The main types of warehouse automation can be grouped by task: mobile transport, automated storage and retrieval, palletizing, conveyors and sortation, and goods-to-person picking. A project can combine several. Start with the loads, peak moves, storage demand and order profile to identify which functions the building needs.
This guide compares those five task groups, their equipment interfaces and selection inputs. Motionwell’s delivered transport, storage-interface and end-of-line work is on our warehouse and intralogistics page.
On the storage work delivered so far, the rack and the crane came from the storage vendor and the scope on our side was everything that touched them. Goods-to-person picking and high-rate parcel sortation arrive as complete systems from the vendor that builds them, and the integration scope around one of those is the equipment on either side: the replenishment conveyors, the takeaway to despatch and the safety scope at the station.
If the equipment type is already clear, the warehouse automation ROI calculation carries an illustrative scenario from measured work through ramp-up to break-even.
What Are the Five Types of Warehouse Automation?
Use these task groups as a starting point. The final selection depends on the load, flow and storage requirements; several types may contribute to the same problem.
| Type | What it moves | What it changes | What it is not for |
|---|---|---|---|
| Mobile transport (AGV, AMR) | A pallet, cage, tote or cart between fixed points | Who carries loads between stations, and how predictably | Storage density or picking by itself; mobile manipulators and automated forklifts add load-handling functions |
| Automated storage and retrieval (ASRS) | A unit load into and out of a rack position | Storage density per square metre and the accuracy of the location record | Check the density benefit against floor storage, building limits and the range of load sizes |
| Palletizing and depalletizing | Cases or bags onto a pallet, or off one | End-of-line labour, stack consistency and pallet handling | Mixed-SKU order building needs additional picking, sequencing and load-planning functions |
| Conveyors and sortation | Product along a fixed path, with diverts at decision points | Continuous flow between processes, and routing by identity at a scan point | Routes that change, loads too varied for one carrier, or a floor that has to stay open |
| Goods-to-person | Stock brought to a stationary pick station, commonly in totes or on shelves | Walking time per order line and pick confirmation | Pure transport tasks without a picking operation; assess utilisation at low order volumes |
These functions are complementary: a completed pallet can travel by vehicle and conveyor into an automated store. Their expansion paths differ. Fleets can add vehicles until traffic or stations constrain capacity; storage, conveyors and goods-to-person systems can also expand through planned modules, aisles or stations. Include those interfaces and the required building space in the first phase.
What Does Mobile Transport Automation Do, and What Does It Leave Untouched?
A mobile robot moves a load from a station where it was placed to a station where something else will take it. That sentence contains both the strength and the limit. The vehicle owns the journey; it does not own either end of it, and neither end is a route edit.
It suits a warehouse with defined stations and a transport demand that is real and not assumed: pallets between a receiving dock and a storage face, totes between a store and a pack bench, cages between a pick zone and despatch. On the fleet we integrated, every route between the storage system, the receiving docks and the shipping staging area was taken over by vehicles, and the design decision that made it work was the handoff, not the navigation. Each station got a mating powered conveyor at matched height and speed, so a pallet crossed from vehicle to station under chain drive instead of on fork tines. The consequence is that adding a station later is a conveyor project, which the AGV fleet and ASRS case study sets out in detail.
What it does not do is anything at the ends. A vehicle does not store: a pallet parked on a vehicle is a vehicle out of service. It does not pick: an arm on a vehicle is a different class of machine with a different safety case. And it does not fix a station that cannot say when it is ready. Where the load has to land inside a fixture that wants a fraction of a millimetre, the tolerance is bought on the station side, and the mechanisms for that are on our AMR and AGV integration page.
The AGV or AMR question inside this type is a selection problem in its own right, decided by layout stability, shared traffic and whether the floor can be cut, and it is worked through in full in the AMR vs AGV selection guide. For the purpose of classification the two are one type: both move a load between defined points, both need a station that can hand it over, and both take their missions from something above them.
What Does an ASRS Change That a Vehicle Cannot?
An automated store changes two things a vehicle never touches: how much stock a square metre of floor holds, and whether the location record is true.
The density case comes from height and depth. A crane or a shuttle works a rack far taller than a reach truck can serve, and double-deep or multi-deep positions put more pallets behind each aisle face. On the store the fleet above feeds, double-deep pallet locations are served by a stacker crane, with a telescopic-fork stacker crane reaching the second pallet position. That is a building decision as much as a machine decision, which is why the ASRS family is compared on building height and floor flatness before any vendor is named, on our ASRS and automated storage page.
An ASRS can improve location control by recording put-away and retrieval transactions. Keep the physical and digital state aligned through load identification, confirmed completion, duplicate-message handling and reconciliation after faults or manual recovery. Floor-stock locations can coexist with automated storage when the stock system tracks both and their transfers explicitly.
An ASRS is designed for a specified range of load dimensions and weights. Mixed formats can require different carriers, handling devices or storage zones. In a crane-served aisle, crane availability and cycle time constrain retrieval; shuttle systems have a different capacity structure. Include the selected system’s recovery and expansion arrangements in the specification.
The interface an ASRS presents to the rest of the warehouse is narrow and unforgiving: an inbound conveyor at the store face with a profile and weight check, an outbound conveyor, and a controller that accepts a put-away or retrieval order and reports its completion. Our scope on storage projects has been exactly that band, the equipment on either side of the rack and the logic between.
Where Does Palletizing Sit Between the Packaging Line and the Warehouse?
Palletizing is the type that belongs to two departments at once. The cases arrive from a packaging line and the pallet leaves into the warehouse, and the cell is the boundary between them. It is bought to remove the last manual station on a line, where stack quality drifts across a shift and a leaning pallet has to be re-squared before it can be wrapped.
Cases arrive in a controlled orientation, are picked singly or in groups, and are placed in a pattern selected for the load. Completed pallets move to wrapping or collection. Conventional layer-forming machines and robotic cells offer different handling sequences; robotic tools can pick a case, a row or a layer. Four-axis robots can handle cases or bags with suitable tooling, while extra wrist freedom helps where a load must tilt or approach a constrained position. Compare those options on our automated palletizing systems page and see the delivered robotic carton palletizing case study.
Distinguish a uniform end-of-line stack from mixed-SKU order palletizing. The latter adds case selection, sequencing and load-stability planning to the robot and conveyor scope. Depalletizing needs its own trials on incoming case condition, visibility and grip access, especially when the inbound stack was built elsewhere.
The upstream interface includes the required buffering. Where a line delivers in bursts, size an accumulation buffer against the robot’s demand and choose zero-pressure zones when loads must be kept apart. The downstream interface is the pallet’s exit, and it is where this type meets the first one: a full pallet on a discharge conveyor is a station a vehicle can be sent to, which closes the loop between the end of the line and the storage face. Where the packaging machines upstream are part of the same project, the interfaces between them are set out on our packaging line integration page.
Robotic palletizer cost follows from the same boundaries. Case rate sets the robot class, SKU count sets the pattern work and the gripper compromise, case condition sets the gripper, stack height decides whether a lift column or a track is needed, and the conveyor and pallet-handling scope on either side is priced separately from the arm. None of those is a catalogue number, which is why the palletizing page lists the drivers and no price; what moves the arm itself is in our note on industrial robot cost drivers.
A cobot palletizer uses an arm with safety functions suited to the intended human interaction. Evaluate sustained case rate using the actual load, path, pallet height and validated operating mode. Power-and-force-limited motion and monitored-separation phases impose different constraints, so there is no universal cases-per-minute boundary between cobot and conventional palletizing. The distinction is explained in our guide to what a cobot is.
What Do Conveyors and Sortation Do That the Other Types Depend On?
Conveyors commonly connect storage, palletizing, transport and picking equipment. They provide a fixed transfer path where the selected equipment uses conveyor handovers; other systems use forks, carts, shelves or direct robotic transfer.
What it changes is continuity. A conveyor turns repeated moves into a flow. With zero-pressure accumulation zones, sensors and release controls, loads can wait without contacting the load ahead until the next process is ready. That buffering is what decouples a machine that works in bursts from one that works at a steady rate, and it is why a palletizer, a store face or a pack bench is only as good as the conveyor feeding it. The families, and what each one carries, are in our guide to the types of conveyor systems, and how zones, transfers and merges are designed is on the conveyor and material handling page.
Sortation is what a conveyor becomes when it has to make a decision. Product is identified at a scan point, a controller decides its destination, and a divert at the right moment sends it there. At the scale of a cell that is a reject lane with a pneumatic diverter downstream of a checkweigher, which is delivered work. At the scale of a parcel hub it is a loop sorter with hundreds of destinations, which is a different industry with its own vendors. The classification point is that sortation is bought for identification and routing, and it fails on identification before it fails on mechanics: a label that will not read is a product that cannot be sorted, whatever the divert can do.
Where conveyors stop: the route is the asset. Moving a conveyor is a mechanical and electrical project, so a flow that changes with the season is a transport problem for a vehicle. Where forklifts cross a proposed conveyor route, compare a controlled crossing, an overhead or retractable section, and mobile transport. A varied load mix may need different carriers or handling paths.
What Is Goods-to-Person, and Why Is It a Different Purchase?
Goods-to-person inverts the pick. Instead of a person walking to the stock, a shuttle, a robot or a crane brings the tote or the shelf to a stationary pick station, and the person picks from what is presented. What it changes is walking time per order line, which in a piece-picking operation can be a large part of a picker’s day, and pick accuracy, because the station can confirm the location, the item and the quantity as each pick is made.
It suits order profiles where presenting stock to a station removes substantial picker travel. Tote and shelf systems are common; pallet-to-person designs can serve larger unit loads. Compare station utilisation, replenishment work and peak presentation demand. Accurate stock identity and quantity records remain essential to useful picks.
Goods-to-person selection includes storage presentation, station ergonomics and order sequencing as one coordinated scope. The shuttle rack, the lifts, the station and the software that sequences orders to stations are designed together, and the integration scope for a company like ours is the equipment on either side: the replenishment conveyors, the pack bench interface, the takeaway to despatch and the safety scope around the station. Where a project needs one, the vendor of the store leads it.
Which Type Fits Which Problem?
The table reads from the complaint, not from the catalogue. Find the row that describes what is actually wrong, and the type follows.
| What is actually wrong | Type that answers it | What a different type would have done |
|---|---|---|
| Forklift drivers and runners spend the shift moving pallets between the same points | Mobile transport | A store would have densified stock the drivers still had to fetch |
| Storage capacity limits output | ASRS where the building and load profile support it | Transport alone does not increase rack capacity |
| The last station on the line is manual, stacks drift, and pallets get re-squared | Palletizing | A conveyor would have delivered cases faster to the same people |
| Product waits between processes, or has to be routed by what it is | Conveyors and sortation | A vehicle would have carried in batches what needed to flow |
| Pickers walk most of the day for orders of a few lines each | Goods-to-person | A store on its own would have shortened nothing about the walk |
| More than one of the above | Two types and an interface | One type stretched past what it changes |
The last row is the common one. A warehouse with a full building and an end-of-line labour problem is a store and a palletizer, joined by a conveyor and possibly a vehicle, and the interface between them is engineering in its own right, never a single line on the storage vendor’s quotation.
Where Do the Types Meet, and What Crosses Each Boundary?
Every pairing in the table above meets at a boundary, and three things cross it: a physical load, a control handshake and a data record. Assign an owner to each boundary so that the connected controllers use the same transfer conditions and completion record.
| Boundary | Physical handoff | Control handshake | Data record |
|---|---|---|---|
| Vehicle to station (dock, store face, pack bench) | A mating powered conveyor at matched height and speed, or a fixture with a defined docking tolerance | Vehicle present and aligned, station ready, transfer start, transfer complete, both sides confirming load presence | Fleet server reports the move; the stock record holder updates the location |
| Store face to the rest of the warehouse | Inbound conveyor with profile and weight check, outbound conveyor | The storage controller accepts a put-away or retrieval order and reports completion; a rejected load has a defined exit | The stock record holder owns the location; the storage controller owns the rack position |
| Line to palletizer | Accumulation buffer, squaring station, a case presented in known orientation | Case available, robot ready, pattern selected, pallet full, pallet removed | The pattern recipe from a barcode or an MES command; the pallet identity leaving the cell |
| Palletizer to transport or store | Discharge conveyor, wrapper, a full-pallet position a vehicle can be sent to | Pallet complete, pallet wrapped, station ready for pick-up | Pallet identity handed to the fleet server or the store |
| Conveyor to sorter divert | Product spaced and oriented so the scan reads and the divert clears | Scan result, destination decision, divert fired against product position | Identity read at the scan point against the order or the route |
Three items in that table decide more than their size suggests. The first is who owns the stock record. A fleet server, a storage controller and a cell PLC each keep their own state, and the authoritative stock record and its updates must be agreed; on the fleet we integrated, the WMS was updated at each handoff. The second is what each side does when the other stops answering, which is a sentence that has to exist before commissioning, because discovering it during commissioning means discovering it on a live floor. The third is that every signal in the handshake column can be exercised on a builder’s floor against a stand-in for the other side, and should be, which is what the interface section of a factory acceptance test checklist is for.
Define the scheduling boundary as well as the stock-record owner. The WMS can issue work while fleet and equipment controllers sequence their own resources. Exchange priorities, status and completion messages so local scheduling remains consistent with the warehouse plan. The related ownership questions are in our guide to what MES, ERP and SCADA each own.
Which Safety Standard Governs Which Boundary?
Each type carries its own safety case, and the boundaries in the previous section are where two of them overlap. Two documents do most of the work on the types we build.
A palletizing cell is an industrial robot application, and the standard for the cell as opposed to the arm is ISO 10218-2:2025, the application and cell half of the 2025 pair. It is also where the collaborative operation requirements now live, since most of ISO/TS 15066:2016 was incorporated into this edition, which is why a cobot palletizer placed beside a pack bench is assessed under the same document as a fenced cell. What changed in the 2025 revision and what it asks of an integrator is in our guide to ISO 10218.
At a vehicle transfer, permit conveyor motion only when the required docking and transfer conditions are satisfied, inhibit departure during transfer and define the response to a lost condition. The risk assessment determines which parts of that sequence are safety functions and their required performance. For an ISO 13849 design, assess the complete sensing, logic and output chain under ISO 13849-1:2023 and validate its behaviour. An edition change calls for a documented gap review of the design and evidence. Shared vehicle and pedestrian areas also need traffic and recovery assessment; the method is in our machine safety risk assessment guide.
| Boundary | What the safety case has to cover | Governing document for the types we build |
|---|---|---|
| Inside a palletizing cell | Guarding, access, the robot’s safeguarded space, collaborative operation if any | ISO 10218-2:2025 for the application; ISO 13849-1:2023 for each function |
| Pallet exit from a cell | Muting or a defined safe opening where the pallet leaves and a person could enter | ISO 13849-1:2023 for the muting function |
| Vehicle docked at a station | Conveyor cannot run unless the vehicle is present and aligned; vehicle cannot leave mid-transfer | ISO 13849-1:2023 for the interlock |
| Shared floor between vehicles and people | Speed, field configuration, traffic rules and recovery behaviour | Risk assessment first; the vehicle standard for the vehicle |
When Is Warehouse Automation Not the Answer?
Five situations put the fix somewhere no machine reaches, and each one is worth recognising before a shortlist is drawn.
The complaint is slotting. If fast-moving stock is stored at the back and slow stock at the front, a machine will move the wrong stock efficiently. Re-slot first, then measure what is left.
The stock record is wrong. Every type above trusts the record. A store filled from an inaccurate count is a store full of surprises, a goods-to-person station brings empty totes, and a fleet delivers pallets the WMS thought were elsewhere. Fix the count before the machine, or the machine will make the count worse at speed.
The volume is assumed instead of measured. A fleet sized from an estimate of moves per shift and a palletizer sized from a nameplate line rate both end up idle or short. What to measure before either is sized is step one of the ROI calculation.
The building cannot take it. Height under the lowest service, floor flatness, column grid and fire protection decide the storage type before any vendor does, and a leased building with a return condition decides whether a conveyor across the floor is possible at all.
Nobody will own it afterwards. Every type above needs a named owner after handover, and what a fleet in particular needs someone to own is one of the selection questions in the AMR vs AGV guide. Where a site cannot yet name that owner, naming and training one belongs in the project scope alongside the hardware, because the fleet, the store and the cell all degrade quietly without one.
Which Complaint Are You Solving?
Use the task groups to build one material-flow specification. Mark where loads enter and leave, who owns each stock transaction and how each interface recovers after a stop. The resulting equipment shortlist should address measured capacity and handling needs, with the boundary work included in the quotation.