AGVs and AMRs: taking the driving out of warehouse travel

Two families with different assumptions

Guided vehicles follow a prepared path, using magnetic tape, embedded wire, reflectors or markers, and they are predictable, repeatable and awkward to reroute because the infrastructure defines the route. Autonomous mobile units build and consult a map of the building, plan their own path and drive around an obstruction, which makes them quicker to deploy and to redeploy. The trade is between the certainty of a fixed path and the flexibility of one that is recalculated, and busy mixed-traffic buildings usually favour the latter.

Fleet size is a traffic question

Adding machines raises capacity only until they begin to obstruct one another at junctions, doors and charging points, after which each extra unit adds congestion instead of movement. Fleet sizing therefore has to consider path layout, passing places, charging strategy and the proportion of journeys made empty. Cutting empty running, by pairing a delivery with a collection the way an operator would, is often worth more than buying another vehicle.

The floor has to be legible to a machine

Robots are defeated by the things human operators simply step around: a pallet left protruding into an aisle, a cage parked across a route, a spill, a damaged floor joint, a door that has been propped open. Deployment therefore comes with a housekeeping standard and a rule that routes stay clear, enforced rather than requested. Lighting, floor condition and the position of fixed reference features all affect navigation reliability, and they deserve a survey before the first machine arrives.

People and machines in the same aisle

Mixed operation is the normal case, and it works through sensing, speed limits, defined behaviour when a person is detected and clear expectations on both sides about who yields. Machinery safety requirements for such equipment are set by the authority responsible in the territory, and the supplier's assessment should be reviewed against how the site actually works rather than against an idealised layout. Training matters in both directions: operators need to know how the machines behave, and the machines' rules need to reflect what operators will really do.

Where the payback actually comes from

The strongest cases are long horizontal moves that recur constantly, replenishment runs between reserve and pick faces, transport between distant zones, and moves that can be shifted into quiet hours when the building is otherwise empty. The weakest cases involve short, dense, highly variable movement where a person covers several tasks in the time a machine completes one. Measuring the candidate journeys before choosing equipment prevents buying a fleet for work that never justified a vehicle in the first place.

Frequently asked questions

Do automated vehicles need a purpose-built building?

No, and that is much of their appeal: they can be introduced into an existing site incrementally. What they do need is a floor in reasonable condition, routes that stay clear, and a housekeeping standard the operation is prepared to enforce every shift.

How is charging handled without losing availability?

Usually by short opportunity charges during natural pauses rather than long charges that remove a unit for hours. Charging points then become part of the traffic plan, since a queue at the charger is as limiting as a queue at a door.