How Many AMRs Do You Need? A Practical AMR Fleet Sizing Guide

When planning an AMR deployment, one of the most common questions is not simply, “Can the robot move the load?” It is: How many AMRs does our factory actually need?

That number directly affects project budget, charging infrastructure, fleet-management strategy, and future expansion. AMR fleet size cannot be estimated reliably from floor area alone, nor can it be based only on total mission volume or a robot’s maximum speed. The factors that really determine fleet size are peak mission demand, actual cycle time, and the effective capacity lost to charging, empty travel, waiting, and traffic.

If you are preparing an AMR project budget, comparing supplier proposals, or checking capacity before a pilot, you can use the method below to calculate a first-pass fleet size, then validate it against the real layout and mission timing.

1. What Is AMR Fleet Sizing? Define the Goal First

AMR fleet sizing is not about determining how many robots can physically fit inside a factory. It is also not about finding the theoretical minimum number of robots that can keep moving.

The real objective is to determine how many AMRs are needed to reliably complete transport missions during peak demand at the required service level, while still maintaining reasonable capacity for charging, maintenance, and recovery from unexpected events.

The core logic can be reduced to two steps:

1. Calculate how many robot-minutes per hour are required during the peak period.
2. Calculate how many effective mission minutes one AMR can actually provide during the same hour.

The word “Initial” matters. This calculation is designed for first-pass capacity planning and supplier proposal comparison. It is not the final purchasing decision.

2. Prepare These 6 Inputs Before Calculating AMR Fleet Size

You Do Not Need a Complex Simulation Model at the Budgeting Stage

If you can establish credible assumptions for these six inputs, you can move from “guessing how many robots we need” to a first-pass fleet size with a clear formula, explicit assumptions, and known limits. As the project progresses, replace those assumptions with actual layout data, mission timing, and pilot results.

3. Step One: Use Peak Mission Demand, Not Just the Daily Average

AMR fleet size is driven first by how many missions must be completed during the busiest period. Daily averages are useful for understanding long-term utilization, but they can hide replenishment peaks, line changeovers, concentrated production output, and shipping waves.

For example, suppose a factory operates 16 hours per day and completes 320 transport missions. The average demand is only:

But if concentrated replenishment between 14:00 and 15:00 raises demand to 30 missions/hour, a fleet sized around 20 missions/hour may perform normally during quiet periods while continuously building a backlog during the peak.

Why Should Critical Material Replenishment Also Be Checked at 10–15 Minute Peaks?

Because 30 missions in one hour do not necessarily arrive evenly. Ten missions may be released in the first 10 minutes, creating a queue immediately.

The purpose of checking short-term peaks is to evaluate mission waiting time, queue length, and line-starvation risk. It does not mean that 10 missions in 10 minutes should automatically be multiplied by six and treated as a sustained requirement of 60 missions/hour for purchasing purposes.

Whether an additional robot is justified depends on how long missions are allowed to wait, how long the surge lasts, and whether the process has enough buffer capacity to absorb the peak.

4. Step Two: Calculate the Full Mission Cycle Time, Not Just Driving Time

For fleet sizing, cycle time means the total time an AMR is occupied from the moment it starts executing a mission until it becomes available to accept the next one.

A complete AMR mission may include:

• empty travel to the pickup point;
• queuing, positioning, and docking;
• manual loading, lifting, roller transfer, towing connection, or PLC / conveyor handshakes;
• loaded travel, including turns, intersection slowdowns, and temporary obstacle avoidance;
• waiting for automatic doors, elevators, AS/RS, or other shared equipment;
• drop-off docking, unloading, clearing the station, and returning to an assignable state.

Why not use the maximum speed listed on the AMR specification sheet? Because maximum speed is a vehicle capability limit, not the average speed achieved on a real factory floor. Two 100 m routes can have very different mission times if they contain different numbers of turns, intersections, pedestrians, automatic doors, or station-control requirements.

During early project planning, use a conservative estimate of expected site-average speed. Once the pilot begins, replace that assumption with measured cycle-time data.

As supporting evidence, a 2026 manufacturing study by Ekim used total robot travel time, plant layout, station operation times, and safety factors to establish a cycle-time-based lower bound for fleet size, then validated operational feasibility using a time-indexed scheduling model. Read the study in Machines.

5. AMR Fleet Sizing Formula: Calculate the First-Pass Fleet Size

If mission structures are reasonably similar, a first-pass estimate can be made with the following simplified formula:

How Should You Choose E?

Values such as 85% or 90% should not be treated as universal industry standards. If no historical data is available, use E as an explicit planning assumption and run a sensitivity analysis—for example at 0.80, 0.85, and 0.90.

Once the project enters the pilot phase, calibrate E using actual charging time, maintenance time, and mission availability.

Avoid Double-Counting Capacity Losses

If traffic delay, docking time, automatic-door waiting, or elevator waiting has already been included in cycle time, do not deduct the same loss again through E. Otherwise, the same constraint is counted twice, and the fleet may be unnecessarily oversized.

6. Worked Example: How Many AMRs Are Needed for 30 Missions per Hour?

The example below walks through the calculation step by step. It is intended to demonstrate the method and does not represent the guaranteed performance of any specific AMR product.

6.1 Calculate Travel Time

Total average travel distance:

At an expected site-average speed of 1.0 m/s, travel time is approximately 240 seconds, or 4 minutes.

6.2 Add Pickup, Drop-off, and Waiting

Pickup 30 s + drop-off 30 s = 1 minute.

Average docking and traffic waiting = 1 minute.

6.3 Convert Cycle Time Into AMR Quantity

Assume an initial planning value of E = 85%:

So, four AMRs are the first-pass fleet size for this example. That is still not the final purchasing decision because the real layout may contain bottlenecks such as intersections, shared stations, or charging infrastructure.

6.4 Why Does Using Average Demand Underestimate the Fleet by One Robot?

If the daily average of 20 missions/hour is used incorrectly:

For the same project, simply replacing peak demand with average demand changes the result from four robots to three. Three robots may be adequate off-peak, but that does not prove they can prevent mission backlog during the peak.

6.5 How Much Capacity Margin Do Four AMRs Have?

Four robots at E = 85% can provide:

At 30 missions/hour, the maximum average cycle time that fits within that capacity is:

This 6.8-minute boundary is often more useful than the statement “we need four robots.” If pilot data shows that actual cycle time exceeds approximately 6.8 minutes, four AMRs may have crossed the capacity limit.

At that point, the project either needs a fifth robot or needs to reduce cycle time through route optimization, better docking, improved task chaining, or revised traffic rules.

7. Multiple Mission Types: Use Robot-Minutes per Hour

When different routes have very different mission times, simply counting “missions per hour” can be misleading. A better method is to convert every mission type into the number of robot-minutes it consumes per hour, then add them together.

Task B has only one-third as many missions as Task A, but both consume 60 robot-minutes/hour. For multi-route projects, build a From-To Matrix and calculate:

Then divide the total workload by 60 × E to obtain the first-pass fleet size.

8. Two Factors That Are Commonly Missed: Empty Travel and Charging

8.1 Empty Travel: Where Does the AMR Go After Drop-off?

If the one-way route from the warehouse to the production line is 150 m, and the next pickup is still at the warehouse, the AMR must travel approximately another 150 m empty after unloading. The true mission loop is close to 300 m, not 150 m.

A good fleet-management system can reduce deadheading by chaining missions—for example, assigning the AMR another pickup near point B immediately after completing A → B. But that benefit depends on mission density, direction, and timing.

Early-stage fleet sizing should not assume that empty return travel is zero unless historical data or simulation supports that assumption.

8.2 Charging: Do Not Treat “8 Hours of Battery Life” as “8 Hours of Productive Work”

The key question is not the battery’s rated runtime. It is how many robots remain mission-capable during the production peak.

Opportunity charging, scheduled charging, battery swapping, state-of-charge trigger logic, charger location, and charging queues can all change the effective capacity of each AMR during peak operations.

Ekim (2026) compared scenarios with replaceable batteries and mandatory charging interruptions, showing that charging assumptions can significantly change the minimum feasible fleet size and operational feasibility. Selmair et al. (2022) used simulation to compare AMR charging and parking strategies using metrics such as delayed orders, average state of charge, traffic density, fleet availability, and charger utilization. Read the Journal of Power Sources study.

Fleet Size and Charger Quantity Are Separate Calculations

After calculating fleet size, charger capacity should be validated separately.

Do not apply a fixed rule such as “one charger for every X AMRs” without checking charging minutes, available charging windows, charger queues, and the non-productive travel time to and from charging stations.

9. Why More AMRs Do Not Always Produce More Throughput

As a fleet grows, the system bottleneck may shift from robot availability to shared-resource capacity. Main aisles, intersections, narrow doors, elevators, pickup/drop-off stations, conveyors, AS/RS interfaces, and chargers can all cause additional AMRs to spend more time waiting.

• More robots → more intersection conflicts → lower average travel speed.
• Fixed station capacity → longer arrival queues → longer cycle time.
• Insufficient chargers → longer charging queues → fewer mission-capable robots.
• Poor dispatching logic → more deadheading and resource competition → lower marginal benefit from each additional robot.

Kim, Kang, and Jung (2026) frame robot fleet sizing as an operational planning problem that changes with order volume, layout, and operating strategy. Their results also show that deploying the full available fleet is not always optimal, and that fleet size should be evaluated against actual operating conditions. Read the study in Computers & Industrial Engineering.

If you want to understand how multi-robot task allocation, path planning, and fleet control can influence congestion and throughput, see Fdata’s guide to autonomous mobile robot control systems and multi-robot management.

So the best optimization is not always “buy another AMR.” In some cases, changing the route, introducing one-way traffic, adding parallel pickup/drop-off stations, adjusting mission-release timing, or changing traffic-priority rules can be more effective than simply increasing fleet size.

10. When Is the Formula Not Enough? Use Simulation or a Pilot

The formula is useful for answering, “Approximately how much robot capacity do we need?” But it does not reproduce every conflict, shared-resource constraint, or random event in a real factory.

Consider discrete-event simulation, digital validation, or an on-site pilot when:

• multiple AMRs frequently share main aisles, narrow aisles, or critical intersections;
• there are many pickup/drop-off points and cycle times vary significantly by route;
• missions depend on automatic doors, elevators, conveyors, AS/RS, or other shared equipment;
• missions are released in concentrated bursts and peak demand is highly variable;
• charger capacity is limited, and charging queues are a realistic risk;
• the loss of one AMR could cause production-line starvation;
• the project is expected to scale quickly from a small pilot to a medium or large AMR fleet.

The number of robots alone is not a reliable threshold for deciding whether simulation is needed. Three AMRs that all depend on the same elevator may require more modeling than a larger fleet operating on largely independent routes.

From a broader planning and control perspective, Fragapane et al. published an invited review in the European Journal of Operational Research covering AMR routing, scheduling, dispatching, and control decisions, and how these decisions affect system performance. Read the EJOR review on ScienceDirect.

If your project is primarily warehouse-based, Fdata’s step-by-step guide to deploying AMRs in a warehouse covers site assessment, pilot deployment, KPI validation, optimization, and phased scaling.

Do Not Measure Only “Missions Completed per Hour”

11. Should You Buy a Spare AMR? Run an N-1 Check

There is no universal rule that says, “If five AMRs are running, always buy six.” A more useful engineering test is to remove one robot and evaluate what happens.

Using the previous four-AMR example:

Peak workload remains:

So under N-1 conditions, the system has a capacity deficit of 27 robot-minutes per hour, equivalent to approximately 4.5 six-minute missions.

Whether the fifth robot is justified depends on:

• how long peak demand lasts;
• the maximum allowable delay for critical missions;
• whether forklifts or another manual fallback process are available;
• maintenance response time and spare-parts strategy;
• the cost of production-line starvation.

Evaluate at Least Three Fleet States

1. Normal peak: N robots available.
2. Single-robot outage: N-1 robots available.
3. Future growth: higher mission volume, longer routes, or more shared resources.

The final purchasing quantity should explain what risk each additional robot is actually solving, rather than relying on one formula result alone.

12. Give Suppliers These 8 Inputs Before Asking Them to Calculate Fleet Size

1. Peak Missions per Hour: how many missions must be completed in the busiest hour; for critical replenishment, also provide 10–15 minute peaks.
2. From-To Points: all pickup and drop-off points, mission directions, and mission types.
3. Real Route Distance: measure actual drivable paths rather than straight-line distances on the drawing.
4. Payload & Carrier: weight, dimensions, center of gravity, and whether the load uses pallets, racks, carts, or other carriers.
5. Pickup / Drop-off Time: actual time required for manual handling, lifting, roller transfer, towing, or custom mechanisms.
6. Facility Layout: CAD, DWG, or PDF drawings showing aisles, doors, intersections, shared equipment, and charging areas.
7. Operating Schedule: 8-hour, 16-hour, or 24/7 operation; available charging windows; whether battery swapping is permitted.
8. Peak Pattern & Service Level: when peaks occur and how long critical missions weg wait.

The more complete these inputs are, the easier it is for a supplier to provide an explainable fleet-sizing result instead of simply stating, “We recommend five robots.”

For Payload & Carrier, do not provide only the cargo weight. The top module, rack, mounting structure, center of gravity, and dynamic operating conditions can all affect final platform selection. For a deeper calculation, see Fdata’s mobile robot chassis payload and selection guide.

13. A Simple, Practical AMR Fleet Sizing Workflow

1. Identify Peak Missions per Hour and check 10–15 minute short-term peaks.
2. Build a From-To Matrix and separate different routes and mission types.
3. Measure real routes, empty travel, and pickup/drop-off time.
4. Calculate the full cycle time for each mission type.
5. Convert mission demand into robot-minutes/hour.
6. Apply charging, maintenance, and reasonable capacity-margin assumptions to calculate the first-pass fleet size.
7. Round up and calculate how much capacity margin remains before another robot is required.
8. Use simulation or a pilot to validate traffic, queues, charging, and shared stations.
9. Validate N-1 and future-growth scenarios.
10. Finalize the AMR purchase quantity, then validate charger quantity and charger locations separately.

Conclusion: AMR Fleet Sizing Is About Stable Peak Performance, Not Buying the Fewest Robots

The core calculation is straightforward:

But that number is only the starting point. A real AMR project must also validate empty travel, charging, intersections, shared aisles, workstation queues, short mission surges, single-robot failures, and future growth.

A fleet that is “just large enough” on paper but immediately builds a backlog during peak demand, charging, or maintenance is not truly deployable.

If your calculation gives 3.4 robots, do not only ask, “Can we make three work?” A more useful question is: What is the fourth robot actually solving? Peak mission demand? Charging time? Traffic delay? Capacity margin? Or system reliability?

Once that reason is quantified, AMR fleet sizing becomes an engineering decision rather than a guess.

Planning an AMR Project?

Start by organizing your From-To points, real route distances, peak mission demand, payload, pickup/drop-off method, and operating schedule. Once these inputs are available, you can contact Fdata to discuss your AMR project.

Note: This article is intended for early-stage AMR capacity estimation and solution evaluation. Factories, warehouses, AMR products, fleet-management systems, and charging strategies vary significantly. Final purchasing quantities should be validated using actual site data, simulation, and/or an on-site pilot.

FAQs

How many AMRs does a typical factory need?