EU vs US: Two Ways to Certify an Autonomous Forklift

The same autonomous forklift, built to the same safety architecture, faces two entirely different doors depending on which side of the Atlantic it ships to. In the EU, a machine must prove it is safe before it enters the market. In the US, a machine is presumed safe until something goes wrong — and then the manufacturer must prove it was safe all along.

That single asymmetry explains almost everything else. It is the difference between a preventive, harmonized regime and a decentralized, liability-driven one.

The philosophical split

The two systems disagree on a foundational question: who carries the burden of proof, and when?

The EU gives you a gate you must pass through. The US gives you a liability landscape you must navigate yourself. Neither is obviously lighter — they just demand different kinds of diligence.

Conformity assessment: a mark vs a listing

The EU's conformity assessment is a ladder. For machinery that follows harmonized standards, the manufacturer self-declares (Module A) — compile the risk assessment, write the technical file, affix the CE mark. But if the vehicle's safety path uses self-evolving machine learning, it climbs to Annex I Part A, and a Notified Body becomes mandatory.

The US has no direct equivalent. There is no single authority that stamps approval on an autonomous forklift. What exists instead is the NRTL listing — third-party testing by a Nationally Recognized Testing Laboratory (UL, ETL, TÜV SÜD America) against standards like UL 583 for electric battery-powered trucks and UL 3100 for automated mobile platforms.

A nuance that trips up a lot of teams: UL 3100's scope explicitly excludes vehicles with forks — it does not cover lifting tines. An autonomous forklift is evaluated under UL 583, not UL 3100.

Functional safety: same math, different container

Here is the counterintuitive part. The US does not have its own functional-safety mathematics. When an American engineer computes a Performance Level or a SIL, they use the same IEC 61508 / IEC 62061 / ISO 13849 framework as their European counterpart.

The difference is the wrapper:

The EU's ISO 3691-4 explicitly maps required Performance Levels to each safety function, removing ambiguity. The US's R15.08 introduces its own typology — Type A (standalone platform), Type B (platform + payload), Type C (platform + robot arm) — where a Type C vehicle's arm falls back to R15.06 and PL d Category 3.

The electrical minefield: color codes and pilot lights

The most concrete day-to-day difference is in the wiring — partly the standards themselves, partly the legacy panel practice each region's technicians grew up with:

The indicator-light conflict is the most dangerous. In legacy practice, a green light means "de-energized and safe" in Europe and "running and energized" in the US. A European-wired vehicle deployed in a US facility without modification can signal the exact opposite of what an operator expects. The practical workaround that global manufacturers converge on: text-labeled digital HMIs instead of colored pilot lights.

Cost and timeline

The economics diverge sharply, and the divergence is driven by when third-party involvement becomes mandatory. (The figures below are typical ranges; they vary with the Notified Body or NRTL and the scope of the assessment.)

The striking result: the EU is the cheaper, faster market for a deterministic vehicle (self-certification costs a fraction of an NRTL listing), but the most expensive path of all is an AI-heavy EU vehicle forced through the Notified Body track. The US cost is roughly constant regardless of autonomy level — because the NRTL listing is driven by the electrical/fire/battery testing, not the navigation logic.

The market-entry decision

This cost structure produces a clean strategic fork:

flowchart TD
    A{"Vehicle autonomy level?"} -->|"High AI / self-evolving"| B["Europe first"]
    A -->|"Deterministic / standard"| C["US first"]
    B --> D["Design to ISO 3691-4<br/>Prepare for Notified Body"]
    C --> E["Leverage self-validation<br/>Rapid revenue pilots + field labels"]

• High-AI vehicles → Europe first. The EU's strict upfront requirements force a clean functional-safety and cybersecurity architecture from day one, and a vehicle certified to ISO 3691-4 with PL d transfers to the US with minimal rework.
• Deterministic AGVs → US first. Self-validation under B56.5/R15.08 means a vehicle can deploy on a US floor — with Field Evaluations running in parallel — and start generating revenue before the full NRTL listing completes.

The recommended endgame for most serious manufacturers is joint engineering: design the physical safety to ISO 3691-4 (PL d scanning), build the electrical cabinet from dual-certified (UL-listed and CE-marked) components with modular wiring, and use text-labeled digital HMIs to sidestep the color-code conflict entirely.

The honest summary

Neither system is "the hard one." They are hard in different places:

• The EU is hard up front. A clear gate, a harmonized standard, a single mark — but if your autonomy is AI-driven, that gate is guarded by a Notified Body and a 12-month review of your learning algorithm.
• The US is hard in the aftermath. No single gate to pass, but a liability landscape where the adequacy of your risk assessment is tested only after an incident — by an OSHA citation or a plaintiff's attorney.

The engineering lesson that carries across both: risk assessment first, autonomy stack second. The zones, the speed limits, the PL targets, and the hardware choices all flow from the risk assessment — whether that work is later checked by a Notified Body in Brussels or a courtroom in Texas.

This is Part 3 of a three-part series. Part 1 — Germany/EU and Part 2 — United States cover each regime in depth. This comparison draws on the Machinery Regulation (EU) 2023/1230, ISO 3691-4, ISO 13849-1, ANSI/ITSDF B56.5, ANSI/RIA R15.08, OSHA 1910.178, NFPA 79, and the UL 583 / UL 3100 listing requirements.