The Use of Autonomous Mobile Robots in Retail

robotics in retail
Photo courtesy of Knightscope

You’re visiting the grocery store when the announcement “spill in aisle four” sounds off. Since you’re on aisle three, you take a peek to see if you might help out. As you round the corner, you notice the spill actually means a customer got sick, really sick, and somebody is going to need to deal with a pretty unpleasant and potentially infectious situation. Then you notice an approaching whirring noise and witness a special test robot roll up, stop, and proceed to handle the situation.

For those of a certain age, you might hear somewhere in the back of your mind, “Danger, Will Robinson!” from the old black‑and‑white Lost in Space television series. Today, however, thanks to advances in technology, robotics has come of age, including applications within retail.

Remaking Retail

Retail is incredibly competitive and perhaps at a fever pitch with reduced margins across brick‑and‑mortar, online, and omni‑channel chains. Convenience and safety rule the day, meaning retailers must gin up their innovation and technology efforts to increase their customers’ in‑store experience and even their company’s long‑term existence.

- Sponsors -

E‑commerce isn’t new, but according to digitalcommerceonline360.com, customers spent an incredible $861 billion online in 2020, an unprecedented 44 percent increase over 2019. The pandemic appears to have boosted a shift to digital shopping over physical stores at a pace that would typically take place over five years, requiring massive innovation.

Different forms of artificial intelligence (AI) including data management and computer vision are part of this innovation. Well‑designed AI models more rapidly detect opportunistic and profitable inventory popularity and movement patterns; building heating, ventilation, and air‑conditioning and lighting efficiency areas; and customer purchase and theft patterns and events.

The pandemic and convenience‑led move from stores means retail executives are needing to remake their companies to better showcase key merchandise products, provide a more convenient and interesting physical store‑visit experience, and identify and retain employees more likely to bolster customer relationships and brand allegiance. A large part of this process includes retailers speeding up their testing and adoption of scalable Internet of Things (IoT) and RFID to improve inventory selection, movement, loss reduction, and margin performance via almost real‑time and prognostic modeling.

Retail Robotics

While not new this year, smart machines including autonomous mobile robots, or AMRs, are generating more brainstorming and actual testing across the entire retail value chain. Robots combine AI, IoT, and mobility for certain use cases. Walmart, for example, started using merchandise‑scanning robots to better maintain on‑shelf availability in 2017 but severed their relationship with the robot’s manufacturer in 2020 and reassigned store associates to shelf scanning possibly for ROI reasons. Human flexibility compared to current AMRs is unique since the same associate assigned to scan bar codes or RFID tags can be quickly moved to the point of sale or stockroom or can provide situational customer service if trained to do so. So robots need to be carefully chosen for those environments or tasks they’re best suited for. For example, Walmart is still deploying robots to clean floors, unload inventory, or even pick and move products in their distribution centers (DCs) or even large or former stores serving as small DCs. Also, robots can conduct more than one duty, such as cleaning while scanning shelves and detecting spills.

Other AMRs like “Marty” are detecting debris, spills, and other hazards in a planned rollout of up to 500 Ahold’s Giant Foods Stores. Lowe’s has selectively deployed “LoweBot” to assist customers with where to find items while helping maintain inventory monitoring, so employees can interact with shoppers while moving across the store finding a range of user‑experience opportunities. Additionally, Giant Eagle and Schnucks have been using a Simbe robot to monitor on‑shelf levels, while Australian retailer Woolworths has been using an AMR to detect and even clean up spills. The UK’s Tesco uses RFID reader robots to maintain high in‑store inventory accuracy. Some companies are exploring using robots to charge electric cars while customers are shopping. By the way, all robots are not AMRs, including a test Lowe’s is doing with Virginia Tech on smart exosuits designed to help employees more safely lift heavier products. Drones are yet another example of smart machines being more frequently used in retail environments.

Parking Lots Are Part of the Store

We’re talking a little about how robotics will and are affecting in‑store and supply chain operations, but we also should discuss how AMRs will be used in what we call Zone 4 or parking lots. Loss Prevention Research Council (LPRC) community members have long discussed how the first and last thing customers and employees experience in a physical store visit is the parking lot. Whether traveling by car, bus, or on foot, the parking area is important to “get right.”

Customers then to avoid places where they feel intimidated by people approaching them for money, dark areas, suspicious vehicles, publicized attacks, stolen cars, loitering teenagers, or vehicle burglaries, so great merchandise, interior store aesthetics, and friendly employees may not be enough to keep them coming back, especially when they can just order items on their smartphones.

The LPRC is preparing to explore parking lot AMR options to greet and assist customers, perhaps providing comfort to employees and customers as their cameras provide real‑time video and audio feeds to place guardians. As previously mentioned, robots might in the future provide charging to electric vehicles, report hazards and violence, collect shopping carts, and clean up broken glass, for example.

Digital Ecosystems

Data around what’s being purchased, lost, and stolen, where and when the events cluster, and adding in situational data around weather, staffing and in‑store customer levels, current state display, and protective action are helping retailers better focus for efficiency and margin. But this process needs data, and the data come from transactions, sensors, and other sources.

The IoT is a network of connected sensors and devices picking up and streaming huge data flows for quick on‑premises or cloud computing. This IoT digital ecosystem might include robots and cameras and leverages machine learning to immediately enable more efficient processes for onsite and executive decision‑making. Retail robots working precise and programmed pathways can provide a critical advantage since they might capture more on‑shelf, purchasing, and human behavior details consistently across place and time than people who are frequently moving more randomly around, not always very accurately, and off responding to situations.

European retailer Auchan is using AMRs with camera “eyes” digitally recording all shelves to provide near‑term in‑ and out‑of‑stock insights for quick action as well as midterm planning. Retailers are exploring numerous digital, visual, and aural sensors to be more responsive to customer and employee shopping, working, and even safety needs. Sensors help establish customer counts correlated with purchases, path‑tracking, loitering, searching but not finding, purchasing, and stealing or aggressive behaviors, attention spans, and emotions. Customer‑scoring systems recognize micro‑expressions to sense potentially unhappy customers and help spur on corrective actions. Combining multiple sensors, including those on AMRs, which provide mobility and consistency, provides a more complete understanding on what, where, and how buying and stealing is or is not occurring and puts it into context, including time periods and display or other factors. Other retailers are considering testing robots that can speed selected apparel items to and from fitting rooms where employees are scarce.

AMRs in Action

Zebra robot
This robotic retail inventory management system was demonstrated in the Zebra booth at the 2020 NRF Big Show in New York City.

Roaming robots can continuously gather data on what products are and are not shopper‑available on fixtures and, if they’re running throughout a store’s open hours, can generate a more complete picture on what customers are dealing with across categories, issues, and places. By tying together the out‑of‑stocks with event video footage, robot‑collected data can provide a more real‑time or emerging picture on what items, like toilet paper, are suddenly selling out or what items have stopped selling but still need to be replenished because of theft. These real‑time detections with other evidence will help leaders more quickly take advantage of new opportunities while rapidly responding to better‑defined problems. Robots might soon be able to respond to unexpected issues themselves. Customer loyalty and company profits are at stake.

Another cleaning AMR benefit is they automatically upload their what, where, and when cleaning information, and the data tend to be more timely, complete, and accurate. Not many employees meticulously report and map their cleaning protocol, so area and place cleaning may not be as complete or even meet expected standards. And in a pandemic or even in heavy competition with other chains or buying channels, extra cleanliness can be a benefit.

Augment not Replace: Five Robot Ds

As expected, some employees and unions have expressed concerns about being replaced by robotics, and in 2017 Forbes listed the frequently used “five robot Ds” to describe some of the reasons robots might be preferrable to human counterparts. These include:

  • Dirty. Robots do the jobs that are really dirty, smelly, or sometimes a bit dangerous, such as inhaling fumes. They do the jobs like spray‑paint cars, move things back and forth through mines, and mine things out of the ground.
  • Dull. Robots are also good at repetitive and boring jobs like daily putting thousands of lids on bottles, putting hundreds of small toys together, or turning millions of items around on a conveyor belt each month.
  • Dangerous. Some robots do the very dangerous jobs like defuse bombs, assist highly contagious medical patients, walk on Mars, handle nuclear plants, and other dangerous jobs that could have taken the lives of many human beings.
  • Dexterous. Robots are used for precision with their arms able to accurately and consistently move very small things and are being used to conduct highly precise surgeries. These robots can be operated across the digital ecosystem, so physicians can conduct surgery on patients in distant lands, for example.
  • Domestic. Increasingly AMRs are used for household cleaning, automatic lawn mowing, elder care, or even children’s toys. They help and entertain humans, do cleaning jobs, and call for help during an emergency.

Robots and People in the Same Aisle

One issue often considered with AMRs is how people respond to them. And as imagined, that depends. A person’s response to their first robot encounter in a store can be excitement, amazement, or uneasiness. Age, situation, and the robot’s physical size, appearance, and even its movements can play a role in an emotional response.

So robot developers and users spend considerable time exploring varied perceptions based on robot appearance and action options to dial in a façade that can accomplish the mission(s) while not “freaking out” legitimate place users like customers, especially since AMRs are designed to enhance store aesthetics. Some research shows people anxious about human‑to‑human interactions are often less nervous with a robot. Kids generally like robots, and regardless, people quickly get used to them.

Part of the decision‑making process includes whether the robot should be clearly store branded or rather convey a high‑tech facade. Should the AMR appear sharp or a little goofy? A lot of objective customer feedback can help this process, and if flexible, the robot’s exterior can be modified as more interaction and response data are collected.

AMR Survivability

Another real‑world consideration is people picking on or even attacking robotic devices. The New York Times featured an article in 2019 describing why some humans are prone to bully robots. The tendency is evidently global with attacks reported from Japan to Moscow and across the US. People have kicked them over, hit them with bats, punched them, intentionally driven into driverless cars, thrown tarps over them, and more. One security robot immediately sent images of its attacker to the police resulting in an arrest. Fear, novelty, showing off for others, or just treating robots the same way they treat other humans are possible explanations. But for retail use, AMRs need to be durable in the same way display fixtures and other items must hold up under constant human exposure.

Robot Care and Feeding in the Real World

Robots, like anything, will break, need updates, be damaged, and need power to operate. Store size, design, constant layout changes, employee size, and capabilities are also considerations. The AMR’s primary and perhaps secondary missions will drive ROI and numerous requirements, but mobile robots tend to be cordless and wireless, so data transfer and even battery charging frequency, type, and location need to be addressed. Companies are working on wireless charging to enable a single robot to operate for much longer time periods, reducing the number of units needed per location.

Robotics, IoT, AI Next

Retailing is rapidly changing, especially brick and mortar. COVID‑19 infections, mobility changes, and evermore convenient buying options are driving retailers to accelerate change. Retail companies need to change a customer’s in‑store experience, including aesthetics, convenience, and even safety and security. A combination of AI, IoT, more powerful on‑premises computers, and AMRs is enabling retailers to much more quickly and accurately understand how store layout, merchandising, transactions, cleanliness, and safety options affect store visit frequency and duration and customers’ buying behaviors while in the store.

With buy online and pick up in store (BOPIS), buy online and return in store (BORIS), curbside, store‑to‑store service, and home delivery constantly evolving as well, robotics offer new ways to understand, serve, and protect employees and customers in varying conditions and environments.

Stay Updated

Get critical information for loss prevention professionals, security and retail management delivered right to your inbox.