today, we joke that we are, on the surface, one startup, but in reality, there are probably like five startups inside,” Skaff said.
Not everyone is so on board with Bossa Nova’s approach.
“I think the shelf bots [like Bossa Nova’s] are good, but they’re expensive, and there’s much more effective ways to do it,” Steven Keith Platt, the research director of the Retail Analytics Council and a professor at Northwestern University, said. “They have video cameras, they have to carry a large CPU to process all that video. And so it makes it extremely expensive.”
“I think, two years from now, the same function will be accomplished, but with a much different robot,” Platt added, noting that robots that are modular and can perform many different functions at once have a better chance of being a net positive for both retailers and robotics companies.
Skaff is undaunted.
“I think as an industry, [retailers] are hungry for data about their store operations,” he said. “They keep asking for more and more and more. So I think we have easily 10 years of innovation [ahead of us] … mapping the retail world.”
“[Every retailer] has the same problem, and they want the same solution.”