What is AGV navigation?

AGV navigation is the method an automated guided vehicle uses to determine its position and follow a path. Common methods include magnetic tape, QR markers, LiDAR SLAM, vision SLAM, and reflective beacons, each trading off accuracy, flexibility, cost, and reconfiguration effort.

Does this page explicitly answer "agv amr navigation" intent?

Yes. The canonical URL /learn/agv-navigation treats "agv amr navigation" as an alias of "agv navigation" and answers both on one hybrid page without creating a duplicate route.

Why is there no separate page for "agv amr navigation"?

"agv amr navigation" and "agv navigation" belong to the same intent cluster. A separate route would create near-duplicate content risk, so the alias is merged into this single canonical page.

What is the difference between AGV navigation and AMR navigation?

AGV navigation follows a fixed physical or virtual path and stops when blocked, waiting for clearance. AMR navigation uses SLAM to localise against a map and dynamically replans around obstacles. The distinction is behavioural (replanning logic), not just sensor hardware.

Which AGV navigation method is the most accurate?

Reflective beacon triangulation typically delivers ±3-8 mm repeatability, followed by QR markers (±5-12 mm) and magnetic tape (±8-15 mm). SLAM methods are more flexible but usually land at ±15-30 mm depending on environmental features.

Is the navigation selector a final engineering approval?

No. It is a deterministic pre-screen for architecture and sourcing decisions. Final selection still requires supplier evidence, an on-site pilot, and internal safety validation per ISO 3691-4 or your local standard.

Can I mix navigation methods in one fleet?

Yes. Hybrid fleets are common: SLAM for flexible zones and markers or beacons for fixed high-throughput loops. Ensure your fleet manager supports multi-method vehicles and that handoff zones are tested for localisation continuity.

How does navigation cost scale with fleet size?

Infrastructure-bound methods (tape, markers, beacons) scale cheaply because the per-vehicle sensor is inexpensive. SLAM methods carry a high per-vehicle sensor cost, so total cost rises steeply with fleet size. A common integrator rule-of-thumb puts the crossover at roughly 20 vehicles (not a published benchmark); beyond it, hybridising SLAM with markers/beacons usually lowers total cost of ownership.

What safety standard applies to AGV/AMR navigation?

ISO 3691-4 (current 2023 edition) is the Type-C safety standard for driverless industrial trucks, including AGVs and AMRs. It adopts EN ISO 13849-1 performance levels — most AGV/AMR safety functions require PL d (Category 3, equivalent to SIL 2). Person detection must use an IEC 61496-1/-3 Type 3 safety area scanner (e.g. SICK microScan3 / nanoScan3, KEYENCE SZ-V, OMRON OS32C). This is separate from the navigation method choice.

What is VDA 5050 and does my AMR need it?

VDA 5050 (v2.0.0 published January 2022 by VDA/VDMA) is an open JSON-over-MQTT interface between AGVs/AMRs and a master control (fleet manager). AMRs are defined as "free navigation AGVs". It does not replace navigation — it lets mixed-vendor fleets share one fleet manager. You need it when running more than one vehicle brand under a single master control.

Why does wheel odometry matter if I use SLAM?

Every navigation method fuses its correction onto a wheel-odometry motion model. Odometry accumulates error from wheel slip, tyre mismatch, wheelbase uncertainty, encoder quantisation and floor roughness; typical un-fused differential-drive odometry drifts on the order of 1–2% distance error and 2–3° heading error over 100 m. SLAM, markers or beacons re-localise to bound that drift, so the drive wheel, encoder and tyre set the accuracy floor any method can reach.

AGV / AMR navigation method selector

AGV navigation & AGV AMR navigation method selector

Pick the right navigation method for your automated guided vehicle or autonomous mobile robot fleet. Enter your fleet, accuracy, flexibility, budget, and environment constraints to get a scored recommendation with accuracy ranges, cost bands, and reconfiguration trade-offs, then read the full method and evidence report below.

Methods scored

tape · QR · LiDAR · vision · beacon

Accuracy range

±3-40 mm

method-dependent

Decision mode

4 criteria

accuracy · flex · cost · env

Deployment

1-3 weeks

mapping → beacon calibration

Run the selector Read the method report

Canonical path: /learn/agv-navigation · covers the alias agv amr navigation

Published 2026-06-17 - Last updated 2026-06-17

Navigation method selector inputs

Enter your fleet and constraint profile. The selector scores five navigation methods and returns a recommendation with a fit band.

Fleet size (vehicles)

Range 1-500

Required positioning accuracy (mm)

Lower = tighter tolerance. Map to actual docking tolerance.

Path flexibility needed

Layout change frequency

Budget tier

Environment type

Recommendation & fit band

Scored output with matched criteria, alternatives, and next action.

Enter your constraints and press Run selector to see the recommended navigation method.

Key conclusions on AGV & AMR navigation

Five evidence-backed conclusions for fleet architects evaluating navigation methods. Each is traceable to the method, accuracy, and source tables below.

1. No single method wins on all axes

High accuracy, high flexibility, and low cost are mutually exclusive. Reflective beacons win on accuracy; SLAM wins on flexibility; tape wins on cost. Pick the two that matter most.

2. SLAM flexibility costs per-vehicle

LiDAR/vision SLAM removes floor infrastructure but carries the highest per-vehicle sensor cost. A common integrator rule-of-thumb puts the total-cost crossover near ~20 vehicles (not a published benchmark); beyond it, hybrid fleets usually win.

3. Accuracy claims need field validation

Published repeatability (±3-40 mm) is vendor nominal under controlled conditions. Feature-poor aisles, lighting, and floor wear degrade real performance. Always pilot before lock-in.

4. AGV vs AMR is a behaviour split, not a sensor split

The core difference is dynamic replanning (AMR) vs stop-and-wait (AGV). Both can use SLAM sensors; the navigation logic decides the category.

5. Hybrid fleets are the pragmatic norm

Mixing SLAM (flexible zones) with markers or beacons (fixed loops) balances cost and flexibility. Test handoff zones for localisation continuity.

6. Reconfiguration effort decides TCO

For frequently changing layouts, low-reconfig methods (SLAM) beat high-reconfig methods (tape) on total cost of ownership even with higher upfront sensor spend.

7. Accuracy is bounded by odometry first

Every method fuses its correction onto wheel odometry, which drifts from slip, tyre mismatch and encoder quantisation (~1–2% distance / 2–3° heading per 100 m un-fused). The drive wheel, encoder and tyre set the floor no method can beat.

8. Safety is set by ISO 3691-4, not the nav method

All five methods still require PL d / SIL 2 person detection (EN ISO 13849-1) via a Type 3 safety area scanner. Picking a navigation method and passing the safety case are separate decisions.

Suitable for

Indoor AGV/AMR programs that can define accuracy, flexibility, budget, and environment constraints and run a mapping pilot before fleet-wide procurement.

Not suitable for

Frozen specs demanding ±5 mm accuracy, free roaming, and low cost simultaneously; outdoor-only fleets without beacon infrastructure; programs that cannot pilot.

AGV navigation vs AMR navigation

The alias "agv amr navigation" spans both categories. The distinction is behavioural (replanning logic), not just the sensor hardware.

Dimension	AGV navigation	AMR navigation
Path definition	Fixed path (tape, wire, markers)	Computed dynamically from a map
Obstacle response	Stop and wait for clearance	Replan around the obstacle
Typical sensor	Magnetic / camera line follower	LiDAR or camera SLAM stack
Reconfiguration	Physical infrastructure change	Software remap
Positioning accuracy	Higher (constrained path)	Moderate (map-dependent)
Best deployment	Predictable, high-throughput loops	Variable, human-shared spaces
Fleet comms	VDA 5050 compliant (orders along fixed nodes/edges)	VDA 5050 compliant (defined as "free navigation AGVs")
Safety sensor	PL d / Type 3 area scanner (ISO 3691-4)	PL d / Type 3 area scanner (ISO 3691-4)

Method, accuracy & evidence

How each navigation method works, its typical repeatability, and the evidence basis. Numbers are vendor-nominal unless stated.

Typical repeatability

Lower bars are better. Ranges reflect published nominal repeatability, not guaranteed field accuracy.

Reconfiguration effort by method

How much work is required to reroute when the layout changes.

Method	Repeatability (mm)	Drift over 30 m	Relocalisation	Evidence
Reflective beacon	3-8	Negligible (absolute fix)	Instant on beacon sighting	vendor-nominal
QR marker	5-12	Reset at each marker	Per-marker reset	vendor-nominal
Magnetic tape	8-15	Bounded by line tracking	Continuous on tape	vendor-nominal
LiDAR SLAM	15-30	5-15 mm if features stable	Scan-match relocalisation	vendor-nominal
Vision SLAM	20-40	10-25 mm, lighting dependent	Visual feature re-detect	vendor-nominal

Method	Per-vehicle hardware	Infrastructure	Fleet scaling	Evidence
Magnetic tape	Low (line sensor)	Tape / embed full route	Low marginal cost	market-range
QR marker	Low-mid (camera)	Marker stickers on floor	Low marginal cost	market-range
LiDAR SLAM	High (2D/3D LiDAR + compute)	Mapping only	High per-vehicle cost	market-range
Vision SLAM	Mid-high (cameras + compute)	Mapping only	Mid per-vehicle cost	market-range
Reflective beacon	Mid-high (laser scanner)	Wall/column reflectors	Beacons shared across fleet	market-range

Odometry: the foundation every method builds on

All five methods fuse their correction onto a wheel-odometry motion model. The drive wheel, encoder and tyre set the accuracy floor no method can beat. Typical un-fused differential-drive odometry drifts on the order of 1–2% distance error and 2–3° heading error over 100 m; pure odometry is unreliable beyond ~10 m without re-localisation.

Error source	Type	Effect	Mitigation
Wheel slip / skid	Non-systematic	Measured wheel travel exceeds ground travel; position overestimated, worst on smooth/wet floors and hard acceleration.	High-traction tyres, gentler acceleration profiles, and IMU/LiDAR fusion via an EKF.
Wheel diameter mismatch	Systematic	Unequal effective radii produce a scale bias that curves every straight path.	Calibrate kinematic parameters (e.g. UMBmark test); use matched drive wheels.
Wheelbase uncertainty	Systematic	Heading bias on every turn; accumulates as lateral drift.	Measure the effective track precisely, or use unloaded passive encoder wheels.
Encoder quantisation	Systematic	Discrete pulse counts add integration noise (~0.3 mm per pulse at 1000 ppr on a 0.1 m wheel).	Higher PPR encoders (≥1000–2000; 4000+ for precision).
Floor roughness / unevenness	Non-systematic	Random heading and distance noise that grows with distance.	Periodic relocalisation on markers, beacons, or SLAM scan-match.

Standards & safety framework

Choosing a navigation method and passing the safety case are separate decisions. The standards below govern AGV/AMR deployments regardless of whether you run tape, markers, SLAM or beacons.

Reference	Scope	Key requirement	Evidence
ISO 3691-4 (2023)	Industrial trucks — safety of driverless trucks (AGV/AMR) — Type-C product-safety standard for AGV, AMR, AGC and similar trucks per ISO 5053-1.	Mandates hazard/risk assessment (Annex B), personnel detection, braking, speed control and stability; adopts ISO 13849 performance levels for safety functions. In an operating hazard zone with no pedestrian escape route (≥0.5 m × 2.1 m), personnel detection must cover to within 180 mm of surrounding objects.	primary-standard
EN ISO 13849-1	Safety-related parts of control systems (performance levels) — Functional-safety framework assigning performance levels a–e from severity, exposure and avoidability.	AGV/AMR safety functions typically require PL d (Category 3), equivalent to SIL 2 — a dangerous-failure rate of 10⁻⁷ to <10⁻⁶ per hour. Determines whether the navigation/safety stack is certifiable, independent of which navigation method is chosen.	primary-standard
IEC 61496-1 / -3	Electro-sensitive protective equipment (ESPE), Type 3 — Construction and testing of safety laser scanners used for person detection on moving vehicles.	Certifies the area scanner used for the safety stop. The protective field (stop) is distinct from warning fields (slow/pre-warn), which must not be relied on for personnel protection.	primary-standard
VDA 5050 v2.0 (Jan 2022)	AGV/AMR ↔ master-control communication interface — Open JSON-over-MQTT interface so mixed-vendor AGVs and AMRs share one fleet manager. AMRs are defined as "free navigation AGVs".	Standardises orders, nodes, edges, state and instant actions — not a full control system. At the VDMA AGV Mesh-up (2023) a heterogeneous fleet was integrated in under two days. Relevant once an AMR fleet needs a single master control regardless of vendor.	primary-standard

Safety-rated area scanner is mandatory, not optional

Under ISO 3691-4, person detection must reach PL d / SIL 2 via an IEC 61496-3 Type 3 safety laser scanner. The navigation LiDAR (used for SLAM) is usually a separate, non-safety-rated sensor. Common safety scanners: SICK microScan3 / nanoScan3, KEYENCE SZ-V / SZ, OMRON OS32C (all Type 3 · SIL 2 · PL d). Detection resolution: ~30/40 mm = hand, ~50/70 mm = leg, ~150 mm = body.

Scanner class	Rating	Detection note	Typical use
SICK microScan3 / nanoScan3	Type 3 · SIL 2 · PL d · IP65	30/40 mm = hand, 50/70 mm = leg, 150 mm = body (per SICK documentation).	Mobile hazardous-area protection on AGVs/AMRs; protective + warning field switching by speed.
KEYENCE SZ-V / SZ	Type 3 · SIL 2 · PL d · Cat 3	Up to four simultaneously monitored protective fields; muting and bank settings.	AGV safeguarding and robotic-cell guarding; alternative to hard guarding and safety mats.
OMRON OS32C	Type 3 · SIL 2 · PL d	Configurable protective/warning fields; PROFINET/EFI options.	AGV/AMR person protection and access guarding.

Research basis updated 2026-06-17. Standard editions: VDA 5050 v2.0.0 (January 2022); ISO 3691-4 current edition 2023. Scanner models listed are representative examples, not endorsements.

Comparison, trade-offs & risks

Side-by-side comparison, the cost-accuracy frontier, deployment scenarios, and the risks of mis-selecting a navigation method.

Method	Accuracy	Flexibility	Cost	Reconfig	Best for	Main limit
Magnetic tape	±10 mm	Fixed line only	$ (lowest)	High (re-tape floor)	High-throughput repeatable loops	Tape wear; no dynamic rerouting
QR / fiducial marker	±8 mm	Grid-based, semi-flexible	$$	Medium (re-stick markers)	Dense warehouses with shelf aisles	Markers can be obscured or damaged
LiDAR SLAM	±20 mm	Free roaming	$$$$ (highest)	Low (remap only)	Dynamic, frequently changing layouts	Feature-poor aisles degrade localisation
Vision SLAM	±25 mm	Free roaming	$$$	Low (remap only)	Cost-sensitive free-roaming indoor fleets	Sensitive to lighting and visual symmetry
Reflective beacon	±5 mm	Open-area triangulation	$$$	Medium (relocate beacons)	Large open bays needing high repeatability	Line-of-sight to beacons required

Cost vs accuracy frontier

The top-left quadrant (high accuracy, low cost) is empty by design, reflecting the fundamental trade-off.

High-throughput pallet loop

FIT

Fixed route, large fleet, tight budget

Input: 24 vehicles · ±15 mm · fixed · low budget

Recommended: Magnetic tape navigation (100/100)

Magnetic tape navigation meets all four constraint criteria for your AGV/AMR navigation profile. Accuracy, path flexibility, budget tier, and environment are all within the method's defensible envelope.

Dynamic e-commerce fulfilment

FIT

Frequent layout change, free roaming

Input: 12 vehicles · ±25 mm · free-roaming · high budget

Recommended: Vision SLAM / VSLAM (112/100)

Vision SLAM / VSLAM meets all four constraint criteria for your AGV/AMR navigation profile. Accuracy, path flexibility, budget tier, and environment are all within the method's defensible envelope.

Precision assembly feed

FIT

High accuracy, semi-flexible, mid budget

Input: 6 vehicles · ±5 mm · semi-flexible · mid budget

Recommended: Reflective beacon / grid navigation (100/100)

Reflective beacon / grid navigation meets all four constraint criteria for your AGV/AMR navigation profile. Accuracy, path flexibility, budget tier, and environment are all within the method's defensible envelope.

Conflicting spec (over-constrained)

RECONSIDER

High accuracy + free roaming + low budget

Input: 8 vehicles · ±5 mm · free-roaming · low budget

Recommended: Vision SLAM / VSLAM (62/100)

Your accuracy, flexibility, and budget constraints conflict. No single navigation method satisfies all criteria at once. Relax at least one constraint before sourcing navigation hardware.

Risk	Trigger	Impact	Mitigation
Accuracy gap in feature-poor aisles	Long straight corridors with few geometric features	High	Add fiducial markers as localisation anchors, or fall back to reflective beacons in those aisles.
Tape / marker damage	Forklift traffic, floor cleaning, spills	Medium	Schedule marker inspection, use recessed tape, and keep a replacement-marking runbook.
Map drift after layout change	Racking moved or pallets repositioned without remap	High	Trigger a remap after any documented layout change and version-control the map.
Cost overrun on large SLAM fleets	Scaling LiDAR SLAM beyond 20+ vehicles	Medium	Hybridise: SLAM for flexible zones, markers for fixed high-throughput loops.
Lighting sensitivity (vision SLAM)	Aisles with strobe, glare, or low-light shifts	Medium	Add active illumination or switch those zones to LiDAR SLAM.
Over-specifying accuracy	Requesting ±5 mm when pick faces tolerate ±25 mm	Low	Map accuracy to the actual docking tolerance, not a round number.

Frequently asked questions

Grouped by topic. Covers both "agv navigation" and the alias "agv amr navigation".

AGV/AMR navigation basics

Choosing a navigation method

Cost and deployment

Accuracy, risk, and validation

Evidence basis & sources

Accuracy and cost claims are traced to source categories. Where open cross-vendor evidence is incomplete, the status is marked explicitly.

Source	Scope	Date	Status	Note
Navigation method accuracy ranges	Vendor datasheets and integration guides	2024-2026	Partially known	Reported as nominal repeatability under controlled indoor conditions; field performance varies.
Per-vehicle cost bands	AGV/AMR integrator quotations and market ranges	2024-2026	Partially known	Cost indices are relative bands, not fixed prices; request live quotes for current figures.
AGV vs AMR behavioural distinction	Industry association definitions	2023-2026	Known	AGV = fixed-path / stop-on-obstacle; AMR = dynamic replanning. Boundary blurs in marketing copy.
Reconfiguration effort ranking	Integration case studies	2024-2026	Partially known	SLAM remap is faster than physical tape/marker relocation, but remap quality depends on operator skill.
Safety framework (ISO 3691-4 · EN ISO 13849-1 · IEC 61496-3)	Primary international standards (type-C + functional safety + ESPE)	13849 / 61496 current; ISO 3691-4 edition 2023	Known	AGV/AMR safety functions require PL d (Cat 3, SIL 2) person detection via a Type 3 safety area scanner. Applies to every navigation method.
VDA 5050 fleet communication interface	VDA / VDMA open standard (JSON over MQTT)	v2.0.0, January 2022	Known	Standardises AGV/AMR ↔ master-control messaging; AMRs are "free navigation AGVs". Enables mixed-vendor fleets under one fleet manager.
Wheel-odometry error model	Mobile-robotics literature (systematic vs non-systematic error)	2018–2025	Partially known	Slip, tyre mismatch, wheelbase uncertainty and encoder quantisation accumulate; ~1–2% distance / 2–3° heading per 100 m is engineering-typical, not a guaranteed figure.

VDA 5050VDA / VDMA · v2.0.0, January 2022
VDA 5050 v2.0.0 — Interface for communication between AGVs/AMRs and master controlChecked 2026-06-17
ISO 3691-4ISO · Current edition 2023
Industrial trucks — Safety requirements and verification — Part 4: Driverless industrial trucksChecked 2026-06-17
EN ISO 13849-1Pilz (standards reference) · Reference page
Safety-related parts of control systems — performance levels (PL d)Checked 2026-06-17
IEC 61496-3SICK (product documentation) · Datasheet compendium
Safety laser scanners (microScan3 / nanoScan3): Type 3 · SIL 2 · PL d dataChecked 2026-06-17
ROS 2 Nav2Open Robotics / ROS 2 · 2025
Navigation stack — AMCL localisation, costmaps, replanningChecked 2026-06-17
Odometry errorMDPI Robotics, 13(1):7 · 2025
Online odometry calibration in low-traction conditions (systematic vs non-systematic error)Checked 2026-06-17

Next steps

Run the selector

Enter your fleet constraints above to get a scored recommendation before talking to suppliers.

Plan a mapping pilot

For review-band results, budget a 2-4 week pilot to measure repeatability over a representative traverse.

Request an architecture review

For conflicting constraints, request a navigation architecture review to find a hybrid method set.

Request navigation architecture review

Related engineering resources

Continue with adjacent drivetrain checks, navigation evidence review, and direct RFQ actions.