Walmart’s test of shelf-scanning robots, Oct. 2017 – Photo: Walmart
Will shelf scanning robots put an end to out-of-stocks?
Of the numerous types of robots that are catching on for use in the front of the store, robotic shelf scanners are the ones that retailers see being most effective, according to a new RetailWire Research study.
The study, conducted in conjunction with Brain Corp, found 59 percent of retailers said shelf scanning is the function or application that can be most effectively filled by robots. Order picking came in second at 47 percent, followed by moving product loads (35 percent), pricing accuracy checks (35 percent), floor cleaning (21 percent) and planogram compliance (21 percent).
Retailers’ emphasis on using robots for shelf scanning likely comes from the perception that the technology is a consistently reliable means of maintaining in-stock positions and the successful implementation of omnichannel tactics, including store pickup and delivery.
At least one test of shelf-scanning inventory robots did not go according to plan. Walmart announced late last year that it was scrapping its pilot of shelf-scanning robots. The chain found that in-stock positions could be more cost-effectively managed by human associates. Walmart did, however, continue to use the floor scrubber robots that it had been piloting to handle front-of-store janitorial tasks.
In addition to shelf scanners, other robots that roam store floors and undertake shelf-related duties have been coming to market recently.
Retailers including Ahold Delhaize have been piloting UV disinfection robots, according to Progressive Grocer. The robots roll down aisles using projected UV light to inactivate potentially dangerous pathogens lurking on shelves.
The sales floor is not the only area of retail where robots are being seen with a greater frequency.
Nearly three-quarters of retailers in the RetailWire study say that their use of robots in warehouses or distribution centers has increased due to the novel coronavirus pandemic.
This trend can be observed in practice with Kroger, which has built large robot hive warehouses in the U.S. in partnership with Ocado, after that provider had success in Europe and Canada.
- Robots in Retail: Examining the Autonomous Opportunity – RetailWire Research
- Will Walmart’s decision to scrap robots have far-ranging effects? – RetailWire
- Get Ready for More Robots Cleaning Grocery Stores – Progressive Grocer
- Will a skyscraper store change how groceries are delivered in cities? – RetailWire
BrainTrust
Gene Detroyer
Professor, International Business, Guizhou University of Finance & Economics and University of Sanya, China.
Shawn Harris
Board Advisor, Light Line Delivery
Ananda Chakravarty
Vice President, Research at IDC
Discussion Questions
DISCUSSION QUESTIONS: Where do you see retailers most likely to put robots to work in stores and other facilities? Do you think the cost-effectiveness question will be answered by robotics companies sooner rather than later?
Robots may help with keeping items in stock, however that presupposes that all aspects of the retailer’s distribution and fulfillment operation are properly optimized. Target is a devil for out-of-stocks (it’s not the only one), and it’s not because no one notices that items are not on the shelf: it’s because they have issues which stretch further back into the supply chain. Robots won’t necessarily solve all of those issues.
The shelves in my local Target are ALWAYS empty!
Target’s in stock conditions are also quite terrible at most of its locations where I am in NV. One low volume location with only a few of everything on the shelf has the fewest out-of-stocks. They have improved in the past month from how they were, though.
Target is supposed to have an employee walking around who scans the out of stocks and it sends a message to the backroom employees to find the item and fill the shelf. Not sure if my local store in Reno, NV which is really poorly run is bothering to do this, but let’s assume they are doing it because Target tracks metrics like that pretty heavily and I’d be surprised if they were skipping it.
Improved visibility/demand sensing in-store, enabled by intelligent automation, provides the deterministic data required to inform net new inventory policies and processes upstream. Helping to answer the question, what needs to change upstream to meet the aggregate needs we now see in-store.
The use of robots in warehouses and backrooms of stores is the most logical location as you don’t have to worry about competing for space with customers. However, with better sensors and scheduling robots during non-busy times, having robots on sales floors has become practical. As labor costs continue to rise and the cost of robot technology becomes more economical due to economies of scale the ROI is becoming positive. Robotics, especially in micro-fulfillment centers and dark stores is a hot trend that is expected to continue.
Empty shelves can be restocked by either human or robot only if the product is actually in the store. Often, the product is not in the store because of ineffective inventory forecasts. This is a function of poor forecasting. Use of AI in forecasting is critical and is being adapted and adopted at spotty rates.
To date, robots in retail seem to have been a solution trying to find a problem. Finally robotics have found a purpose in shelf scanning. It was nice to have a robot tell the manager to get someone to clean up an aisle that had a spill, or that greeted a customer, but we all knew that those applications were not viable long term. If retailers can do a cost/benefit analysis of the revenue they lose with out-of-stocks, then they may see that the shelf scanning application, long term, makes sense.
How about actually putting the goods on the shelves, not just advising that there is an out-of-stock. Have you ever noticed in an apparel store the sales person refolding the merchandise after it was handled by shoppers? Can’t a robot do that?
We’ve been talking about shelf-scanning robots for a decade now. Out-of-stocks and perceived out-of-stocks are a huge problem for retail, so there should be adequate incentive to make robots work. One of these days we’ll get it right, assuming that we can fix the longer-term supply chain issues.
Shelf scanning and driving a more optimized replenishment process is ideal for grocery operations to leverage robotics. In what was once a very human and manual process, led by intuition and experience, robotics, combined with data, analytics, and AI, would absolutely help to mitigate the out-of-stock situation at the grocery and retail pharmacy operations.
There are several leading solution providers, including Zebra Technologies, that are driving this change. The SmartSight robot is a hardware, software, and services solution that roams the store’s aisles, checking the shelves, using a combination of computer vision, machine learning, workflow automation, and robotic capabilities. Zebra claims it increases available inventory by 95 percent while reducing human time spent wandering the aisles to do inventory manually by an average of 65 hours per week.
There are clear labor savings in leveraging robots but, more importantly, the grocery stores will be able to be more predictive to keep up with the changing consumer demands.
I agree with this post.
There are many potential uses for robots in retail (that has a ring to it). Checking shelf inventory levels requires RFID tagging or the ability for the robot to recognize holes in the shelf visually. The former is not universal and the latter is not cheap. On the other hand, picking and floor cleaning and shelf disinfecting are easier to implement and probably less expensive.
I don’t think robots are a great solution for in-store. Distribution or fulfillment centers maybe, but you certainly wouldn’t want them in the store during open hours, when they would become toys for kids passing by or an object for theft.
I would expect the growing use of robots in DCs and warehouses where workflow can be steady and predictable. I also expect to see robots in stores, but the sales floor is starting to get very crowded with the growth of fulfillment teams and carts picking orders. If it gets too congested with added robots it will have a negative impact on traffic and sales and therefore cost-effectiveness.
“Great, they’re out of Cheerios again, the aisle ahead is blocked like the Suez Canal by a shopper who brought 3 kids, the Instacart shopper snatched the last pack of hash browns right as I got there, and now a Dalek is screaming at me.” OUT OF STOCKS WILL BE EXTERMINATED, indeed.
As Neil says, only if the entire distribution system is aligned and, gee, nobody’s is. Also, don’t discount the creepiness factor of robots in the aisle while you’re trying to shop. Backroom and warehouse/DC makes more sense to me.
Having the robots collect inventory in the store is only the tip of the iceberg. There are many supply chain processes that need to support moving goods to the shelf that must be connected and probably modernized to make the collected shelf data effective. And as Walmart discovered, their customers are not comfortable. It is more likely that robots will work in the backrooms or warehouses where they can be quite effective at moving products where they need to go next without getting in the way of the shopper.
Where? In the warehouse/distribution center, in the back room, in the store. In fact, anyplace where there is a mundane and repetitive job. The capabilities of robots are increasing and the cost is decreasing. If for a specific task, the ROI of robot implementation does not pay out today, be assured it will pay put tomorrow.
But, please, like the fulfillment shoppers that clog the aisles in the supermarket, keep the robots out of the aisles during shopping hours.
Robots seem suited for replacing humans with repetitive tasks such as continuous heavy lifting, picking, packing, labeling, etc. Likely an entire warehouse operation can be fully robotic.
I suspect in the near-term they are effective for counting or managing inventory but with computer vision and other innovations, cameras will be able to monitor inventory levels from a ceiling or elsewhere; no need for roaming robots.
Cost effectiveness comes at sufficient scale and when tasks can be done more efficiently than humans.
Robots in-store solely for the purpose of inventory detection may not provide the best ROI. Also, they’d never be able to manage inventory in real-time. If the bot is scanning aisle 2, it’s not checking aisle 10, so that’s hardly an advancement. And they can be intrusive to shoppers.
Alternatives are cheaper, multipurpose, and offer lower maintenance. Consider that most retailers have loss prevention cameras already in the ceiling. Those could be used to detect shelf voids too (with the right resolution and cognition). And advances in battery powered electronic shelf labels can also double as stock detectors and planogram compliance.
Robots do have a place in retail, however. They can be used for repetitive tasks like order packing and fulfillment in the back room. And why not use small autonomous bots to deliver groceries for pickup in the parking lot?
If Walmart discontinued the use of robots last year, they did not see the long term value of having them on the sales floor, although I can see the value of robots in fulfillment centers or dark stores to process orders for customers.
In terms of looking for shelf issues (stock-outs, planogram compliance) robots make a lot of sense – firstly they can relay a “restock now” signal immediately and the restock can be carried out while they are looking at other areas. Additionally they can be added to an existing store far more easily than smart shelves or other approaches that detect the amount of stock actually on a shelf.
In a word – no. When you think of the number of people who go up and down the aisle with high powered computers in their hands (customer order pickers) or the number of video cameras that are and can be placed in the store (for a lot less than a robot), you have to wonder why out-of-stocks aren’t a quaint old tradition no one remembers. Until some of the fundamentals are addressed (like what’s in the back room vs. whats on the shelf, and the supply chain issues behind the back room,) I’m afraid the robots will just keep passing the same stock-out over and over again, dodging Instacarters and shoppers while desperately heading back to the charging station.
Excellent point Peter. In my experience just finding the inventory in the back room would require some really advanced AI and probably adoption of RFID for everything in the store.
Robots are technological tools, sadly dependent on human tool users. Knowing that you are out-of-stock on a shelf isn’t useful in real-time if additional inventory isn’t physically present in a store because somebody ordered incorrectly. Walmart’s decision makes sense if you think in terms of the fuzzy logic that schedules elevators. Humans randomly passing by shelves should be more efficient at spotting stock-outs than robots operating on a fixed schedule. Cost effectiveness is not a fixed measure. If we deconstruct it we see in many – if not most – cases, there are more nuanced elements that contribute to total cost savings such as the ability of a human associate to upsell, deliver service, listen to a consumer, etc. It isn’t all about the direct cost of performing a single action.
We are in interesting times where retailers are having to more aggressively figure out how to get a tube of lip balm to a front door, instead of a pallet to a store. This is a deep structural change. In addition, an increase of minimum wage to $15 per hour is looming. In order to have the resource to apply to addressing immediate customer experience needs while also innovating forward, retailers need significant advances in productivity. I have been calling this transformational productivity. Shelf edge scanning robots can help with the immediate need of providing the inventory visibility required to support retailers in keeping brand promise for those shopping in-store and at home. Shelf edge robotics can also provide that productivity, which can be applied to customer service, BOPIS, store safety, and invested into innovation. Solution pricing has reached cost effectiveness, with a growing list of use cases to drive more value.
Robots will not eliminate out-of-stocks at the shelf in retail. Period. Out-of-stocks in retail are almost always caused by forecast inaccuracy and supply chain issues upstream from the store. Robots are of little use except perhaps for data integrity. If the goods aren’t in the warehouse or at the store there’s little that can be done.
I do see more use cases for robots. Warehousing for sure, that’s one place where they’re more efficient, can run 24/7 and perform activities efficiently and accurately.
Inventory management is heading into the future. Robotics will help manage the shelves and the warehouse. Combine that with AI and you’ll see advantages in the form of no over-buying or over-stocking and out-of-stock mitigation.
The big if here is accurate forecasts. In many categories, that just not going to happen consistently.
Can robots give a bald man a haircut?
No single solution can solve for out-of-stock challenges in stores. More likely, it takes a multi-faceted solution where robots can play a role. This might mean more strategic use, such as during non-peak hours in customer-facing areas of the store. As technology advances, other solutions to detect empty shelves may become more practical and cost-efficient. However if out-of-stocks are a persistent issue, reporting of them might be redundant and offer little value.
I don’t know if robots in the aisles is at all a good idea. My experience around robots is that while humans in the aisles coordinate well with each other, a robot doesn’t comprehend situations quickly enough or accurately enough to be a subtle presence. Humans will always have to give the robot primacy.
That said, I also wonder whether there really is a financial value here. Out-of-stocks on the shelf happen for two reasons: high demand which needs to be re-supplied from the back room and high demand which outpaces the DC’s ability to re-stock. My personal experience is I rarely find an out-of-stock which can be re-supplied from the back room (except with produce).
So once we eliminate all of the out-of-stocks due to DC problems (no extra product in the store), is there still financial justification?
I struggle with the ROI on a robot vs a front-line retail worker, even at $15.00/hr plus benefits.
Since grocery retail is a low 2 percent margin business with direct exposure to end-consumers, it makes huge business sense to drive smart automation that is a win-win for the store and the customer. For customers, the increasing presence of helpful robots in stores will translate into hassle-free, personalized shopping experiences leading to higher satisfaction. As for retailers, apart from driving conversion, they can increase their profitability through improved productivity, lower costs and better inventory management. I wrote two articles last year about how robots are being piloted in retail stores – one that covers general use cases and the second one specifically related to micro-fulfillment.
We’ll see more robots behind the scenes in warehouses and DCs before we see them in the store. For the store, they are expensive, cost-inefficient, and require additional human resources to manage. That said, so were the first trains that launched the industrial revolution or the first of any tech advancement. Each subsequent set of robots brings us closer to an effective solution in retail robotics. It won’t be long before they become standard in some form or another- maybe a mechanical device that doesn’t scan from the floor, but from the shelves themselves or from a camera positioned above. Companies like Walmart et. al. will continue to experiment and we’ll see more effective solutions brought to the market.
Ananda, I think you’re spot on.
I see a catch 22 here. The robot needs a “neat” store to work with. Walmart is not really a neat store most of the day. It is a mess most of the day. Pallets all over even in the daytime, customers all over, misplaced items all over. The robot may pick up an empty shelf but maybe the item is just in the wrong slot, or sees stuff on a shelf (which is actually clutter) that is actually missing what it is supposed to have and gives that shelf a pass.
So the catch of course is the other stores that may be “neat” also tend to be more service focused and I am not sure how comfortable customers in a higher end store would be with a robot walking around the store looking at the shelves. Customers expect service in higher-end stores.
Also, I will point out here that Walmart continues to use a competing robotic device for floor cleaning.
In-store sensing robots are mere toys unless the data they gather is linked to a robust perpetual inventory and automated ordering system on the back end. Detection and remediation is the wrong operational strategy to address out-of-stocks. By the time they are discovered, the damage is done!
Forecasting plus inventory optimization are the correct strategies for fast-turning product categories. The latter may be enabled by any number of sensing mechanisms – handheld scanners have worked beautifully for this purpose for many years.
Can semi-autonomous hardware successfully replace fleshware in this scenario? Robotics investors surely hope so. Machine vision tech is advancing at an astonishing pace, and those rolling shelf scanners are close to delivering data quality that rivals human-gathered. So are stationery shelf cams and other types of sensors, including mobile phones.
So a fair argument may be made that tireless ‘bots can replace people for those tasks. But they can’t search the back rooms or restock shelves or refold sweaters. And they can’t drive effective decision-making unless they are linked to forecasting, PI and CGO systems that already work.
Robots may have a permanent place on a grocer’s list of employees. Some of them may do a good job and ultimately may be cost effective. But two things will never change: one, robots take jobs away from people; two, shoppers — at least the ones I know — do not like seeing robots in the aisles. I don’t think grocers care about either one.
I agree with you all: the problem would be almost solved if the supply chain is totally optimized. But, concerning the robots, wouldn’t it be a better decision to use a control and tracking via RFID inventory? Cheaper, more efficient and more scalable.