What Should an AMR Site Survey Measure Before a Robot Is Selected

An autonomous mobile robot should not be selected from a brochure and then introduced to the building. The building must be measured first. A decision-grade AMR site survey converts routes, floors, interfaces, utilities, radio conditions and traffic behavior into evidence that engineering, EHS, IT/OT, operations, procurement and the integrator can approve. Its purpose is not to prove that a robot can move during a polished demonstration. Its purpose is to reveal the physical and digital changes required for a repeatable production service before those changes become late-stage cost, delay or safety surprises.

The direct answer is simple: survey the complete mission in its real operating state. Measure the loaded vehicle envelope, every route transition, every station interface, charging access, shared doors and lifts, wireless behavior and mixed traffic during representative shifts. Record each observation against a requirement source, an owner and a release status. Do not accept a room-wide average where a local defect can stop a mission, and do not use a supplier’s generic minimum where the actual payload, top module, protective-field configuration or recovery method changes the requirement.

A Site Survey Is a Release Decision, Not a Factory Walk

A casual walk-through produces impressions: “the aisle looks wide,” “the floor seems level,” or “Wi-Fi is available.” A release-quality survey produces traceable statements: route segment R-17 has a measured clear width of X under normal staging conditions; the configured robot-and-load envelope needs Y under the supplier’s stated test assumptions; the difference is Z; the finding is red until Facilities relocates the barrier and Operations prohibits temporary storage there.

This distinction matters because the operating zone is part of the system. The official scope of ISO 3691-4:2023 explicitly recognizes that operating-zone conditions can significantly affect the safe operation of driverless industrial trucks, including autonomous mobile robots. A survey is therefore not merely a commercial discovery exercise. It is an engineering input to application design and the site-specific risk assessment. It does not replace that risk assessment, supplier validation or formal acceptance testing.

Keep the boundary clear. A survey determines what exists, what varies and what must change before selection or detailed design. The broader question of whether the organization can deliver the entire program—including governance, process ownership and implementation capability—is addressed separately in this AMR delivery feasibility assessment. Navigation selection, radio engineering and production acceptance also deserve their own detailed work. The survey feeds those decisions; it should not pretend to finish them.

Build the Evidence Chain Before Entering the Site

The most useful AMR site survey checklist is not a long list of yes-or-no questions. It is an evidence chain with eight actions: scope, observe, measure, challenge, classify, remediate, recheck and sign. If any action is missing, an apparently complete report can still fail at procurement. A measurement without a requirement cannot be judged. A nonconformance without an owner does not get corrected. A correction without a recheck remains an assumption.

1. Scope the Missions and Time Windows

Start with missions, not square meters. Give every proposed flow a mission ID and record its origin, destination, material, load carrier, maximum loaded mass, load dimensions, required orientation, release logic, expected trips per hour, peak window, empty-return logic and manual fallback. Mark the route on a controlled drawing. If inbound and outbound travel use different lanes, treat them as different route segments.

Time is also part of the scope. A route inspected at 10:00 may be unrecognizable during shift change, replenishment, sanitation, trailer unloading or production changeover. Define the operating windows that must be observed. Record future-state missions separately from present-state missions so that a planned rack line or machine does not disappear from the design simply because it was absent on survey day.

2. Establish IDs Before Collecting Evidence

Use stable IDs for missions, route segments, stations, doors, lifts, chargers, network zones and findings. A practical finding ID can follow site-area-asset-condition-sequence. Photographs should be named so another reviewer can locate the subject without opening every file: plant-a-route-r07-floor-joint-20260910-01.jpg. Record the direction of view, the marked location on the layout and a scale reference where dimensions matter. Avoid filenames such as IMG_4832.jpg, which break the connection between the image and the decision.

3. Define the Requirement Source

Each limit must identify its source: the shortlisted robot configuration, top-module drawing, load specification, supplier application manual, facility engineering standard, electrical code, fire/EHS requirement, cybersecurity rule or risk-assessment output. Write the configuration and document revision beside the value. “Vendor says okay” is not an auditable requirement. If candidate robots have different limits, preserve all candidate columns until selection instead of silently using the most permissive number.

4. Use One Survey Record for One Decision

A robust AMR facility assessment record should contain at least the following fields. The same structure can be used in a spreadsheet, mobile inspection form or database.

Measure the Floor as a Route, Not a Room

Floor suitability is often reduced to a single flatness statement. That is inadequate for a mobile robot. The mission experiences a sequence of local events: slab waves, joints, cracks, thresholds, drains, ramps, dock plates, coatings, contamination and transitions between floor materials. A single defect at a station approach can matter more than a good average across the warehouse.

Translate AGV floor requirements into a route-based measurement plan. Divide each mission into meaningful segments, then add event points wherever the wheel, caster, load or sensor response may change. Survey both travel directions when the vehicle dynamics or load orientation differ. Measure under representative floor conditions; a clean, dry shutdown survey may not describe a route exposed to oil mist, washdown water, shrink-wrap fragments or cardboard dust during production.

Level, Flatness and Local Discontinuities

Level describes the overall tilt of a surface; flatness describes local variation. They are not interchangeable. For AMR floor flatness, the correct tolerance depends on wheel diameter and material, suspension, ground clearance, load stability, localization method, vehicle speed and the supplier’s validated application limits. Do not publish or purchase against a universal millimeter value. Instead, establish the candidate-specific measurement method, sample spacing and acceptance limit, then map every out-of-tolerance location.

Pay special attention to expansion joints, damaged joint edges, covers, cable protectors, drainage grates, embedded scales, elevator sills and transitions to dock equipment. Record gap width, height difference, edge shape, approach angle and whether the condition changes under load. A small vertical step can create impact, wheel slip, payload oscillation or a localization disturbance even if the robot clears it physically.

Slopes, Compound Grades and Stopping Behavior

Measure slope length, gradient, cross-slope, transition geometry, surface condition and the available level area before and after the grade. Identify whether the robot must stop, turn, dock or queue on the incline. The relevant limit is the configured vehicle under the specified load and speed—not an unloaded brochure maximum. Challenge the route for a safe stop, restart, loss-of-power response and recovery scenario. A robot that can climb a ramp is not automatically suitable for holding a tall or liquid load on it.

Traction and Changing Surface Conditions

Record coatings, polished areas, worn zones, metal plates, anti-static flooring, moisture, oils, dust and cleaning chemicals. Ask how the surface varies by season, sanitation cycle and production process. Photograph the condition and collect maintenance history where available. If the survey cannot reproduce the worst credible state, classify the finding as not yet verified and define a controlled test. Visual inspection alone cannot establish braking or traction performance.

Floor Evidence to Retain

• An annotated route drawing with measurement stations and defect IDs.
• Instrument type, calibration status, sampling method and raw readings.
• Scaled photographs of joints, gaps, transitions and damage.
• Surface-condition observations tied to shift and production state.
• The candidate-specific limit, source document and configuration.
• A repair method, area, shutdown requirement, cost owner and recheck date.

Prove the Loaded Motion Envelope

The robot chassis is not the moving envelope. The real object includes the top module, load carrier, payload overhang, tolerances, protective fields and any deformation or sway. A tugger train, forked load, high rack, roll cage or offset pallet can produce a very different swept path from the base shown on a data sheet.

Start with a controlled configuration drawing for every load family. At each restriction, measure fixed obstruction clearance and the space normally consumed by pallets, bins, hoses, waste containers, work-in-process and parked manual equipment. A dimension taken between building columns is meaningless if normal operations reduce the usable opening.

Aisles, Turns and U-Turns

Static geometry can be screened with a simple relation:

screening width = configured robot-and-load width + left allowance + right allowance

That relation is only a screening tool. Final AMR aisle width requirements must account for the dynamic swept path, localization tolerance, protective-field behavior, opposing traffic, fixed and temporary obstructions, load movement and the recovery method. Map ninety-degree turns, S-curves, U-turns, merges, blind corners and approaches to stations separately. The narrowest straight aisle may not be the controlling point; a corner with an overhanging load often is.

Passing, Waiting and Recovery Space

Measure where robots can pass, yield, queue or be removed without blocking production. A route that fits one moving vehicle may fail as a service when two vehicles meet or a stopped vehicle needs recovery. Document the proposed right-of-way rule and verify the physical space for it. Include pedestrian escape paths and required access to fire, electrical, first-aid and process equipment; an efficient robot route cannot consume protected access.

Challenge Geometry with Representative Loads

Where risk or tolerance is tight, use a full-scale envelope mock-up or the intended vehicle configuration under controlled conditions. Challenge the largest, most offset and least stable approved load—not the smallest demonstration tote. Record speed, direction, load state and observers. A video can support evidence, but the test must still be tied to measured dimensions and acceptance criteria.

Audit Every Mission Interface

Most missions fail at interfaces, not in the middle of an empty aisle. The site survey should therefore spend disproportionate effort at pickup, drop-off, docking, transfer and shared-building interfaces. Convert AMR infrastructure requirements into an interface control record for each station rather than a generic facility declaration.

Pickup, Drop-Off and Process Stations

For each station, record datum location, approach direction, usable floor area, height, lateral and angular tolerance, load-carrier condition, mechanical stops, sensors, guarding, buffer capacity and operator access. Measure several real pallets, carts or racks; nominal drawings often omit bent frames, wheel wear and accumulated manufacturing tolerances. Note whether operators can accidentally move the interface and how its position is restored and verified.

Observe the process sequence. Does the machine release a load before the robot is ready? Can an operator enter the transfer zone? What happens when the downstream station is full, a barcode is unreadable or a carrier is misaligned? The survey is not the final control design, but it must expose the physical states and exception paths that the design must handle.

Doors, Shutters and Elevators

Record clear opening, sill condition, approach space, opening and closing time, obstruction detection, control ownership, manual mode, fire-mode behavior and failure recovery. For elevators, also record car dimensions, leveling variation, rated load, traffic policy, communications availability, dispatch interface and what happens during an alarm or loss of service. Treat the door or lift as a dependency with an owner and availability requirement, not as empty space on a floor plan.

Station, Power and Network Readiness Matrix

Treat Charging as a Place, a Utility and a Recovery Zone

Charging readiness is more than finding a nearby outlet. AMR charging station requirements should cover the electrical source, docking geometry, surrounding traffic, thermal and environmental conditions, access for maintenance, fault isolation, queuing behavior and the safe recovery of a depleted or disabled robot.

Verify supply voltage and frequency, dedicated circuit expectations, protection devices, disconnect location, cable route, grounding or bonding requirements and available panel capacity with the responsible electrical engineer. Confirm fire, ventilation, temperature, humidity and hazardous-area constraints with EHS and the applicable local rules. Record who may isolate the charger and how the area remains accessible during production.

Measure the complete docking approach and the space needed to clean, inspect, repair or replace the charger. Protect the area from pallet staging and forklift impact. Check whether a waiting robot blocks an aisle or emergency path. If multiple robots share a charger, the site survey should reserve the physical queue and maneuvering space; the operational capacity and charging policy should then be engineered using a separate AMR charging strategy.

Finally, challenge failure. Can a low-battery robot reach the location under its approved operating rules? Can technicians recover a disabled unit without entering an uncontrolled traffic stream? Is there a temporary charging or tow method, and where is the recovered robot parked? These are site facts with layout and ownership consequences, not software settings to be solved later.

Survey What the Sensors Will Actually See

A floor plan cannot show the full sensing environment. Walk every route from the approximate height and direction of the intended sensors. Record glass walls and doors, mirrors, glossy metal, deep black or highly absorptive surfaces, direct sunlight, steam, dust, smoke, rain exposure, feature-sparse corridors, repeated rack geometry, dense pedestrian occlusion and objects below or above the primary sensing plane.

Document changes over time. An empty rack face may become a repetitive wall of identical cartons. A loading-bay door may create harsh backlighting at a particular hour. Temporary curtains may remove features used for localization. Hanging straps, fork tips or conveyor projections can enter a part of the envelope that a casual standing-height inspection misses.

The output is a map of sensing conditions and test scenarios, not a premature declaration of the winning technology. Where a condition may influence localization or detection, flag it for candidate-specific testing and link the finding to the controlled photograph. The separate AMR/AGV navigation guide explains how site characteristics should influence navigation selection; the survey supplies the evidence for that choice.

Follow the Mission with the Wireless Test

A phone showing Wi-Fi bars beside a route is not a wireless assessment. A useful AMR wireless site survey follows the mission in both directions, at the device’s expected antenna height and orientation, while the plant is in representative operation. It records where performance changes, which access points serve the path, how roaming behaves and what production equipment or structures may alter propagation.

The survey scope should include coverage, signal quality, interference, channel use, retries, packet loss, latency variation, roaming events and relevant network segmentation or authentication constraints. Do not turn one generic signal-strength threshold into a universal acceptance criterion. The required evidence depends on the robot communications architecture, the criticality of messages, application response to disconnection and site design. NIST’s Industrial Wireless Systems program emphasizes reliability, latency, resilience and the measurement of propagation and interference in industrial deployments—exactly why route and operating context matter.

Test during the production states defined in the survey plan. Include doors open and closed, high-bay inventory present, nearby wireless devices active, and representative machine operation where practical. Tie anomalies to route coordinates and timestamps so IT/OT can correlate them with controller and network logs. For RF design, roaming validation and remediation methods, use the site’s detailed guide to industrial wireless networks for mobile robots rather than duplicating it inside the physical survey.

Observe Mixed Traffic by Time, Not from a Quiet Snapshot

Traffic is a time-varying condition. Schedule observations during representative production peaks, breaks, shift changes, replenishment, waste collection, sanitation and maintenance. At defined route points, record pedestrian flow, forklift crossings, manual-cart movement, vehicle speed, direction, queue formation, blind approaches, temporary staging and rule compliance. A short time-lapse or fixed observation period can reveal patterns that disappear during a guided tour.

Separate flow evidence from the safety decision. The survey may count crossings, map conflicts and identify visibility constraints. Qualified safety work must then decide protective measures, right-of-way rules, signs, barriers, speed zoning, access controls and validation. Do not label a mixed-traffic route “safe” solely because no incident occurred during the visit.

Use a Time-State Observation Matrix

Record the Future State Before It Invalidates the Survey

A survey begins aging as soon as the building changes. Ask Facilities, IT/OT and Operations for approved and probable changes over the design horizon: new racks, machines, mezzanines, fire doors, floor repairs, production lines, access points, network policies, staging rules, load carriers and traffic volumes. Mark them on a future-state drawing with an implementation date and confidence level.

Also define change triggers that require partial resurvey. Examples include resurfacing a route, changing wheel-contact materials, moving a station, increasing payload dimensions, modifying a door controller, replacing access points, changing the charger model or introducing a new forklift flow. The survey report should state its baseline date and configuration; it is not a permanent certificate for every future layout.

Demand growth belongs in the evidence package as well. Site observations—route congestion, station dwell, door-cycle time and shared-resource availability—become inputs to the AMR fleet-capacity model. Keeping those inputs traceable prevents a later fleet estimate from being based on an idealized CAD route with no operating delays.

Convert Findings into Red, Amber and Green Release Conditions

Color is useful only when it changes a decision. Define the rule before applying it, retain “to be verified” as a separate state, and do not average findings into a comforting site-wide score.

• Green: the named configuration meets the sourced requirement in the representative condition, the evidence is complete and no open action affects selection.
• Amber: a feasible remedy or controlled design assumption exists, with an owner, budget, due date, dependency and mandatory recheck. Amber is not permission to forget the issue.
• Red: the condition blocks the proposed mission, invalidates the current candidate or creates an unresolved safety or infrastructure dependency. Selection or route release must stop until the decision is changed or evidence closes the finding.
• To be verified: the necessary operating state, measurement, requirement or specialist review was unavailable. Lack of evidence is not green.

Worked Release Examples

This is a pre-selection release model. It should not be confused with the later proof that the integrated system is ready for production. Once the robot, controls, interfaces and operating procedures exist, the program needs a separate AMR go-live readiness process with witnessed tests and production evidence.

Build the Evidence Pack and Remediation Budget

The survey deliverable should allow a reviewer who did not attend the visit to reproduce the logic. At minimum, issue a controlled mission register, present-state and future-state layouts, route-segment register, measurement files, station matrix, traffic observations, wireless findings, photo index, requirement-source register, RAG log, remediation plan, exclusions and signatures.

For every amber or red item, separate the expected remedy into:

• Scope and location of work.
• Responsible function and approving function.
• One-time capital cost, recurring operating cost and contingency.
• Required shutdown or production-access window.
• Dependency on a supplier, landlord, utility, IT change or permit.
• Target completion date and consequence if late.
• Reinspection method and release authority.

This converts site readiness into the commercial baseline. Procurement can compare robot proposals against the same site facts, Finance can see enabling cost before approval, and project management can place infrastructure work on the critical path. Do not hide facility remediation inside a robot price comparison; a lower vehicle price can be economically irrelevant if it requires more floor repair, network work or station reconstruction.

Copy-Ready Survey Form

The following compact form can be copied into a spreadsheet or mobile inspection application. Use one row per measurable condition, not one row per room.

Who Signs What

Use discipline sign-off rather than one undifferentiated approval. Operations confirms missions, loads, time windows and temporary staging. Facilities owns floors, doors, lifts, power and building modifications. EHS confirms that the evidence has entered the site-specific risk process. IT/OT owns network, identity, segmentation and industrial wireless work. The integrator verifies interface assumptions and integration dependencies. The robot supplier confirms configuration-specific application limits. Procurement controls the assumptions included in the quotation. No party should sign outside its competence.

The resulting mobile robot deployment checklist is therefore not a procurement tick box. It is a controlled bridge between site evidence, technical selection, remediation funding and later validation.

Focused FAQ

When Should the Site Survey Happen?

Perform the first decision-grade survey before the final robot and integrator selection, after candidate missions and load families are defined. Revisit candidate-specific limits during detailed design, then recheck every remediated or changed condition before commissioning. A sales walk-through can precede this work, but it should not substitute for it.

Can a CAD Drawing Replace On-Site Measurement?

No. CAD is essential for route control and geometry, but it rarely captures temporary staging, damaged joints, real carrier variation, door timing, reflective materials, wireless behavior or shift-dependent traffic. Reconcile the controlled drawing with field measurements and record discrepancies.

Is There One Standard Minimum Aisle Width for Every AMR?

No. Required space depends on the configured robot, load and top module; the maneuver; protective-field behavior; localization tolerance; traffic policy; obstacles; and recovery method. Use supplier-validated configuration data and challenge tight maneuvers with representative loads.

What Floor-Flatness Number Should Procurement Specify?

Specify a measurement method and candidate-configuration limit, not an unsupported universal number. Wheel design, ground clearance, suspension, speed, payload stability and local discontinuities all influence suitability. Retain the source document and revision with each result.

Should Wireless Testing Wait Until a Robot Arrives?

No. Pre-selection measurements can expose coverage, interference, roaming and network-policy risks early. Candidate-specific validation is still required because antenna placement, radio hardware, software behavior and message criticality vary. The early test reduces uncertainty; it does not certify the final configuration.

How Long Should Mixed-Traffic Observation Last?

There is no meaningful universal duration. Observe every representative operating state that can change the mission: production peaks, replenishment, breaks, shift changes, sanitation and maintenance. Continue until the survey captures stable patterns and credible extremes, then record what was not observed.

Who Owns Infrastructure Remediation?

The owner should be the function with authority and competence over the asset—commonly Facilities for floors and utilities, IT/OT for networks, Operations for staging and standard work, and Engineering for stations. The survey must name both the action owner and the person authorized to accept the recheck.

Can an Amber Finding Remain Open When the Purchase Order Is Issued?

Only if the organization explicitly accepts the dependency and records the remedy, owner, funded cost, due date, consequence and mandatory recheck. Any finding that can change the selected configuration, route feasibility or safety concept should remain a gate, not a soft action item.

How Often Should the Facility Be Resurveyed?

Use event-based triggers plus a planned review. Resurvey affected areas after route resurfacing, layout or station changes, load-envelope changes, access-point replacement, charger changes or new traffic flows. The report should state the baseline date, configurations and future-state assumptions so users know when it no longer applies.

What Proves That a Survey Is Complete?

Completion means every scoped mission and representative time state has traceable evidence; each requirement has a source; every red or amber finding has a decision and owner; unknowns remain visible; remediation has a recheck method; and the responsible disciplines have signed their portions. A high percentage of checked boxes alone proves nothing.

Do Not Let the Robot Quote Arrive Before the Site Facts

The strongest buying sequence is evidence first, configuration second and price comparison third. A disciplined survey makes the building visible to the selection process: not as a clean layout, but as a changing production environment with measurable geometry, utilities, interfaces, radio behavior, traffic and ownership.

That evidence changes the commercial conversation. Suppliers respond to the same mission and site baseline. Remediation appears in the budget before approval. Red conditions stop unsuitable configurations. Amber conditions acquire owners and rechecks. Green conditions retain their proof. The result is not a promise that deployment will be effortless; it is a defensible decision about what must be true for the selected system to work.

#AMRSiteSurvey #AMRSiteSurveyChecklist #AMRInfrastructure #AGVFloorRequirements #AMRAisleWidth #AMRFloorFlatness #AMRWirelessSurvey #AMRChargingStation #MobileRobotDeployment #AMRFacilityAssessment #FactoryAutomation #Intralogistics