Is an autonomous forklift right for an e-commerce fulfilment warehouse?

It depends on whether your constraint is horizontal and vertical movement rather than picking itself. Where let-down, replenishment and despatch trunking queue behind a small pool of shared trucks, autonomous forklifts usually earn their place. Where the bottleneck is pick-face design or packing capacity, they will not fix it. A site survey and a look at movement data answer this quickly.

Do we have to replace our WMS to run warehouse robots?

Usually not. A fleet management layer sits above your warehouse system, takes work from it over a documented interface and writes completions back, so your existing system stays the record of stock. Where no live interface exists, a scheduled export and import handled by middleware is often enough to start. Replacement is rarely the cheapest route.

Can autonomous forklifts and AMRs from different manufacturers run in the same building?

Yes, where they support a common interface. Open standards such as VDA 5050 let one fleet manager issue orders to vehicles from different suppliers in the same message format. Traffic rules still need setting centrally and each vehicle's capabilities checking, but a separate integration per manufacturer is no longer required. As an independent integrator, this is how FlyWei keeps operators free to add the right machine later.

What drives the cost of automating an e-commerce fulfilment site?

Fleet size and mix, integration scope, and the support model — not a headline price per robot. Fleet size comes out of the flow simulation, integration scope out of how your existing system exposes work, and the support model out of how the site runs at peak. FlyWei quotes these as visible line items and the site survey is free.

How do robots and people share the same aisles safely?

Mixed traffic is normal in fulfilment buildings and is designed in from the survey. Vehicles carry certified safety laser scanners and functional-safety rated controllers, travel at reduced speed in shared pedestrian zones, and follow centrally set traffic rules. Zones, crossings and pedestrian routes are agreed before rollout rather than retrofitted afterwards.

How long does a deployment take, and can it be ready before peak?

The vehicles are rarely the long pole; data access is. Mapping order and stock fields, agreeing exception handling and testing against live data take the time. Projects move fastest where the existing system already exposes a documented interface and a test environment is available, and phased rollouts mean one flow can be live well before the whole scope is.

Peak Surge Capacity for UK E-Commerce Fulfilment | Case Study

Case Studies › Peak-Season Surge Capacity Without Replacing Your WMS for a UK E-Commerce Fulfilment Operation

Peak decides an e-commerce fulfilment centre's year, and it is also when the building runs furthest from its design assumptions. Volumes climb, agency drivers arrive who have never seen the site, and the constraint quietly moves from picking to the pallet flow feeding it. This illustrative case study looks at how an independent integrator approaches that for a UK e-commerce operation: not by replacing the warehouse management system, but by automating the repetitive moves around it with autonomous forklifts and mobile robots drawn from several manufacturers.

Illustrative scenario. This is a representative application study, not a named-client reference. The operator described is a generic composite of UK e-commerce fulfilment sites, and all figures are indicative capability ranges, not measured project results.

Operation profile

Operator: A mid-sized UK e-commerce fulfilment centre — illustrative and unnamed, serving direct-to-consumer and marketplace orders
Building: Typically 15,000–40,000 m², selective and narrow-aisle racking over a ground-floor pick face, with packing benches and despatch lanes at one end
Shift pattern: Commonly two shifts for most of the year, extending to double or triple shifts through peak
Throughput band: Typically a few thousand to several tens of thousands of units a day, with peak multiples of roughly two to four times baseline for several weeks
Load types: Mixed — full and part pallets, cages and roll containers, tote and carton replenishment

At-a-glance application snapshot

Typical capability ranges for the machine classes used here — what the equipment can generally do, not what any site achieved.

Payload: autonomous pallet trucks typically 1,500–3,000 kg; jacking and tote-handling robots typically 150–1,000 kg
Lift height: pallet movers at ground level; autonomous stackers typically to around 6 m; autonomous reach trucks to around 10 m for high-bay work
Aisle width: narrow-aisle machines generally from roughly 1.6–2.0 m; counterbalance classes typically 3.0 m or more
Travel speed: generally 1.0–2.0 m/s laden, automatically reduced in shared pedestrian zones
Runtime: typically 6–10 hours per charge on 48 V lithium iron phosphate packs, with opportunity charging in breaks usually sustaining continuous running
Navigation and safety: natural-feature navigation placing loads to commonly around ±10 mm, with certified safety laser scanners and functional-safety rated controllers, typically SIL 2 / PL d by class

The challenge

Fulfilment sites rarely struggle at peak because the pickers are slow. They struggle because the material flow behind the pick face cannot keep up — and because the extra hands hired to fix that arrive with the least site knowledge at the busiest moment.

Replenishment is the hidden constraint

Every unit picked has to be put there first. Through peak, let-down and replenishment grow in proportion to picking, but are served by a handful of shared trucks also covering goods-in, put-away and despatch. When those trucks saturate, pick faces run empty and productivity falls for reasons that have nothing to do with picking.

Seasonal labour is expensive in more ways than one

Counterbalance and reach-truck work needs a licensed, site-inducted operator — a hard resource to flex up for a couple of months and then release, with the induction burden landing on supervisors at the worst possible time. Sites in this position describe the same pattern: recruitment pressure, more near-misses in shared aisles, more damage in the weeks when damage costs most.

A system that must not be disturbed

Any fix carries a hard constraint: the warehouse management system is the record of stock and the source of order truth, and nobody sensibly replaces it in the run-up to peak. Automation has to take work from the existing system and hand results back to it.

The solution

The design principle is straightforward: automate the repeatable moves, leave judgement work with people, and choose each machine class on merit rather than on who supplies it. As an independent, vendor-neutral integrator, FlyWei is tied to no one product family — and in a mixed-flow building that matters, because no single manufacturer makes the best machine for every move on the list.

Matching the machine class to the move

Goods-in to reserve racking: autonomous forklifts in the counterbalance or reach-truck class. Reach trucks are usually preferred where aisle width is already committed.
Let-down and replenishment: autonomous stackers and narrow-aisle pallet trucks on scheduled and demand-triggered loops — typically the highest-value single flow to automate on a peak-constrained site.
Totes, cages and roll containers: lifting robots and jacking mobile robots, which move sub-pallet loads far more efficiently than a forklift will, and fit where a truck should not go.
Despatch marshalling: pallet-truck class robots running finished pallets from packing to loading lanes on a repeating circuit.

Integration without replacement

The robots do not talk to the warehouse system directly. A fleet management layer sits between them, taking work from the existing WMS or ERP over a documented interface — or, where none exists, a scheduled export handled by middleware — and translating each task into missions, routes and traffic rules. Completions and load confirmations are written back, so the existing system stays the record of stock, and the same layer handles PLC handshakes for doors and conveyors.

Because that layer speaks open standards such as VDA 5050, machines from different manufacturers can run under one traffic controller rather than needing a separate integration each — the difference between an operator who can add the right machine next year and one locked into whatever their original supplier now sells. Vehicle-side, safety and navigation behaviour comes from certified robot controllers, not bolt-on software.

How a deployment runs

Free site survey. Engineers walk the building, measure aisles, floor condition and level changes, and record the real traffic pattern — including where pedestrians and manual trucks will keep working.
Flow analysis and simulation. Movement data models which flows to automate first and how many vehicles each needs, tested against peak multiples rather than average days — so fleet size is decided before anything is ordered.
Interface design. The WMS or ERP interface is scoped: which fields carry the task, what a completion looks like, how exceptions reach supervisors, and what happens if the link drops.
Phased rollout. One flow goes live first — usually replenishment or despatch trunking — with a small fleet in a defined zone, alongside manual trucks. Teams are trained on exception handling, not just the robots.
Live operations and scale. Vehicles and flows are added against measured behaviour, typically ahead of the next peak, on the same control layer.

Typical results

Outcomes are best described qualitatively: they depend on the building, the order profile and how much of the flow is automated. What operations of this type generally report:

Replenishment becomes a scheduled, predictable flow rather than a queue behind shared trucks, so pick faces run empty less often.
Long, repetitive travel legs move to robots, and travel time on them generally falls because routing is centrally optimised rather than left to judgement.
Night-shift and shoulder-hour running becomes feasible without a fully crewed shift — often where peak capacity is actually recovered.
Operators are typically redeployed to higher-value tasks — quality, exceptions, goods-in checking — rather than displaced.
Damage and near-misses in automated lanes generally reduce, because robot travel is consistent, speed-limited and logged.

What to consider for your site

Which flow is genuinely the constraint? If picking is not the bottleneck, automating it will not help; movement data answers this quickly.
Are locations uniquely identified and stock balances trusted? Automating on top of unreliable master data multiplies errors rather than removing them.
Can your existing system expose open work and accept confirmations? A documented interface is ideal; a scheduled export is often enough to start.
What are your real aisle widths, floor tolerances and door constraints? These decide the machine class before anything else does.
Buy or lease? Leasing suits operators who want the fleet to flex with contract length; purchase suits stable, long-horizon flows.
Who owns the control layer? Ask any supplier whether you can add another manufacturer's vehicle later without a new integration.

Talk to an independent integrator

Every fulfilment building is different, and the honest answer to "would this work here?" starts with a walk round the site. FlyWei is an independent UK systems integrator: we specify autonomous forklifts and mobile robots from multiple manufacturers and integrate them with the systems you already run, with no incentive to favour one product family. If peak is exposing a material-flow constraint, book a free site survey and we will tell you which flows are worth automating — and which are not.

Peak-Season Surge Capacity Without Replacing Your WMS for a UK E-Commerce Fulfilment Operation