AGV vs AMR: real differences between guided vehicles and autonomous mobile robots, and which one to choose

Two vehicles can run the same aisle, on the same job, and end up costing very different amounts over ten years. The reason is how they find their way. One reads the building and the other follows a line laid into the floor. That single difference decides what gets cut into the concrete, how many days the plant stands still to install it, and what happens the next time the flow changes. Pick wrong in one direction and a company pays for guidance infrastructure it never needed. Pick wrong in the other and it runs a fixed path through a warehouse that gets reshuffled every season.

How an AGV finds its way, and how an AMR does

An AGV (automated guided vehicle) follows a predetermined path defined by guidance infrastructure. The technologies came one after another: a wire buried in the floor, magnetic tape or spots, floor QR codes, and laser guidance that triangulates on wall-mounted reflectors, typical of the so-called LGVs. The vehicle knows where it is relative to the path, not relative to the building. When something blocks the route it stops safely and waits for someone to clear it. Vis's survey in the European Journal of Operational Research (2006) captures the logic well: designing an AGV system starts with designing its path network, and everything else follows from that.

An AMR (autonomous mobile robot) builds and updates a map of its surroundings with SLAM (simultaneous localization and mapping), usually with lidar or cameras, and localizes on the building's natural features such as walls, racks and columns. No route is wired anywhere. The robot plans one for each mission and recalculates it on the move, slowing down and steering around people and obstacles. In their 2021 review in the same journal, Fragapane, de Koster, Sgarbossa and Strandhagen identify this decision-making autonomy as what sets AMRs apart from AGVs, and as the source of a new set of fleet planning and control problems.

The acronyms don't settle the question anymore, either. Plenty of recent vehicles navigate on natural features but stay on constrained virtual paths, and what one supplier sells as an AMR another calls an AGV. What holds up in a comparison is the set of questions in the table further down: infrastructure, how often the layout changes, mixed traffic, takt time. And how honestly a given machine answers them.

The differences that move the project

The last row matters most on synchronized lines. Re-planning is an advantage in variable flows, yet it adds variability to cycle times. When a line is fed at a fixed takt, the absolute predictability of an AGV can be worth more than all the flexibility of an autonomous robot.

Safety: same standard, different behavior

On the regulatory side, industrial AGVs and AMRs both fall under ISO 3691-4:2023, the standard for driverless industrial trucks and their systems: person detection, safety laser scanner protective fields, speed control, protective stops and the preparation of operating zones. When the topic is CE marking, the version that counts is the harmonized EN ISO 3691-4:2023, cited in the Official Journal by Implementing Decision (EU) 2024/1329, which gives presumption of conformity with the Machinery Directive 2006/42/EC. Risk assessment under ISO 12100:2010 sits upstream. From January 20, 2027 the Directive gives way to Regulation (EU) 2023/1230, which for the first time deals explicitly with autonomous mobile machinery. In the United States the references are ANSI/ITSDF B56.5-2024 for guided vehicles and the ANSI/A3 R15.08 series for industrial mobile robots.

So safety doesn't depend on the acronym. It depends on vehicle compliance, on the quality of the integration and on the risk assessment of the real site, with its intersections, doors, shared areas and overhanging loads. The behavioral difference still matters for throughput in busy areas: where an AGV stops and waits, an AMR slows down and drives around. How much that changes injury figures is covered in our piece on AMRs and warehouse injury reduction.

When the AGV is still the right call

The AGV earns its place where flexibility isn't needed. A handful of A-to-B routes repeated thousands of times, in an area with little mixed traffic, is its natural ground: the protective stop becomes a rare event instead of a bottleneck, and the simple vehicle pays off. The same goes for heavy or oversized loads such as reels and coils, where heavy-duty AGVs and LGVs have decades of installations behind them, and for feeding lines at a fixed takt, where cycle-time predictability is the first requirement. A new building designed around flows that will stay put for years can also amortize guidance infrastructure without strain.

When the autonomous mobile robot wins

The picture flips as soon as the flow changes often. Mission mixes, seasonal peaks, e-commerce orders: re-planning is the AMR's structural advantage here. In existing buildings, laying wire or reflector fields is often impractical, and a robot that navigates on natural features goes in without building work. In busy areas, where people and manned trucks cross the routes all day, stopping at every interference would wreck throughput. And a company that wants to grow in steps can start with a few vehicles on one flow and add robots, not infrastructure. For pallet handling in a warehouse that is already running, the starting path is described in our article on automating pallet moves with AMRs without stopping operations.

Where the cost sits

The useful economic comparison is about where the spending concentrates rather than which vehicle is cheaper. With an AGV the cost shifts to infrastructure and project engineering: route studies, installation work, commissioning. The hidden cost is the cost of change, because every layout change reopens the job. With an AMR the cost concentrates in on-board intelligence (sensors, computing, navigation software) and in fleet software, often licensed or sold by subscription, while infrastructure is minimal and change is cheap. Payback profiles end up very different depending on the flows, and the method for working them out, with recurring costs and stated assumptions, is in our analysis of AMR ROI, TCO and payback.

One simple question helps settle it. If the routes two years from now will be the same as today, an AGV is defensible. If nobody in the company can answer that with confidence, the AMR's flexibility already works as room to maneuver.

Decision table

The last row needs a caveat. The VDA 5050 specification, which reached version 3.0.0 in March 2026, standardizes communication between fleet control and vehicles from different makers, but it states that it covers neither safety nor traffic management logic. Who signs off on the conformity of a mixed fleet has to be settled in the contract. To put numbers behind the table, the feasibility study method is in our article on evaluating an autonomous forklift and AMR project.

What to send us for an assessment

To tell you whether your flow calls for an AGV, an AMR or a mixed fleet, a few figures are enough, as long as they are measured rather than estimated:

• missions per hour, with the worst hour kept apart from the average;
• origins and destinations, with the distances of the main routes;
• load type: pallets, bins, roll cages or totes, with weights and dimensions;
• a dimensioned floor plan with aisle widths, doors, ramps and intersections;
• mixed traffic: people and manned trucks sharing the routes;
• how often the layout has changed over the last two or three years;
• systems to integrate: WMS, ERP or MES, and any lines to be fed.

You can send them by email to info@dipitech.com or on WhatsApp, with the floor plan and photos of the critical spots. The technologies DIPITECH represents in Italy are described on the pages about robotic forklifts and AMRs for pallet handling and the goods-to-person automated warehouse, part of our industrial automation and material handling robotics offering.