Automated Guided Vehicles (AGVs) Explained

Long before anyone was talking about robots that map a building on their own and reroute around a pallet dropped in the aisle, warehouses were already moving loads automatically. The technology that did it was the automated guided vehicle, and it did so by following a physical path laid into the floor: a strip of magnetic tape, a wire buried in a slot, or a grid of reflective targets on the walls. It was not clever in the modern sense. It did not need to be. It needed to move the same load along the same route, thousands of times a day, without a driver and without complaint, and at that narrow job it was, and remains, extremely good. This guide sits inside the broader complete guide to warehouse automation, and it goes deep on the AGV specifically: what it is, how it finds its way, the types you will actually see, and the honest trade against its more fashionable successor.

The message up front: an AGV is not an inferior AMR. It is a different tool for a different job. When your routes are fixed, your volumes are high, and your loads are heavy, the AGV's rigid, path-bound simplicity is a feature, not a limitation. The mistake is not choosing an AGV. The mistake is choosing an AGV for work that keeps changing, or choosing an AMR for work that never will.

1. What an AGV is

An automated guided vehicle is a driverless industrial vehicle that transports material along a predefined path between fixed points in a facility. The two words that matter most in that definition are "guided" and "predefined." An AGV does not decide where to go in any meaningful sense. It follows a route that a human laid down in the physical world, whether as tape stuck to the floor, a wire cut into it, or a set of reflector targets it triangulates against. Take away that infrastructure and the vehicle is lost. That dependence is the single defining characteristic of the whole class, and everything else about how AGVs behave, where they excel, and where they frustrate flows directly from it.

Mechanically an AGV is straightforward. It has a drive system, a battery, a load-handling mechanism appropriate to its job (forks, a flat deck, a hitch, a roller conveyor top), a guidance sensor that reads the path, and a safety system, almost always including obstacle-detection scanners that bring the vehicle to a controlled stop if something enters its path. It communicates with a central controller, sometimes called a fleet manager or traffic manager, that tells it which route to run, coordinates it with other vehicles at intersections, and manages battery charging. The intelligence in an AGV system lives more in that central controller and in the fixed infrastructure than in the vehicle itself. The vehicle is deliberately simple, which is part of why AGVs have historically been so reliable and so long-lived on the plant floor.

It is worth placing the AGV in the wider family early. It is one member of the broader field covered in warehouse robotics explained, sitting alongside its more autonomous cousin the autonomous mobile robot and the fixed, non-vehicle option of conveyor systems. The AGV occupies the middle ground: more flexible than a conveyor because it can be rerouted (by moving the tape or reprogramming the path), far less flexible than an AMR because it still needs a physical path to follow at all.

2. How AGVs are guided

Guidance is the heart of the AGV, so it is worth understanding the main methods properly. Every one of them solves the same problem, keeping the vehicle on a known route, but they solve it with very different infrastructure, cost and flexibility profiles. Before the table, here is the essential picture: an AGV senses a fixed path element, steers to keep itself centred on it, moves between fixed stations along that path, and stops when its safety scanner detects an obstacle blocking the way ahead.

The four guidance methods below are the ones you will encounter in practice. Each reads a different kind of path reference, and the choice drives cost, flexibility and how disruptive installation is.

Notice the pattern running through the table. Magnetic tape and buried wire are true fixed-path methods: the route lives in the floor. Laser and vision guidance loosen that a little, moving the reference off the floor and letting you change routes in software, which is why people sometimes call laser-guided vehicles the bridge toward autonomous mobile robots. But even the laser and vision methods still depend on installed infrastructure, targets on the walls or codes on the floor, and a route that a human defined. That distinction, path defined by fixed infrastructure versus path decided by the vehicle in real time, is exactly the line between an AGV and an AMR, which we come to later.

3. Types of AGV

Under the single label "AGV" sits a range of vehicle types, each shaped by the load it carries and the job it does. Knowing the types keeps a specification conversation grounded, because a vendor pitching "an AGV solution" is really pitching one of these forms, and the right one depends entirely on what you are moving and how.

• Tugger (tow) AGVs: a driverless tractor that pulls a train of carts or trailers behind it. Tuggers are the classic answer to moving many loads over a long, fixed route, for example feeding a production line from a central store. One vehicle hauls several carts, so the throughput per vehicle is high, which makes tuggers efficient for repetitive line-side replenishment across a big plant.
• Forklift AGVs: automated versions of the counterbalance or reach truck, able to pick pallets from the floor or from racking and place them at height. These handle the same lifting tasks a manned forklift does, but on fixed routes and repetitive moves, and they are the natural fit where loads must be raised and lowered rather than just carried at floor level.
• Unit-load AGVs: vehicles with a deck, rollers or a lift table that carry a single load, typically a pallet or a large container, directly on the vehicle body. They shuttle full loads point to point, often between staging areas, and are common where the movement is a straightforward carry between two fixed stations.
• Cart and light-load AGVs: smaller vehicles that carry totes, bins or light assemblies, frequently used in assembly and kitting flows and in goods-to-person picking. This is the category that overlaps most with the newer generation of robots, and the boundary between a small AGV and an AMR is genuinely blurry here.

The practical point when specifying: pick the type by the load and the movement first, then choose the guidance method to suit the environment. A heavy pallet between two fixed docks all day is a unit-load or forklift AGV on a robust guidance method. Line-side replenishment across a large plant is a tugger. Light totes to a pick station is a cart AGV, and that is exactly where you should also be asking whether an AMR would serve better.

4. Where AGVs still win

It would be easy, reading the industry press, to conclude that AGVs are legacy technology overtaken by autonomous robots. That conclusion is wrong, and expensively so. There are whole categories of work where the AGV remains the better economic and operational choice, and they share a common shape: the route does not change, the volume is high, and the load is demanding.

• Fixed, unchanging routes: when the same load moves along the same path day after day, the AGV's rigid adherence to that path is exactly what you want. There is nothing to gain from a robot that could dynamically reroute, because the route never needs rerouting. The AGV does the fixed job with less complexity and less to go wrong.
• High, steady volume: on high-throughput lanes, a fleet of AGVs running a tight, well-managed loop delivers predictable, high sustained throughput. The determinism is a genuine advantage: you know exactly where every vehicle will be and when, which makes the whole flow easy to plan and audit.
• Heavy and oversized loads: AGVs scale up to move very heavy loads, pallets, coils, engines, whole assemblies, that are beyond the comfortable range of most light autonomous robots. Where the load is measured in tonnes, the robust industrial AGV is the mature, proven answer.
• Harsh or controlled environments: cold stores, hazardous areas and clean processes often suit a purpose-built AGV whose behaviour is fully deterministic and whose infrastructure is fixed and inspectable, rather than a vehicle making its own real-time decisions.

The insight worth keeping: predictability is a feature. In a high-volume operation the fact that an AGV always does the same thing, along the same path, at the same speed, is precisely what makes the flow plannable and reliable. The flexibility of an AMR is only valuable when you actually need to change the flow. If you never need to, that flexibility is complexity you are paying for and not using. Match the tool to the stability of the work, not to the fashion of the market. For the full landscape of options, keep the warehouse automation pillar open alongside this decision.

5. AGV versus AMR

This is the comparison everyone actually wants, so let me put it plainly. The defining difference is navigation. An AGV follows a fixed path built into the environment, tape, wire, targets or codes, and cannot leave it. An autonomous mobile robot carries a map of the space, senses its surroundings in real time, and plans its own route dynamically, so it can navigate freely, avoid obstacles by driving around them, and adapt when the layout or the task changes. The AGV stops and waits when its path is blocked; the AMR, in principle, finds another way around.

That single difference cascades into everything else. AGVs need physical guidance infrastructure and are disruptive to reroute, but they are mechanically simpler, extremely deterministic, cheaper per vehicle at the heavy end, and mature. AMRs need no floor infrastructure and are reconfigured in software, which makes them ideal for changing layouts and mixed tasks, but they cost more per unit, are more complex, and their real-time decision-making, though impressive, introduces variability that a tightly-timed high-volume line may not want.

The honest decision rule I give clients is about the stability of the work, not the sophistication of the machine. If your routes are fixed and your volumes are high and steady, the AGV is very often the better and cheaper answer, and the AMR's flexibility is a premium you will not use. If your layout changes often, your tasks are mixed, or you are scaling into a space you cannot afford to cut and re-cut floors in, the AMR's ability to be redeployed with a software change earns its higher price many times over. Neither is universally superior. The question is whether the work will change, and how often.

6. The honest limits

An expert guide has to be as clear about where AGVs frustrate as about where they win, because the failure stories are real and predictable, and almost all of them trace back to the same root: the fixed path that is the AGV's great strength is also its great constraint.

• Inflexibility: changing an AGV route means changing physical infrastructure, relaying tape, re-cutting a floor for wire, moving reflector targets, or re-marking a code grid, and reprogramming. In a facility whose layout keeps evolving, that rigidity becomes a recurring cost and a drag on change. An AGV rewards a stable operation and punishes a volatile one.
• Infrastructure dependence and disruption: the guidance path has to be installed before the vehicle is useful, and for buried-wire systems that installation means cutting the working floor, which is disruptive and costly. The infrastructure also has to be maintained; damaged tape or a blocked reflector can stop a vehicle dead.
• Blocking behaviour: a classic AGV meets an obstacle on its path and stops, then waits, because it cannot leave the path to go around. On a busy floor shared with people and manual trucks, one blocked vehicle can stall a whole lane until the obstruction is cleared, whereas a well-implemented AMR would route around it.
• Upfront cost and lead time: between the vehicles, the guidance infrastructure, the central controller and the integration, an AGV deployment is a capital project with meaningful lead time, not a plug-in purchase. That is fine for a stable high-volume flow that will run for years, and poor value for a flow you are still figuring out.

The caution: do not buy an AGV for a flow that is not yet settled. The technology assumes the route is fixed and worth committing infrastructure to. If you are still redesigning the process, the layout, or the volumes month to month, an AGV locks you into decisions you have not finished making. Prove the flow is stable first, then automate it with the tool whose whole premise is stability.

7. AGVs, the WMS and orchestration

An AGV moving loads around a floor is only useful if it moves the right loads, to the right places, at the right times, and that coordination comes from the software above the vehicles. This is the part of an AGV project that is easy to underestimate and expensive to get wrong, and it is the part I spend most of my own time on as an integration specialist. Hardware that works and software that does not orchestrate it is a very common and very frustrating outcome.

There are two software layers to keep distinct. The first is the AGV fleet controller (the traffic manager), which owns the vehicles: it assigns routes, sequences vehicles through intersections, manages charging, and handles the safety and deadlock logic of many vehicles sharing fixed paths. The second is the warehouse management system, which owns the work: it knows what needs to move, from where, to where, and in what priority, because it holds the inventory, the orders and the storage logic. The AGV fleet moves things; the WMS decides what should move and why.

The value, and the difficulty, is in the integration between those two layers. The WMS has to hand transport tasks down to the fleet controller, the fleet controller has to report completion and status back up, and both have to agree on where inventory ended up so the stock record stays accurate. When that integration is clean, an AGV fleet becomes a smooth, invisible extension of the warehouse process. When it is not, you get vehicles running the wrong moves, stock records drifting out of truth, and operators back to manually directing the automation they paid to remove. This is the same operational-technology to enterprise-IT bridge that recurs across every automation project, and it is where an AGV programme most often succeeds or fails, well after the vehicles themselves are working perfectly.

8. A practical adoption approach

If an AGV is on your agenda, the sequence matters more than the vendor. The approach I would advise any operation to follow keeps the risk low and the decision honest:

• Step 1: confirm the flow is stable. Identify the specific transport routes that are genuinely fixed, high-volume and unlikely to change. Those, and only those, are AGV candidates. If you cannot name a route that will still be running unchanged in three years, you are not ready to lay infrastructure for it.
• Step 2: characterise the load and the environment. Weight, dimensions, lift or carry, floor condition, sharing space with people and manual trucks. These determine both the vehicle type and the guidance method, and they surface the safety requirements early.
• Step 3: choose vehicle type, then guidance method. Match the type (tugger, forklift, unit-load, cart) to the movement, then pick the guidance method (tape, wire, laser, vision) to suit durability, disruption tolerance and how permanent the route is.
• Step 4: design the WMS integration first, not last. Decide how transport tasks flow from the WMS to the fleet controller and how completion and inventory feed back, before the vehicles arrive. The integration is the hard part; treat it as a first-class workstream.
• Step 5: pilot one route end to end. Prove a single lane, vehicle plus infrastructure plus fleet controller plus WMS integration plus safety, closes the loop and holds up under real volume before scaling.
• Step 6: scale deliberately along stable routes. Extend only to other routes that meet the same fixed-and-high-volume test. Resist the urge to automate flows that are still changing; those are AMR territory, or not yet ready for automation at all.

The through-line of that sequence is discipline about where the AGV belongs. Steps one and two cost almost nothing and prevent the most expensive mistake in the whole field, which is committing fixed infrastructure to a flow that turns out not to be fixed. Get the targeting right and the AGV is one of the most reliable, lowest-drama pieces of automation you can own. Get it wrong and you have poured infrastructure into a process that has already moved on.

9. References

For readers who want to go to the source material, the most useful anchor is the international safety standard for the class. ISO 3691-4 covers driverless industrial trucks and their systems, and it is the reference that governs how AGV safety, obstacle detection and system behaviour are specified and assessed. Any serious AGV specification or vendor conversation should be able to point to conformance with it. Beyond the standard, vendor technical documentation for the specific fleet controller and vehicle range, and your own facility's functional safety assessment, are the documents that turn the general principles in this guide into a concrete, compliant deployment. Treat the standard as the baseline and the vendor and safety documentation as the project-specific detail on top of it.

Final thoughts

The automated guided vehicle is often described as the older, simpler technology that autonomous mobile robots have superseded, and that framing does the AGV a disservice. It is not a worse robot. It is a different answer to a different question. When the work is fixed, high-volume and heavy, the AGV's rigid dependence on a physical path stops being a weakness and becomes the source of its reliability, its determinism and its cost advantage. The newer autonomous robots are genuinely better where the work changes, but a great deal of warehouse and plant transport does not change, and for that work the AGV remains, decades on, the right tool.

The judgement that matters is not which technology is more advanced. It is whether the flow you want to automate is stable enough to commit infrastructure to. Answer that honestly, match the vehicle type and guidance method to the load and the environment, design the WMS integration as carefully as the vehicles, and the AGV will quietly move your loads for years. Skip that judgement, automate a flow that has not settled, and you will learn the hard way why the fixed path is both the AGV's greatest strength and its sharpest limit.

Related reading: Warehouse automation: the complete guide, Autonomous mobile robots (AMRs) explained, Warehouse robotics explained, Conveyor systems in the warehouse, What is a WMS?.