How to Validate AMR Stopping Distance on Real Floors

What is the right acceptance test on day one?

Validate an autonomous mobile robot by measuring how it actually stops on the exact floor, route, payload, and speed it will use in production. A sound site acceptance test does not ask whether the robot can stop in theory. It asks whether this unit, on this floor, in this state, stops inside the safeguarded distance you approved for go-live.

According to OSHA's robotics technical manual, site acceptance testing should confirm that equipment performs as expected with site utilities, interfaces, and environmental characteristics before startup. OSHA also says stopping ability, safety distances, and safety function settings should be checked again through periodic performance testing, which is the right frame for AMR stopping distance validation.

That discipline is not paperwork for its own sake. In the Bureau of Labor Statistics 2023-24 release, falls, slips, and trips accounted for 479,480 private-industry cases involving days away from work, restricted activity, or transfer, and the median days away from work for the days-away subset was 18. If your AMR needs more floor than your risk assessment assumed, the exposure is operational, not academic.

Why do real floors rewrite the brochure number?

Stopping performance lives at the tire or wheel contact patch, and that makes the floor part of the machine. Polished concrete, sealed concrete, epoxy, VCT, quarry tile, and worn coatings can all produce different traction even when they look similar to the eye. Joints, dust, patch repairs, and areas that stay damp near dock doors or scrub sinks can turn one route into several distinct braking environments.

According to the Tile Council of North America, ANSI A326.3 can be measured in the lab and in the field, and its current product-use categories include a wet DCOF threshold of 0.42 for an Interior, Wet classification. That is useful context, but it is not a robot pass line. The same TCNA guidance says DCOF is not the same thing as slip resistance and should not be the only factor in deciding fitness for a specific application.

How much do speed, payload, and load position matter?

The short version is: a lot. The Federal Highway Administration notes that when speed doubles, braking distance grows fourfold when braking conditions are otherwise unchanged. An AMR moving at warehouse pace is not a tractor-trailer, but the same physical trend applies. Small speed increases can expand stop distance faster than operators expect, which is why the test has to be run at every approved speed band, not just a comfortable middle setting.

Payload matters for the same reason momentum matters, and load placement matters because braking is not only about weight. OSHA's powered industrial truck guidance says load weight, size, position, and weight distribution affect stability, and it warns that dynamic forces from motion and braking shift the combined center of gravity. For AMRs, that means you validate with the heaviest real payload and the most forward or awkward load center the application allows, not an empty deck and a neat demo carton.

What does contamination do to stopping performance?

Contamination changes the floor faster than any other variable you do not formally control. OSHA's walking-working-surfaces rule requires floors to be kept clean and, to the extent feasible, dry, and maintained free of hazards such as spills. That matters for people and it matters for robots, because less traction means longer stops and more variability from run to run.

TCNA's current A326.3 guidance says contaminants such as dirt, water, soap, oil, and grease change DCOF. NIOSH found just how much traction conditions matter in the human case: in a field study covering about 17,000 food service workers across 226 school districts, highly rated slip-resistant shoes were linked to a 67 percent reduction in slip injury claims on wet or greasy floors. That study was about people, not AMR wheels, but the operating lesson is the same. Contamination deserves its own stopping test, not a footnote under clean-floor data.

A repeatable floor test protocol

Keep the method simple enough that operations can rerun it after a floor recoat or route change. According to NIST, repeatable robot test methods work because they use a defined apparatus, a defined procedure, and a quantitative metric, and repeated trials are used to establish confidence. For AMR stopping distance, you want the same discipline applied to your own site acceptance packet.

Start with a witnessed baseline on the production route, then expand the matrix only where the floor or task truly changes. You do not need to test every square foot. You do need to test every materially different combination of route surface, speed setting, payload class, and contamination state that the robot is approved to see.

• Freeze the test matrix before the first run. Identify the route segment, floor type, approved speed setting, payload case, and floor state being tested.
• Mark one trigger point and one stop endpoint. Measure from the same command or e-stop trigger every time, and define full stop the same way in the work instruction.
• Run a clean and dry condition first, because that becomes your service baseline. Then run the worst credible contamination state for that area, such as water at a dock threshold or detergent residue near washdown.
• Use repeated trials instead of one lucky stop. NIST notes robot trials are commonly run to 10, 20, or 30 repetitions depending on failures, so 10 runs per condition is a practical minimum starting point.
• Record the longest stop, the shortest stop, and the spread. Do not rely on an average alone, because the long stop is what collides with your safety distance.
• Have the integrator perform the test and the user witness it, matching OSHA's site acceptance expectation.

What should count as pass or fail?

There is no single OSHA stopping-distance number that fits every AMR application, and OSHA explicitly says safe speed depends on the vehicle, the load, stopping distance, operating surface conditions, and pedestrian traffic. So your acceptance limit has to be application-specific. The right limit is the one that keeps the robot inside the safeguarded stopping zone your risk assessment approved on that floor, at that speed, with that payload.

The operational recommendation I use is simple. Approve a separate stop limit for each tested condition, set it from the longest witnessed stop in that condition, and carry the measurement uncertainty of your tape or laser method into the record. Then treat any stop beyond that approved line, or any stop that breaches the safeguarded distance, as a fail that blocks go-live or triggers retest after go-live.

• Do not average unlike conditions together. Clean epoxy at low speed is not the same acceptance case as damp concrete at max speed with a loaded deck.
• Set separate approved stop limits by route condition, speed band, and payload class.
• Keep the approved line inside the safety distance already established in the risk assessment and safeguarding layout.
• Treat any overrun past the approved line as a failure, even if the average still looks acceptable.

What belongs in the record package?

OSHA says maintaining records of testing performed and the results is an effective way to track robotic system safety, and it specifically points users to monitoring safety-parameter checksums to see whether settings changed since the last inspection. That makes the stopping-distance record a control document, not a ceremonial attachment. If the checksum, speed cap, route map, or approved payload changes, the old stop data may no longer describe the live system.

This is one place a full-service, vendor-neutral robot integrator earns its keep. Service Robot Co. can write one site acceptance packet across manufacturers, tie it to site assessment mapping and robot deployment and integration, then hand operations a record set that still works months later when service, training refreshers, or a route edit happens. One partner, one method, and one audit trail is much easier to defend than a stack of mixed vendor forms.

• Robot ID, software version, and safety-parameter checksum.
• Route segment, floor material, floor finish, and photos of the test area.
• Test date, witnesses, and the exact work instruction used.
• Payload description, payload mass, and load position or envelope.
• Approved speed setting and stop trigger used for the run.
• All measured stop distances for the repeated trials, not just a summary value.
• Pass or fail decision, signoff, and the next scheduled review or retest condition.

When do you retest?

OSHA's robotics guidance says performance testing should verify that conditions of use are unchanged from the original installation and that documentation should be reviewed if changes are made to the robot application. In plain operating language, that means you retest whenever the floor, the robot, the task, or the safety settings are no longer the same ones you accepted.

Retesting is not just for obvious incidents. Quiet drift is the bigger risk. A new floor sealer, a wheel compound change, a higher speed cap during peak season, or a payload that hangs farther forward can all lengthen the stop before anyone notices it in production.

• After floor resurfacing, recoating, polishing, patching, or any change in routine cleaning chemistry.
• After a route edit, a map remaster, a speed-limit change, or a safety-parameter checksum change.
• After a payload change that increases mass, shifts the load center, or changes how the load overhangs the chassis.
• After wheel or tire replacement, unusual wear, or recurring contamination in the route.
• After any collision, near miss, unexpected overrun, or operator report that stopping feels different.
• At the periodic review interval your safety program assigns, even if nothing obvious has changed.