Resources / Picking
Piece Picking Automation: Use Cases, Requirements, and Deployment
Piece picking automation uses robots to pick single items from bins. Learn why eaches are hard, which grippers suit which items, and how to prep a pilot.
6 min read
· Updated
November 2026
At a glance
The definition
Robots that pick individual units (eaches) from bins and place them into orders, then confirm the pick in the WMS.
Why it matters
Every each looks and behaves differently, which is why piece picking automation has lagged pallet and case automation.
Where it fits
Ecommerce, DTC, store replenishment, 3PLs, and high-SKU retail with polybags, small items, and frequent launches.
Piece picking automation uses robots to pick single units, or eaches, from bins, totes, and shelves and place them into order containers. The robot has to find the item, choose a grasp, lift exactly one unit, verify it, place it, and confirm the pick to the warehouse management system.
Pallets and cases have had mature automation for decades, because cartons and pallets are uniform. Piece picking is different. Every order line is a single item, or "each," and every item looks, feels, and behaves differently. A lipstick, a bag of coffee, and a pair of folded jeans can land in the same order. That variety is why piece picking has been slower to automate than pallet and case handling.
This guide covers what piece picking automation has to do, why eaches are hard, which grippers work on which items, and what you need in place before deployment.
What piece picking automation actually has to do
A person picking an each makes a string of decisions without thinking. A robot has to make every one of them explicitly:
Find the item. Locate the right bin or slot and identify the target among everything else in it.
Choose a grasp. Decide where and how to hold the item based on its shape, weight, packaging, and position.
Pick it. Lift one item, and only one, without damaging it or disturbing the rest of the bin.
Verify it. Confirm it is the right item, usually with a barcode scan or visual check.
Place it. Put it into the correct tote, order container, or downstream process, in the right orientation if that matters.
Confirm it. Report the completed pick to the warehouse management system so inventory and order status stay accurate.
Why eaches are hard to automate
Packaging variety. Polybags, blister packs, shrink-wrapped bundles, glass bottles, and reflective packaging each challenge cameras and grippers differently.
Deformable items. Soft goods, pouches, and apparel change shape when touched.
Cluttered bins. Real bins are not neatly arranged. Items overlap, labels face the wrong way, and the last few units hide in corners.
Size extremes. Very small items are hard to see and isolate; large or heavy items push the limits of reach and payload.
Constant change. New SKUs, packaging refreshes, and seasonal products mean the catalog the robot picked last month is not the catalog it picks today.
Grippers: matching the end effector to the item
Research competitions show how much gripper choice matters. Cartman, the robot that won the 2017 Amazon Robotics Challenge, combined a suction tool and a parallel gripper in one end effector, and its designers describe that redundancy as a deliberate design choice. A varied catalog can need more than one approach:
Suction grippers work best on flat, sealed surfaces such as boxes, cartons, and rigid packaging. They struggle with porous materials, mesh, and loose or crumpled polybags.
Two-finger grippers handle rigid items with clear edges, such as bottles, tubes, and small boxes, and can pick items suction cannot seal on.
Paddle-style grippers support items from underneath, which helps with flat, soft, or awkwardly shaped products that are hard to pinch or seal.
The more of your catalog a system can cover with the right end effector, the fewer items fall back to manual picking. That is why it pays to ask every vendor which end effectors its system supports, and how quickly it can switch between them.
SKU fit check
Send an email to
Where piece picking automation works today
Ecommerce and DTC fulfillment: single-unit and multi-line orders picked from bins and shelving.
Store replenishment: eaches picked into store totes for restocking.
3PL fulfillment: multiple brands and catalogs in the same building.
Beauty, apparel, and high-SKU retail: small, varied items with frequent launches.
Kit component picking: pulling individual components for kits, sets, and subscription boxes.
Requirements checklist before you automate
A piece picking pilot can stall on missing basics as easily as on robot capability. Before a pilot, confirm you have:
Item data: dimensions, weights, and packaging types for the SKUs in scope, even if approximate.
Barcode coverage: scannable labels and a known label location on each item.
A defined storage medium: the racks, shelves, or bins the robot will pick from, and the containers it will pick into.
A task interface: a way for the robot to receive pick tasks and confirm completions with your WMS.
Floor, network, and safety readiness: clear aisles, reliable Wi-Fi in the pick zone, and a safety plan agreed with your EHS team. For mobile robots, that plan should reference ANSI/A3 R15.08, the safety standard for industrial mobile robots.
Accurate inventory in the pilot zone: a robot sent to an empty slot creates an exception, so cycle count before you start.
An exception path: a defined process for items the robot cannot pick.
Fit call
Send an email to
Deployment: from SKU audit to production
SKU audit. Segment the catalog into items that are easy to pick, items that need testing, and items that should stay manual for now.
Workflow mapping. Walk the site, map how work reaches the pick zone today, and agree where the robot's task starts and stops.
Zone pilot. Run a defined zone with representative items and real orders, measuring accepted picks rather than attempts.
Coverage expansion. Add SKUs from the "needs testing" group as performance holds.
Shift scaling. Extend into nights, weekends, and peak once day-shift output is stable.
Plan exception handling before you need it
Every piece picking system will meet items it cannot pick on the first try: a failed grasp, a dropped item, a mislabeled product, or an empty slot that inventory says is full. What matters is how fast the system recovers. Before production, decide:
Which exceptions the robot retries on its own, and how many times.
Which exceptions route to a person, and who that person is on each shift.
How exceptions are logged so you can spot patterns by SKU, location, or time of day.
Track exception rate and time to recover alongside throughput. A system with fewer exceptions and faster recovery usually beats one with a faster pick cycle.
Metrics that matter
Accepted picks per hour, sustained across a full shift
Pick accuracy and downstream error rate
SKU coverage, by count and by share of order lines
Exception rate and average recovery time
Uptime by shift
Site walkthrough
Send an email to
Common piece picking automation mistakes
Testing only top sellers. High-velocity items are often the easiest to pick. Coverage of slow movers decides how much manual picking remains.
Ignoring bin condition. Robots pick from the bins you actually have. Overfilled, mixed, or damaged bins belong in the test, or should be fixed before it.
Skipping placement requirements. If items must land in a certain orientation or position, test that explicitly.
Leaving out the floor team. Pickers know which items are hard and why. Their input makes the SKU audit more accurate and the rollout smoother.
How Avatar approaches piece picking
Avatar robots pick from traditional racks and bins, so they drop into your existing workflows without a facility redesign or re-racking. Every shift generates training data that makes the robots faster and more accurate on your work, which matters in a catalog that never stops changing. Learn more on the Avatar picking page.
Frequently Asked Questions
What is the difference between piece picking and case picking?
Case picking moves full cases of a single product, usually for retail or wholesale orders. Piece picking selects individual units, or eaches, usually for ecommerce, DTC, or store replenishment. It involves far more item variety, which is why piece picking automation needs flexible grasping, vision, and a plan for exceptions.
Can robots pick polybagged items?
Yes, but polybags are among the hardest packaging types because they wrinkle, reflect light, and deform. Test them explicitly in any pilot, and favor systems with more than one gripper type.
What share of my SKUs can a robot pick?
It depends on your catalog, which is why the SKU audit comes first. Segment items by packaging and handling difficulty, then measure coverage by share of order lines as well as SKU count.
Does piece picking automation replace pickers?
It usually changes where people work. Robots take on repetitive picks and hard-to-staff shifts, while people handle replenishment, quality checks, and exceptions.
Sources
1. Wade-McCue, S., Kelly-Boxall, N., McTaggart, M., Morrison, D., et al. (2018). Design of a Multi-Modal End-Effector and Grasping System: How Integrated Design Helped Win the Amazon Robotics Challenge. arXiv:1710.01439.
2. A3 (July 31, 2020). Robot Safety Update: FAQs. Association for Advancing Automation.
Keep Reading
Solution
Picking
See how Avatar works here.
Visit page →
Picking
High Mix Order Picking Automation
High mix order picking automation buyer checks: SKU coverage, new-SKU onboarding, reslotting, exceptions, and capacity, plus how to build a fair test set.
4 min read
Read guide →
Picking
Robotic Picking Systems for Warehouses
Robotic picking systems for warehouses compared: goods-to-person, AMRs, fixed cells, and mobile manipulation, plus pilot metrics and vendor questions.
5 min read
Read guide →
Ready To Scale Your
Workforce?
Tell us your hardest jobs. We’ll show you what an Avatar robot can do on day one.
