Jobs

Shop floor jobs robots can take on

Four jobs fill most of a grocery store's floor hours outside the checkout. Here is what each involves step by step, what a robot has to get right to take it on, and how to tell whether it did.

Updated 11 November 2026

Restocking shelves

Getting product from the backroom onto the shelf, faced and in the right place.

  • Restocking means taking delivered cases from the backroom or a roll cage, opening them, and putting each item in its place on the shelf according to the planogram, label facing out.
  • It is the job that decides whether a shopper finds the product. A shelf scanner can tell you a gap exists; only restocking closes it.[5][1]

What automating it really involves

  1. Know what is missing: a shelf scan, a gap list from the store system, or the robot's own view of the shelf.
  2. Find the stock: in the backroom, on a cage or in an overstock slot above the shelf.
  3. Move it to the aisle without blocking shoppers or fire exits.
  4. Open cases and handle each item: boxes, bottles, bags, jars and trays.
  5. Place and face items in the planogram position, at every shelf height.
  6. Clear away empty cardboard and report anything it couldn't do.

The hard parts

  • The variety of the range: a robot that handles cereal boxes may not handle eggs or bagged salad.
  • Shelf heights from floor to above head height, built for a standing person's reach.
  • Doing it during trading hours, with shoppers in the aisle.

Robots that do it

How to measure it

  • Facings or cases restocked per hour, timed on your own shelves
  • Share of tasks completed without a person stepping in
  • On-shelf availability before and after, measured the same way

Stock rotation and freshness

Moving older dates to the front and pulling expired product, so fresh food sells before it spoils.

  • Stock rotation means putting new stock behind old stock so the earliest date sells first (often called FIFO or FEFO), and checking for and removing out-of-date items.
  • It matters most in chilled, fresh and bakery, where dates are short and waste is visible to shoppers.

What automating it really involves

  1. Read the date on each item or know it from the delivery record.
  2. Pull existing stock forward and place newer stock behind it.
  3. Find and remove items past their date, and record what was removed.
  4. Work in chilled cabinets and freezers, where it's cold and space is tight.

The hard parts

  • Reading small, varied date labels reliably.
  • Reaching into the back of cabinets and handling soft or fragile packs.
  • It is a physical job: a robot that only looks at the shelf can flag dates but not rotate stock.[1]

Robots that do it

How to measure it

  • Expired items found on shelf in spot checks, before and after
  • Minutes per cabinet rotated
  • Waste written off in the categories it covers, from your own records

Order picking for online grocery

Collecting the items for online orders, from the store's shelves or from a dedicated store of stock.

  • When a store fills online orders, someone walks the aisles with a cart or handheld and collects each item, chooses substitutions, and packs chilled and frozen items to keep them cold.
  • There are two broad ways to automate it: pick from the shelves shoppers use (a mobile robot in the aisle), or move stock into an automated grid and pick there (micro-fulfilment).[7][12]

What automating it really involves

  1. Decide where stock lives: on the shop floor, in a grid, or split between them.
  2. Locate each item, grasp it without damage and place it in the right tote and temperature zone.
  3. Handle what the robot can't pick: fresh produce choice, bulky items, substitutions.
  4. Hand the packed order to delivery or pickup.

The hard parts

  • Picking on the shop floor competes with shoppers for aisle space; picking from a grid means building and stocking the grid.[12]
  • A grid is a single point of failure: a 2019 fire at Ocado's Andover centre took it out of action until 2021.[11]
  • Produce selection and substitutions still involve judgment.

Robots that do it

How to measure it

  • Order lines picked per hour, including the items a person had to finish
  • Pick accuracy and substitution rate
  • Orders a day the store can fill before and after

Shelf availability checks

Finding gaps, wrong prices and misplaced products before shoppers do.

  • Store staff walk the aisles to spot empty facings, wrong or missing price labels and products in the wrong place, then fix them or raise a task.
  • This is the job robots do most often in stores today: shelf scanners photograph every shelf and send staff a prioritised list.[4][5][7]

What automating it really involves

  1. Capture images of every shelf: a roaming robot, fixed cameras, or a robot that also restocks.
  2. Recognise products, labels and gaps against the planogram.
  3. Turn findings into tasks, ranked by what matters (a best-seller gap before a slow line).
  4. Make sure someone, or something, acts on each task.

The hard parts

  • The last step: a report doesn't fill a shelf. Walmart ended its shelf-scanning contract with Bossa Nova in 2020 after a multi-year test, a reminder that detection alone has to earn its keep.[3]
  • Accuracy on promotions, end caps and fresh counters, which change often.

Robots that do it

How to measure it

  • Detection accuracy against a manual audit of the same aisles
  • Time from detection to the gap being filled
  • Staff hours spent acting on alerts

Turning these measures into money? See how to model payback with your own figures.

Sources

Every numbered claim on this page links here. Open the source to check it.

  1. Automating the last 30 metres of retail fulfilment (Theseus Robotics)
    Humanoids for shelf restocking, stock rotation and order picking; real-time shelf intelligence acted on; Zürich and San Francisco; backed by Y Combinator.
  2. Bossa Nova Robotics (Wikipedia)
    Shelf-scanning robots tested by Walmart in 50 stores from 2017, 300 more added in 2019; Walmart ended the contract in late 2020.
  3. Simbe: physical AI and robotics for retail store intelligence (Simbe Robotics)
    Tally autonomous inventory robot, Tally Spot fixed sensors and Tally RFID; inventory, pricing and planogram checks with task lists for store teams. Performance figures on the page are the vendor's own.
  4. Retail automation (Badger Technologies)
    Robot detects out-of-stocks and pricing discrepancies, prioritises restocking, sends tasks to a mobile app, and detects spills and hazards.
  5. Robots, AI and autonomous technology for smarter operations (Brain Corp)
    BrainOS autonomy for commercial floor cleaning and shelf scanning; announces an autonomous mobile manipulator for stocking and picking.
  6. Ocado (Wikipedia)
    Ocado Smart Platform licensed to Casino, Sobeys, Kroger and Coles; 2019 fire at the Andover fulfilment centre caused by a fault in a battery charging unit.
  7. AutoStore: grocery and retail automated storage and retrieval (AutoStore)
    Cube storage system of robots on a grid, with grocery as a listed industry, a multi-temperature option and robotic picking at ports.