Bossa Nova: Robotics, AI and Groceries
Bossa Nova is deploying robots and AI technology in grocery superstores to reduce the chronic challenge of inventory management.
What is Bossa Nova?
Bossa Nova is a Silicon Valley Series B startup, valued at $179M, with a mission to deliver advanced robotic technology and artificial intelligence (AI) solutions to retailers [1]. Currently, Bossa Nova is focused on solving the inventory management challenges faced by large brick-and-mortar grocery stores. Its signature product is an autonomous robot that can deftly navigate the aisles of a grocery store (using LIDAR maps) while also collecting 2D and 3D images of all the shelves, freezers and produce displays [2]. The images are sent to the cloud where they are processed rapidly by Bossa Nova’s AI software and translated into actionable data points – including the identification of out-of-stock, misplaced or mispriced products. This data is sent back to store associates, helping them efficiently manage the inventory available and improve the customer experience [3, 4].
While the robotic element of this solution is itself noteworthy, the AI used by Bossa Nova for item identification is considered leading edge, which earned the company a place on Forbes’ 2019 “AI 50: America’s Most Promising Artificial Intelligence Companies” list [1]. It is not a simple task to classify the various SKUs on grocery shelves using digital images, since product packaging changes frequently and items are often rotated or knocked over by customers [5]. Bossa Nova is still iterating on this challenge, but a 2018 acquisition of an AI lab called HawXeye provided a critical boost. The HawXeye team had spun out of Carnegie Mellon University and specialized in deep machine learning, facial and object recognition. According to Bossa Nova’s Chief Technology Officer (CTO) Sarjoun Skaff: “it turns out that… the way you recognize faces is fundamentally the same way you recognize product” [6]. With this new methodology, Bossa Nova’s AI software can now identify its customers’ SKUs with an impressive degree of accuracy.
Bossa Nova reports to have six major grocery customers across the US and UK already signed up, with the two known customers being Walmart and Albertsons. At Walmart, 350 robots are being used across stores so far [3]. Interestingly, from a value capture perspective, Bossa Nova does not charge retailers for the robots or their maintenance. Instead, it collects service fees for the data and analysis that it provides [6].
How Does Bossa Nova Create Value?
Inventory management in brick-and-mortar stores is a “constant problem” for virtually all large retailers [7]. For instance, store personnel must spend the majority of their time stocking and rearranging items on shelves, rather than assisting customers [7]. This is usually not an efficient system – employees can mistakenly overlook problems on crowded shelves, and do not know where to focus their energy for maximum value. Inevitably, as product is displaced throughout the store, retailers also face the issue of “phantom inventory.” This occurs when a product is technically in-stock based on the point-of-sale database, but it is not actually available on the shelf. Phantom inventory is a difficult problem to spot and results in heavy missed sales for grocers.
Compounding these inventory concerns is the pressure on retailers to embrace omni-channel initiatives. Walmart, for example, has introduced the options of store pickup or local delivery for online orders [3]. These experiences hinge on an accurate understanding of the inventory on each store’s shelves.
These are some of the reasons why Bossa Nova’s technology has generated so much excitement – it has the potential to automate a difficult task, improve the accuracy of inventory tracking, and empower store employees to prioritize problem areas.
Opportunities for Bossa Nova
Bossa Nova has a vision to apply their technology in retail contexts outside of grocery supercenters. They anticipate that home improvement stores and office supply stores could become customers in the near future. They have even been working on developing a smaller robot that could navigate tighter spaces to support these plans [3]. There is also speculation that Bossa Nova could abandon its expensive robot fleet entirely, opting to install permanent cameras throughout the stores instead so that data collection can occur in real time [8].
Challenges for Bossa Nova
First, while Bossa Nova’s customers believe that this technology, in its current state, is worth implementing in stores – there is still plenty of space for improvement with this young product. Bossa Nova continues to refine its AI algorithms to handle a host of edge cases that can arise in a retail setting. Second, as with most AI solutions, Bossa Nova has encountered criticism that it will replace valuable human jobs. For this, Bossa Nova points to its pilot of the technology with Walmart, which showcases how the robots operate as a complement to the existing store employees – helping them accomplish more in a given shift. Tom McGowan, a manager of a Walmart store using Bossa Nova, explained: “employees have embraced the robot… because it performs a tedious task that no one likes – cataloging out-of-stock items” [3]. Finally, there have been some reported concerns about privacy. Given that the HawXeye team originally specialized in unconstrained AI face detection and matching [9], analysts wonder if the Bossa Nova robots will eventually be used to monitor customers and their shopping habits. CTO Skaff has not commented on whether this could be a future application, but clarified that today, Bossa Nova is “not a surveillance robot. It can only see people as a sort of digitized outline… the images are stored in its map as an obstacle… our high resolution cameras are pointed only at the items on the shelves” [3].
Work Cited
[2] https://www.wired.com/story/please-do-not-assault-the-towering-robot-that-roams-walmart/
[4] https://www.bossanova.com/customer-experience
[6] https://www.businessinsider.com/bossa-nova-robot-walmart-uses-2019-4
[9] https://www.hawxeye.net/technology-1
Great article! It is exciting to see robots change the way stores and warehouses manage inventory. I am curious why Bossa Nova chose to specialize in retail store inventory management, and not warehouse management, which is likely a bigger market.
Also, how scalable is Bossa Nova’s solution? Since its robots rely on LIDAR maps for navigation and localization, how easy or hard is it for the robots to adapt to a new retail setting and to understand new product SKUs? Are the LIDAR maps also generated and refined by Bossa Nova’s robots, or does it incur additional costs?