Hacker News Reader

Or it's an experiment. They're giving it a lower importance job during the pilot program. Hard to imagine the initial deployment being on a mission critical path for the store's operation.

It's probably because stock checking is a simple repetitive task that humans tend to fail at due to boredom and mental fatigue while robots would accel at it because of the fact that it's simple, well defined and boring.

It's traditional to spend 45-180 seconds in the back room, staring at your phone or talking to a coworker, before returning to the sales floor to tell the customer that you're sold out. It's like the Hello World of stocking product.

In my experience about a 10-15 years ago when I worked in a super-market there was no "back room" really - it was just a staging-post for getting stuff off the truck, putting it onto trolleys, then taking it out onto the shop floor to put on the shelves as quickly as possible.

I dunno if people expect there to be this huge warehouse of stock "out the back" of every supermarket with loads of shelves holding tins of beans or loaves of bread or whatever just sitting there. The shop floor was the "back room"!

I used to just tell folks that what they see is everything we had, so if they cant see it then we dont have it.

Even with fancy modelling and "just-in-time" delivery stock-outs still happen - I remember one year working on the last weekend before Christmas. We "sold out" of turkeys - in reality it was just that people had them in their trolleys and hadn't checked-out yet so people were stealing them out of other customer's trolleys when they weren't looking! Happy Xmas! :-)

The exception to this is milk and bread. There’s always more bread and milk in the back.

Or somewhere else. Some stores have the same item at multiple locations which one may not know.

Ahh yes.... "Stealing turkeys out of fellow customers trolleys", the TRUE spirit of Christmas.

When I worked in a supermarket in the late 90's, we did have a pretty expansive back room filled with stuff that had been taken off trucks but not stocked on shelves. However, it was next to impossible for a front-end person to actually get to any of the products - they were stacked on pallets 3 to 5 rows high and about 5 rows deep with no clearance between them. Occasionally someone would ask me if something was in the back, but unless it was in the dairy fridge, or a case of it was actually sitting on a cart in the hallway, I'd tell them I couldn't find it.

It only checks inventory at this point because there are no robots that exist that are capable of doing the rest of it safely and effectively such as stocking the shelves in those uncontrolled dynamic environments.

When they get more capable robots, they will replace more jobs.

Once robots can fix the shelving problems cheaper than humans then they will replace those jobs.

As it stands total cost to fix an individual shelving problem likely dropped. In this case it'd be surprising if Walmart employed more people to fix shelving problems, but in other contexts we may have seen induced demand for more employees because the ROI for employee time improved.

Not all increases in efficiency reduce the # of total jobs, we should expect to see the total # of jobs increase in some circumstances.

And the workers are now focusing more on processes that could be argued are more skilled, like restocking the shelves, fixing prices, etc., than they were doing before, like identifying issues.

They are trying to increase the productivity of the stock crew when the store is closed, enabling them to shed a couple of relatively expensive overnight stockers per store. These robots won't be operated during business hours, but take over or make easier some of those stocker duties.

The robot will scan the shelves in a pattern, followed by a stocker with a cart to collect all the misplaced items. It will print shelf labels, without the need for a stocker to use a cumbersome printer relatively slowly.

It also to me seems it might be a way for them to in-house inventory control services like RGIS, using robots to replace human counters. If it can detect misplaced items, one day it might be able to count placed items.

It could also be for the same reasons you stated: it's low-controversy and low-responsibility.

Low-controversy, but visible. Wich will introduce people to the idea gently. They can add features to the robots later with less resistance.

availability is a psychologic factor in purchase decisions. The more in the soup pot, the more soup people will eat. Whereas the beloved low stock fear of missing out trigger might be too hectic for retail. The warehousing aspect is interesting too.

Also consider that better inventory data allows better just-in-time inventory refill, increasing $/ft^2 inventory turnover and therefore total $ sales per store. I suspect this is about driving analytics with higher quality, more frequent data.

Yes, I think this is more likely what they are doing. With people shopping the store, they can never really be sure about on-hand inventory even with robot checks. The robot could confirm that an item is on the shelf, and then a moment later a customer could put that item in a shopping cart and not check out for another 30 minutes or more. During that time the on-hand count is wrong.

Stock levels are important for home delivery but this robot is also a logical first step for a fully fledged shelf packing robot.

If the current robot signals something is out of stock but a human determines the robot is incorrect then it means the store map is wrong or the product identification needs work. Once the error rate is low enough then full shelf packing can be implemented.

Have you ever worked in a store like Walmart? Robots won't be stocking shelves without major(!!!) changes to how things are displayed.

Considering I've actually called stores to check whether they have a particular item in stock or not, I imagine you are not too far off in your hypothesis.

in the distribution system I am in, excess inventory is concern. not only identifying over stocked items but rounding up all obsolete inventory taking space in stores and warehouses. then understanding why it is there.

since wal mart doesn't close stores often or deal with private stores I am curious how they account for take backs from such. that alone can upset the inventory of any business.

accurate inventory is so key with both the seller and supplier. I am not surprised they want to focus there. plus another issue is shrink, by having robots fulfill the role the chance of it vanishing on a count is gone

Relatively more real-time big data, AI/ML or at least sophisticated algorithms to detect or even predict shoplifting.

Stock checking seems vastly easier to accomplish electronically (a la RFID tags) than with dedicated robots.

Porque no los dos?

As other jobs get automated, inventory checking could get more important.

The answer is that this is one of few jobs that is simple enough that it can be automated.