- AMRs use onboard SLAM and sensor fusion to navigate dynamically — no floor modifications required beyond clear aisle widths (typically 1.5–2× robot width).
- A single AMR replaces 1.5–2.5 FTE forklift or tugger operators over two shifts, with payback typically 18–30 months.
- Fleet size is driven by throughput demand, route distance, and robot duty cycle — not by how many people you currently employ.
- Most modern AMRs connect to WMS/MES via REST API or ROS 2; integration adds $15K–$60K to project cost.
- Factories with mixed human/robot traffic, variable layouts, or high-mix material flow are better AMR candidates than rigid single-path facilities.
What AMRs Do in a Warehouse or Factory Floor
An autonomous mobile robot navigates using onboard sensors — typically a combination of 2D lidar, depth cameras, and wheel odometry — to build and maintain a map of its environment. Unlike AGVs (automated guided vehicles), AMRs do not follow magnetic tape or inductive wire guides embedded in the floor. They plan paths in real time, detect obstacles, and reroute dynamically.
In a warehouse or manufacturing context, the AMR's job is transport: picking up a load at point A and delivering it to point B without human intervention. The robot does not pick, sort, pack, or process the goods — that work happens at fixed workstations. The AMR handles the connecting material flow between those stations.
This matters because transport is often a significant fraction of operator time. In a study of manufacturing operations, operators in mixed-task roles (machine tending plus material handling) spend 30–50% of their time walking with carts, staging, or waiting for material. Automating that transit frees operators for higher-value tasks.
Task Types and Fit Assessment
AMRs are well-suited to a specific profile of material handling tasks. Before specifying a fleet, map your current material flow against these criteria.
Tasks Where AMRs Excel
- Point-to-point transport: Raw material from receiving dock to line-side buffer; WIP from machining cell to assembly; finished goods to staging or palletizer. Fixed origin/destination pairs with consistent load profiles.
- Kitting and replenishment: Delivering pre-picked kits to assembly stations on a takt-driven schedule. AMRs can be called on demand via workstation button or WMS trigger.
- Milk-run loops: Circulating through multiple pickup and drop-off points on a repeating sequence. Equivalent to a tugger train route but without a driver.
- Empty container return: Collecting empty totes, pallets, or racks from line-side and returning them to a staging area. Often done in background low-priority mode.
- Inter-building transport: Connecting adjacent buildings or crossing a yard (outdoor-capable AMR variants, IP65-rated).
Tasks That Are Poor Fits
- Loads that require positioning accuracy below ±15 mm: Most AMRs achieve ±10–25 mm docking accuracy. If your station requires a more precise load placement, a conveyor or fixed automation is better.
- Very high throughput requirements: If you need 50+ moves per hour from a single route, a conveyor is more cost-effective and reliable than a fleet of AMRs.
- Loads over 1,500 kg: Heavy-payload AMRs (tugger-style) reach 1,000–1,500 kg. Beyond that, fixed conveyance or fork AGVs are necessary.
- Highly cluttered, narrow, or unstructured environments: AMRs need clear aisles. Cluttered floors, racking that changes weekly, or aisle widths under 1.2 m make navigation unreliable.
- Loads requiring lift or racking access: Standard AMRs are floor-level movers. Forklift AMRs exist but are substantially more expensive and require different infrastructure.
Feasibility Checklist
| Factor | Go | Caution | Stop |
|---|---|---|---|
| Minimum aisle width | > 1.8 m | 1.3–1.8 m | < 1.2 m |
| Floor condition | Level, sealed concrete | Minor cracks, joints | Ramps > 5%, gravel, open grating |
| Load weight | < 600 kg | 600–1,200 kg | > 1,500 kg |
| Route distance | 30–200 m per trip | 200–500 m (cycle time impact) | < 15 m (person faster) |
| Trips per shift | 10–80 | 80–150 (need multiple robots) | > 200 (consider conveyor) |
| Layout change frequency | Quarterly or less | Monthly (re-map required) | Weekly or more |
| Mixed human traffic | Yes — AMRs designed for this | Dense pedestrian areas (slow) | Active forklift priority zones |
| Shift pattern | 2–3 shifts | 1 shift (longer payback) | Seasonal < 6 months/year |
AMR Categories by Payload and Use Case
| Category | Payload Range | Typical Form Factor | Primary Use | Unit Price (as of Sep 2026) |
|---|---|---|---|---|
| Shelf-carrier | 100–600 kg | Flat platform under shelf/cart | Goods-to-person picking, kitting | $25,000–$55,000 |
| Conveyor-top AMR | 200–500 kg | Platform with integrated roller/belt conveyor | Automated load transfer at stations | $40,000–$75,000 |
| Tugger AMR | 500–1,500 kg | Tractor unit towing cart train | High-volume milk runs, heavy loads | $35,000–$90,000 |
| Forklift AMR | 1,000–2,500 kg | Counterbalanced fork truck | Pallet transport, racking access | $80,000–$180,000 |
| Outdoor/yard AMR | 300–1,000 kg | Ruggedised platform, IP65 | Inter-building, yard moves | $50,000–$120,000 |
| Collaborative mobile manipulator | 5–25 kg (arm payload) | AMR base + cobot arm | Mobile machine tending, inspection | $80,000–$160,000 |
For most manufacturing AMR projects, the shelf-carrier and tugger categories cover 80% of applications. Conveyor-top AMRs are increasingly popular where automated load transfer eliminates manual docking at workstations.
Fleet Sizing Methodology
Fleet size is a throughput calculation, not a headcount substitution. The correct approach is to model your material flow demand and divide by what a single robot can deliver per shift.
Step 1: Map Your Demand
For each route, establish:
- Moves per shift — total transport missions required in an 8-hour shift
- Load per move — weight and volume per trip (determines robot type)
- Route distance — one-way meters from pickup to drop-off
- Time windows — are moves evenly distributed or do they peak at shift start?
Step 2: Calculate Single-Robot Throughput
A single AMR's cycle time per mission:
Typical values:
- Robot speed: 1.0–1.8 m/s (automatically reduced around people to 0.3–0.8 m/s)
- Docking time at each end: 15–60 seconds depending on mechanism
- Queue/wait time: 5–15% overhead for traffic and charging
A robot covering a 100 m route (200 m round trip) at an average effective speed of 1.2 m/s, with 45 s docking each end:
In an 8-hour shift with 90% uptime and 15-minute charging break: effective operating time = 420 min. Missions per robot per shift = 420 / 4.3 = ~98.
Step 3: Size the Fleet
Round up to the nearest whole number. Add one spare robot if fleet size is 5 or more (maintenance buffer). For peak-demand scenarios, size to meet peak throughput — fleet sits idle at non-peak times, which is acceptable.
Worked Example
| Parameter | Value |
|---|---|
| Shifts per day | 2 × 8 hours |
| Route: receiving to line-side (one-way) | 120 m |
| Demand: pallet moves per shift | 60 |
| Robot travel speed (effective average) | 1.1 m/s |
| Docking time per end | 50 s |
| T_cycle | (240/1.1) + 100 = 318 s = 5.3 min |
| Missions per robot per shift (90% uptime) | 432 min / 5.3 = 81 |
| Robots required (× 1.15 contingency) | 60 / 81 × 1.15 = 0.85 → 1 robot |
| Conclusion | 1 tugger AMR replaces 2 FTE drivers across 2 shifts |
In this case, a single robot handles both shifts with capacity to spare. If demand grows to 100 moves per shift, a second robot is needed. The fleet management software handles priority queuing automatically.
WMS and MES Integration
An AMR fleet operating on a fixed repeating schedule is the simplest integration scenario and requires no WMS connection — you program routes and schedules directly in the fleet management software (FMS). However, most manufacturing deployments benefit from demand-driven dispatch, where the AMR is called when a task actually arises rather than on a timer.
Integration Levels
| Level | How It Works | Integration Effort | Best For |
|---|---|---|---|
| Standalone schedule | Robot runs fixed milk-run loop independent of WMS | None — configure in FMS UI | Simple repeating routes, no WMS |
| Button/call station | Operator presses physical button to call robot to their station | Low — wired or wireless call buttons ($500–$2K per station) | Assembly lines, replenishment on demand |
| REST API integration | WMS sends dispatch task via HTTP API when order is released; FMS assigns robot | Medium — 2–4 weeks development; WMS must support outbound calls | SAP/Oracle WMS, intralogistics systems |
| MES integration | MES triggers material delivery based on production order state (e.g., kitting when job reaches station) | Medium-high — 4–8 weeks; requires MES job state visibility | Manufacturing cells with MES (Plex, Epicor, etc.) |
| Full VDA 5050 / MassRobotics | Standardised AMR interface; multi-vendor fleet interoperability | High — both AMR and FMS must support protocol | Large mixed-vendor fleets, third-party FMS |
The VDA 5050 Standard
VDA 5050 is a communication interface standard developed by VDA (German automotive industry association) and VDMA for AMR/AGV fleet management. It defines JSON messages over MQTT for order assignment, state reporting, and error handling. Key AMR vendors (MiR, Omron, OTTO Motors) support it, enabling a single third-party fleet manager to coordinate robots from different manufacturers.
For most SME deployments (single vendor, < 10 robots), VDA 5050 is unnecessary complexity. Use the vendor's own FMS unless you have a specific multi-vendor or third-party FMS requirement.
Integration Cost Budget
- No integration (schedule only): $0 additional
- Call stations: $3K–$15K depending on station count
- REST API to WMS: $15K–$40K (development, testing, UAT)
- Full MES integration: $35K–$80K
- Annual maintenance/support: 15–20% of integration development cost
Infrastructure Requirements
One of AMRs' key selling points is minimal infrastructure change. That is broadly true — but "minimal" is not "zero." Plan for the following before deployment.
Physical Infrastructure
- Aisle width: AMR width + 600 mm minimum clearance each side for pedestrian co-presence. A 600 mm wide AMR needs a 1.8 m aisle minimum. Tighter is workable but will increase pedestrian-triggered slowdowns.
- Floor flatness: Standard concrete floors are fine. Surface FR3 or better per ACI 117. Avoid expansion joints wider than 12 mm without a ramp strip.
- Lighting: 100 lux minimum for camera-based AMRs. Lidar-only robots operate in darkness.
- Fire doors: Most AMRs integrate with fire door systems via dry contact or Modbus. Budget $2K–$5K per automated door if not already automated.
- Charging stations: One charger per 1–2 robots is typical. Each charger needs a dedicated 240V/32A circuit (for fast-charge models). Site chargers at natural idle points between routes.
Network Infrastructure
- Wi-Fi coverage: AMRs require consistent 5 GHz Wi-Fi coverage across all operating areas. Dead spots cause connection drops and mission failures. Budget a Wi-Fi site survey and access point additions ($5K–$20K for a typical factory floor).
- VLAN segmentation: Separate AMR traffic from production OT network. Most IT security policies require this.
- Latency: < 50 ms round-trip from robot to FMS server for reliable operation. On-premise FMS server is standard for manufacturing; cloud FMS available but adds latency.
Mapping and Commissioning
Initial map creation takes 1–3 days per building depending on size and complexity. The robot is driven manually (or teleoperated) while building the SLAM map, then landmarks are added, routes defined, and safety zones configured. After initial deployment, maps must be updated whenever racking or major equipment moves — this takes 2–4 hours per re-map and can usually be done by an internal technician after training.
Total Cost of Ownership
Upfront Costs
| Item | 2-Robot Fleet | 8-Robot Fleet |
|---|---|---|
| AMR hardware (per unit, mid-range) | $80,000 ($40K × 2) | $320,000 ($40K × 8) |
| Charging infrastructure | $8,000 | $24,000 |
| Fleet management software (FMS) licence | $12,000 | $30,000 |
| Site preparation (floor, doors, signage) | $5,000 | $15,000 |
| Wi-Fi infrastructure upgrades | $8,000 | $18,000 |
| WMS/MES integration development | $0–$30,000 | $20,000–$60,000 |
| Commissioning and training | $10,000 | $25,000 |
| Total installed | $123,000–$153,000 | $432,000–$492,000 |
Annual Operating Costs
| Item | Per Robot per Year | Notes |
|---|---|---|
| Software licence / FMS subscription | $2,000–$5,000 | Per-robot or fleet-level pricing |
| Maintenance contract | $3,000–$8,000 | Typically 8–15% of hardware cost |
| Battery replacement (cycle 3–5 years) | $800–$2,000 annualised | Li-ion; cycle 1,500–3,000 charges |
| Electricity (charging) | $400–$900 | 2-shift operation, $0.12/kWh |
| Internal technician time | $1,500–$3,000 | ~40–80 h/year at $35–45/h |
| Total per robot per year | $7,700–$18,900 | Wide range; use vendor quote for your region |
Labour Savings
Each AMR on a two-shift schedule replaces approximately 1.5–2.5 FTE, depending on route complexity and load. At a US all-in forklift operator cost of $55,000–$75,000 per year (wages, benefits, overtime, training):
- 2 FTE replaced = $110,000–$150,000 per year saved
- Annual AMR operating cost = $10,000–$20,000 per robot
- Net annual saving per robot = $90,000–$140,000
- Payback on $120K installed cost = 10–16 months
These are best-case numbers. Real projects with single-shift operation, lower labour rates, or complex integration typically show 24–36 month payback.
3-Year TCO Comparison: AMR Fleet vs. Two Forklift Operators
| Item | 2 AMRs (3-year) | 2 Forklift Operators (3-year) |
|---|---|---|
| Upfront capital | $130,000 | $30,000 (2 × forklifts, basic) |
| Annual labour cost | $0 | $140,000/year × 3 = $420,000 |
| Annual operations (fuel, maintenance) | $20,000/year × 3 = $60,000 | $12,000/year × 3 = $36,000 |
| Integration / IT | $20,000 | $0 |
| 3-year total cost | $210,000 | $486,000 |
| 3-year saving | $276,000 | — |
Key AMR Vendors for Manufacturing
| Vendor | Key Products | Payload Range | Strength | Typical Use |
|---|---|---|---|---|
| MiR (Mobile Industrial Robots) | MiR100, MiR250, MiR600, MiR1350 | 100–1,350 kg | Broad payload range, strong FMS, wide integrator network | Manufacturing, logistics, healthcare |
| OTTO Motors (Rockwell) | OTTO 100, OTTO 1500 | 135–1,500 kg | Heavy-duty, automotive-grade, enterprise support | Automotive, heavy manufacturing |
| Fetch Robotics (Zebra) | CartConnect, HMIShelf, Freight | 50–1,500 kg | Cloud FMS, strong WMS integrations, goods-to-person | Warehouse, e-commerce fulfilment |
| Omron (LD series) | LD-60, LD-90, LD-250 | 60–250 kg | ROS-based, open SDK, Sysmac integration | Electronics, light manufacturing |
| Locus Robotics | LocusBot | — | Order picking assistance, goods-to-person | E-commerce, pharmaceuticals |
| Geek+ | P-series, T-series, S-series | 300–2,000 kg | High-density shelf systems, strong Asia-Pacific presence | Retail warehousing, apparel |
| Dematic (KION) | Dematic AMR | 200–1,500 kg | Full system integration with Dematic WCS/WMS | Distribution, FMCG |
| Seegrid | Palion Lift, Palion Tow | 1,000–1,500 kg | Vision-only navigation (no lidar), strong in North America | Automotive, heavy manufacturing |
For manufacturing SMEs, MiR and Omron LD are the most common starting points — both have a broad integrator network, proven fleet management software, and fleet sizes suited to 2–15 robot deployments.
Is an AMR Fleet Right for You?
Choose AMRs if:
- You have repetitive point-to-point transport tasks eating 1+ FTE per shift
- Your facility layout changes occasionally but not weekly
- You have clear aisles of 1.8 m or wider and level concrete floors
- Two-shift or three-shift operation makes labour savings substantial
- You need flexibility to redeploy robots when production mix changes
- You want to redeploy displaced operators to value-added tasks rather than reducing headcount
- Labour availability is the constraint, not budget
Stick with manual or fixed automation if:
- Your routes are single-shift only with < 20 moves per shift (person faster)
- You need very high throughput (> 150 moves/hour on one route — conveyor is better)
- Your facility has narrow aisles, active forklift zones, or highly variable floor conditions
- Your layout changes more than once a month (re-mapping overhead kills ROI)
- Load weight exceeds 1,500 kg (fork AGV territory)
- You need sub-15 mm placement accuracy at docking stations
- Budget is under $80K all-in (not enough for a robot + integration)
Frequently Asked Questions
How do AMRs handle safety around pedestrians and forklifts?
AMRs use layered safety: a safety-rated lidar field detects obstacles and triggers slowdown (typically at 1.5–2 m) then stop (at 0.5–1 m). These fields are certified to ISO 3691-4 (industrial trucks) or ISO/TS 15066 for collaborative applications. The robots do not operate in zones with active counterbalanced forklifts — traffic management requires physical separation (barriers, floor zones, or traffic light systems) between AMR paths and forklift aisles. Mixed pedestrian/AMR areas work well; mixed AMR/forklift areas require careful zoning.
How long does AMR deployment take?
A straightforward 2–4 robot deployment with no WMS integration typically takes 8–16 weeks from purchase order to go-live: site survey and prep (2–4 weeks), robot delivery (4–8 weeks lead time), mapping and commissioning (1–2 weeks), operator training (1 week). Add 4–12 weeks for WMS/MES integration if required. Total project timeline for a mid-size factory with REST API integration: 16–28 weeks.
What happens when a robot breaks down or needs charging mid-mission?
Fleet management software monitors battery state continuously. Robots return to charge when battery drops below a configurable threshold (typically 20–30%) between missions — not mid-mission. If a robot fails mid-route, the FMS reassigns the mission to another available robot. The failed robot raises an alert and can usually be manually pushed to a service area; most AMRs are under 250 kg and manually moveable. With a 2+ robot fleet, single robot downtime does not halt operations.
Do AMRs require specialist maintenance staff?
No. Day-to-day operation requires no robotics expertise. After a 1–2 day training course, an internal technician can handle: map updates after layout changes, adding/editing routes, diagnosing and clearing obstacle alerts, replacing worn bumpers or lidar covers, and swapping batteries. Major mechanical repairs (motor, gearbox) are handled by the vendor under maintenance contract. Most manufacturers with a basic maintenance team can support an AMR fleet internally after the first year.
Can one fleet management system handle robots from different vendors?
Theoretically yes — VDA 5050 enables multi-vendor fleets under a single FMS. In practice, most SME deployments use a single AMR vendor, so this isn't needed. If you anticipate mixing vendors (e.g., MiR for light loads and OTTO Motors for heavy loads), verify both support VDA 5050 and test interoperability before committing. Alternatively, run two vendor FMS instances with manual task routing between them, which is simpler but not fully automated.
What is a realistic payback period?
Two-shift operations replacing fork truck operators: 14–22 months. Single-shift operations: 28–42 months. Projects with complex WMS integration: add 4–8 months to payback. Projects where robots supplement rather than replace labour (freeing operators for higher-value tasks): payback depends on value attributed to the freed capacity, which varies by application. Use our robot ROI calculator with your specific labour rate, shift pattern, and installed cost for a tailored estimate.
Sources and Further Reading
- ISO 3691-4:2020 — Industrial trucks: Safety requirements for driverless trucks
- VDA 5050 v2.0 — Interface for AGV/AMR communication (VDA/VDMA, 2022)
- MassRobotics AMR Interoperability Standard v1.0
- MiR Fleet Technical Specification, Mobile Industrial Robots (2025)
- Interact Analysis — Global AMR Market Report (2025)