Georges Mirza advises companies on developing and prioritizing strategies in computer vision, AI and robotics. The following submitted piece reflects his points of view.
“Timing is everything” is an ironic truth. In my early days, developing the robot solution for retail and defining its road map, I would say it was the beginning of embracing the technology and its presence to pave the way for adoption. There was reluctance and worry of the image a robot solution would project in stores: Would it take away jobs? Interfere with shoppers? and so on. Concerns faded away as developments evolved and time passed. Some even humanized the robot and gave it a name, such as Marty from Badger. Others used a more inconspicuous design like Tally from Simbe to mold with surroundings and provide less interference to shoppers and employees. There has been progress in the area of hesitation as solutions evolve and competition from the likes of Amazon motivate retailers to integrate new technologies faster than usual. I call that advancement and realization maturity.
Highly complicated solution
The road to automating retail stores with robots, as with any changing technology, will be prolonged and challenging. Any paradigm shift for retailers and everyone in their ecosystem requires management and leadership. Understandably, typical startup challenges can lead to costly delays as well. Add the multiple, significant components needed to build a robotic solution for each retailer and you compound these challenges and can create a recipe for disaster. The components are:
1) Robot hardware and autonomous navigation in a congested environment.
2) Collecting and processing imagery, building and growing the AI and learning capabilities.
3) All the infrastructure needed to oversee, maintain and support to solve real retail business problems.
Who is on the right track?
All the current players have achieved various levels of success in accurately, repeatedly and reliably scanning retail shelves at scale. Overall, I believe they have gained reasonable reliability in autonomous navigation. Image processing is the area I consider the most complex, it is lengthy in effort and is the bulk of the complete solution along with scale challenges. We are still a distance attaining 80% from a usable shelf compliance solution, but it is within reach, more so for some. My perspectives on where we are to date follow herein.
Bossa Nova Robotics
Bossa Nova is the longest-running company to bring realization to a robotic solution for retail and the apparent leader, completing major milestones in installation of autonomous data collection robots in 350 Walmart stores and just receiving a commitment for an additional 650 stores to be complete by summer of 2020. This is great news and a sizable milestone that Bossa Nova can use in its next funding round. As Sarjoun Skaff, chief technology officer of Bossa Nova, put it, “This is a very complex solution and, as we found out, there are no shortcuts. Everyone has to go through the same challenges to get here. We are focused on solving problems to deliver a scalable solution and providing true value to our customers.”
Building a solution with cost to scale has been an issue for all players. With the release of the new 2020 hardware, Bossa Nova claims that it comes at a lower cost. I believe this is an area of opportunity for Bossa Nova to break away and add additional retailer banners.
There is big news in Simbe raising $26 million in funding, and a partnership with Softbank inventory financing to expedite deployment of robots globally. Simbe believes it has the largest geographical deployment of robots. “We have been focused on operationalizing the data as a priority in getting ready to scale. We feel we have done a good job diversifying our customer base and making the solution work for all departments and partners, this puts us in a good position to grow to chain wide rollouts this year, announcements forthcoming,” said Brad Bogolea, CEO of Simbe Robotics.
Simbe has a presence in the U.S. at Schnucks and Giant Eagle using the vision solution, and now with Decathlon using recently added RFID capabilities. Simbe also has coverage in Europe, the UAE and Asia with notable large global retailers. Diversifying and capturing a wide group of global retailers might prove to be a winning strategy, as Simbe gets ready to scale. There is no real clarity on U.S. activity beyond the 15 Schnucks and Giant Eagle locations. I still look for larger store count rollout beyond the 50 stores publicly announced and expansion in the current installs at Schnucks and Giant Eagle.
The recent money raised and Softbank partnership to increase production efficiencies could give Simbe the boost needed to mature its capabilities and scale. Simbe did not participate with a presence at the national Retail Federation show this year; with current global retailer coverage, this could be a strategy to go dark with heads down using the latest infusion to build up their abilities and get ready for the next phase to scale. We will wait and see.
I recognized this company early on for its accomplishment of building a spill-detection solution for retail stores, in a short couple of years. Although less complex, Badger has been able to scale it and deploy it to more than 500 Giant Food Stores/Martin’s and Stop & Shop stores.
Tim Rowland, CEO of Badger Technologies, says, “We see a need for a multifunction robot. We were surprised by the demand for a combined spill detection and inventory data collection robot as we expand our discussion with retailers. We also notice retailers looking beyond the hype of the robot idea and to focus more on how to operationalize the data collected to improve efficiencies and shopping experiences.”
I think Badger starting out with a spill-detection machine is an advantage, getting experience testing in live environments, and monetizing early on to help go further without having to rely on constant money-raising efforts. It helps as well to be part of Jabil, a $26 billion manufacturing solutions provider. If the Badger/Jabil combo is not able to bring manufacturing costs down, no one can. Badger has an impressive local seasoned team that works well together and has the feel-good sense of a small town in Kentucky coming together and getting it done.
They are now testing Badger Retail insight, a robot that addresses out-of-stocks, plan-o-gram compliance and price integrity issues; it competes with Bossa Nova and others. The processing capability is the biggest challenge and is a one- to three-year effort to complete and achieve an acceptable compliance reporting functionality. This will be key to how fast the solution penetrates and wins market share. It is testing at Walmart and a half-dozen other retailers globally; we will continue to observe how Badger matures its solution.
Zippedi has been stealthily building solution capabilities leveraging local university talent and getting reasonable store coverage in South America.
“We have tried several approaches to bring a real solution to our customers, from fixed cameras to robots. I think with the current robotic solution capability and tools we provide the expanded retailer partners, we are minimizing the out-of-stock problem for them by allowing them to respond quicker to shelf conditions,” said Luis Vera, CEO of Zippedi.
Zippedi has built a solution for the retail ecosystem to scale from the ground up. It capitalized on retail expertise from the leadership’s previous venture called Scopix that provided video analytics for retailers to improve business operations, sales and profits. Zippedi has validated the robotic solution in several retailers in grocery and home improvement across five countries. It is now testing in several major U.S. retailers, and we can look forward to seeing if the model will gain traction.
The best-kept secret that everyone knew is out. Zebra rolled out its version of the shelf-scanning solution at the NRF Big Show. Known as SmartSight, Zebra boasts its ability to integrate with its store solutions portfolio such as mobile computers and the EMA50 to integrate to other store systems.
“With the SmartSight solution integrated with the rest of our portfolio, we can prioritize the tasks as they are pushed to the store associates’ mobile computers, increasing their availability to interact directly with shoppers. We feel confident our existing capabilities are strong enough to optimize replenishment, reduce out-of-stocks and provide value around compliance while reassigning labor to higher value assignments that enhance the shopper experience,” said Rob Armstrong, VP of portfolio marketing for Zebra Technologies.
Normally I would say it is a bit late to the party, but Zebra leveraging its heritage, infrastructure, portfolio and reach might be the game changer. What Zebra lacks in agility because of its size, it makes it up with maturity, certainly a rare ingredient in the startup world of robots. Image processing and computer vision proficiencies could have gotten a boost as well from a recent acquisition of Cortexica and an investment in Focal Systems. Although shelf insights are a more complex problem to solve, it’s possible the new talent will help mature the solution. These are very interesting moves. We will watch closely how it all comes together over the next year.
Brain Corp. and Savioke
Brain Corp specializes in robot operating systems, with its early success of implanting its technology into floor scrubbing machines and successfully rolling out with Walmart. It is now collaborating with Savioke, a maker of robots for hospitality. Together they are working on a shelf-scanning solution. As with Zebra, these are early days, and they will soon learn the complications of collecting and processing retail data at scale.
The field is getting crowded, and prospects of collaborating and simplifying efforts to accomplish fast results are growing.
What to Look Out For Next
Bossa Nova’s rollout to 1,000 Walmart stores by end of summer, and indications that its model is viable for other retailers to consider. Simbe convergence to expand chain wide in its current retailer presence. Zippedi solidifying its presence in Latin America and penetrating the North American market. The progresses to be made by Badger building the processing capabilities of its insight solution and gaining wider retailer store presence of its new robot.
All are progressing at different speeds maturing their capabilities with differing models. No question about it, robots are here to stay and will be a permanent moving fixture in every retail aisle. They will have various robot duties, providing real data to feed many solutions and finally making them usable. The future that we dreamed about is here.
As we get to scale, the impact of robotics in retail will start to touch the whole ecosystem beyond the obvious data collection efficiencies and labor savings. Retail will not be the same. Current solutions in retail will either be enhanced and made more effective or rendered obsolete. Regardless, this opens the door to countless possibilities that many in the industry await. This realization occurs when robots reach the threshold of truly scaling and collecting data from “the whole store in all stores,” as Danny Sacco, former retail leader at Nielsen, keeps reminding me. I agree and say: “This is when the fun begins.”