Four Jobs Grocery Store Robots Do on the Sales Floor

The Four Jobs Grocery Store Robots Do

Four distinct machines run in supermarkets today, and each is bought by a different person for a different reason. Shelf auditors answer questions about stock and pricing for the merchandising team. Hazard detectors watch the floor for the safety and operations team. Autonomous scrubbers replace a nightly cleaning shift. Backroom and micro-fulfilment picking systems serve the online order business, which usually reports to a different budget line altogether.

The split matters when you evaluate quotes, because a supplier selling one of the four will describe benefits that belong to another. A shelf auditor does not clean anything and a scrubber cannot tell you a price tag is wrong. Ask which of the four jobs a machine performs before the demo begins.

Our wider service robots hub covers the same machine classes across hotels, hospitals and restaurants, where the constraints differ again.

Shelf Scanning and Price Audit

A shelf-scanning robot drives the aisles taking images and returns a list of gaps, wrong prices and misplaced items. Simbe Robotics builds Tally, the best-known machine in this class. Tally 4.0 was announced in January 2026 with up to 12 hours of runtime, and units reached customers from the middle of that year. A full shift of runtime matters because it lets one robot cover a store without a mid-day charge stop.

Simbe describes deployments across grocery, club, farm supply and home improvement chains in 10 countries, and B&R Stores put Tally into its Lincoln, Nebraska supermarkets in June 2026. The output feeds on-shelf availability tracking, price and promotion checks, item location, planogram compliance and replenishment. The value shows up as fewer empty facings rather than as saved labour hours.

In March 2026 Simbe became the first retail robotics company to earn UL 3300 certification for Tally from UL Solutions. That certificate exists because untrained members of the public share the aisle with the machine, which is the constraint shaping every store-floor robot.

Spill Detection and Floor Scrubbing

Hazard detection and floor cleaning are the two grocery robot jobs with a direct safety return. Badger Technologies put its Marty robot into Stop & Shop stores from January 2019. That was the first large-scale in-store robot rollout in United States grocery, and the fleet was upgraded across more than 300 stores in 2023. Marty uses lidar and 3D depth cameras. Some of its eight cameras point at the floor and others at the shelves, so one machine reports both a spill and a missing item on the same pass.

Marty does not clean anything. It raises an alert and a person cleans up, and Badger has used human reviewers to check imagery before a cleanup call goes out over the store loudspeaker. That is worth knowing before a demo, because a hazard detector reduces the time a spill sits on the floor rather than removing the cleaning task.

Floor scrubbing is the job robots do end to end. Tennant sells autonomous scrubbers running the Brain OS autonomy stack from Brain Corp, and Avidbots sells its own scrubber into the same accounts. These convert a nightly cleaning shift into a supervised machine run, and our industrial floor cleaning robot guide covers the machines and their coverage rates in detail.

Why Walmart Ended Its Bossa Nova Robotics Contract

Walmart ended its shelf-scanning contract with Bossa Nova Robotics in November 2020, pulling the robots out of the stores that had them. Walmart did not state a reason in its own announcement. Reporting at the time, first in The Wall Street Journal and picked up by CNBC, said the retailer had found simpler and cheaper ways to get the same information.

The mechanism behind that is specific and useful. Online grocery volume during 2020 put far more staff into the aisles picking orders for pickup and delivery. A person already walking every aisle with a handheld device can record shelf conditions on the way. The robot's unique advantage, covering the whole store systematically, stopped being unique. Walmart's United States chief executive John Furner had also raised concern about how shoppers reacted to the machines.

Read the exit as a warning about the business case rather than about the technology. Bossa Nova's robot worked. It lost to a cheaper path to the same data, which is the outcome any store robot faces if the store already has people covering the same ground. Ask a vendor what happens to the value if headcount in the aisle goes up rather than down.

Store-Floor Robots Answer to a Different Buyer Than Warehouse Robots

A supermarket aisle and a distribution centre floor demand different machines, and the difference is the public. When we reviewed store-floor robots against warehouse hardware for this site, the split that kept showing up was the safety standard each one is built to. Warehouse mobile robots are designed against ANSI/A3 R15.08, written by the Association for Advancing Automation for industrial mobile robots operating among trained staff. Retail robots have to work near people who received no training at all.

Simbe's UL 3300 certification in March 2026 is that difference made concrete. UL 3300 covers service, communication, information, education and entertainment robots, which is the standard family that applies once a machine shares space with the public. A robot that meets it is being certified for an environment containing children, shopping carts, wheelchairs and shoppers reading their phones.

The liability profile diverges too, and this is the part vendors rarely raise. A warehouse worker injured by a machine is handled through workers' compensation and an internal safety process. A shopper knocked over in an aisle brings a claim against the retailer, and a machine with cameras on it generates a full record of the incident. That asymmetry is why store robots move slowly, sit low, and are built to yield. Speed and accuracy come second to it.

The warehouse side of the business runs on entirely different economics, which our pages on warehouse robots for sale and warehouse picking robots cover. Online grocery orders are increasingly assembled in micro-fulfilment rooms attached to stores, using systems from suppliers such as Ocado, which is warehouse automation in a retail building rather than a sales-floor robot.

What Grocery Store Robots Cost

No grocery robot vendor publishes a list price, and the reason is that almost none of them sell hardware outright. Simbe, Badger Technologies and the scrubber suppliers all work on subscription or robotics-as-a-service terms, where a store pays monthly for a machine, its software, its updates and its support together. Tennant sells its autonomous scrubbers through its own dealer network on lease or service terms rather than as a catalogue item.

That structure changes what you should ask for. A capital price for one robot means little when the contract is a monthly fee per store. Ask instead for the cost per store per month, the contract length, the renewal terms, and who pays when a customer or a pallet jack damages a machine.

Then pin down the output. A shelf auditor is worth the monthly fee only if the report changes what someone does. Ask how many scans a week the fee covers, and who on the store team acts on the exceptions. A scrubber replaces a defined number of cleaning hours, which is arithmetic you can do against your existing cleaning contract before signing anything.

• Cost per store per month, not a hardware price
• Contract length, renewal terms and any minimum store count
• Scans per week included, and the report format the store team receives
• Who pays for damage caused by customers, carts or store equipment
• For scrubbers: cleaning hours displaced, priced against your current contract

Which Stores Should Not Buy Grocery Store Robots

Three kinds of store should pass. A single independent supermarket rarely clears the fee, because the reporting value comes from comparing execution across many stores and one store has nothing to compare against. A chain that already has staff walking every aisle to pick online orders is in the position Walmart found itself in, and a handheld scanner in those hands costs far less. A store with narrow, cluttered or frequently re-merchandised aisles will fight the machine constantly, since a robot that cannot complete a route produces a partial report nobody trusts.

Two developments would change that answer. If a vendor published a per-store price low enough for a single site, the independent case would open up. If shelf auditing were delivered by ceiling-mounted or shelf-mounted cameras at lower cost, the mobile robot would lose its remaining advantage, and several suppliers are already selling fixed-camera systems into exactly that gap.

A chain that does clear the bar should pilot one job in a handful of stores. Measure the exception reports against what the store team fixed, and expand only where that number moved.