Robotics — Good-to-Know Rules of Thumb

Robotics — Good-to-Know Rules of Thumb

Robotics has no practitioner license — no state, anywhere in the US, licenses robot engineers or integrators. What functions as law is the safety-standards stack (ISO 10218 / ANSI R15.06 / ISO 13849), enforced in the US through OSHA's General Duty Clause and in the EU through the Machinery Regulation. Careers run on three currencies: vendor training (whose robots you can run), safety competence (whose cells won't hurt anyone), and systems integration judgment (whose cells actually produce). This guide maps all three, plus the standards and regulations you'll be held to.

1. Vendor training and credentials (the practitioner layer)

There is no vendor-neutral robotics license. In practice, employers hire for their robot brand, and the vendor academies are the credentials that matter:

• FANUC Certified Robot Operator / Certified Technician (FANUC America, developed with Nocti Business Solutions) — the closest thing to a national operator credential: two Operator levels and two Technician levels, built after ~18 months of industry collaboration. FANUC is the largest industrial-robot maker in the US, so these carry the most weight in FANUC-equipped plants (automotive, general industry). Covers operations, integrated vision, and programming. Renewal: not published on verified sources.
• FANUC CERT (Certified Education Robot Training) — institution-level, not individual: FANUC certifies instructors at high schools, community colleges, and universities (which purchase the CERT tooling package — robot + vision + PLC + ROBOGUIDE) to train students on FANUC robots. If you learned robots at a CERT school, this is why.
• KUKA College — KUKA's in-house training arm: operator/programmer/technician courses (e.g., Robot Programming 1 — KRL programming, subprograms, PLC I/O simulation). Issues certificates of completion, not third-party credentials; no renewal (training certificates don't expire). High employer value in KUKA shops.
• ABB University (ABB Robotics Training) — ABB's equivalent: operator/programmer courses with completion certificates. Same standing as KUKA College — employer-valued vendor training, not a portable credential.
• Yaskawa Academy / Yaskawa MERIT (Motoman Endorsed Robotics Instructor Training) — operator/programmer courses plus MERIT, which prepares educational institutions to deliver Yaskawa-certified curriculum (e.g., YRC1000 basic programming) through their own instructors. Strong in Yaskawa/Motoman shops (welding, material handling).

How to use them: match the academy to the installed base you're targeting. None of these is portable across brands the way a PE license is portable across states — a FANUC Technician cert doesn't make you a KUKA programmer, and employers know it. The integrator-level credential that is brand-neutral is the A3 RICP (organization-level) — see the RICP study guide.

2. The ROS fact (memorize this)

No official ROS or ROS 2 certification exists. Open Robotics, the OSRF, and the Open Source Robotics Alliance issue no certification — ROS is an open-source framework maintained as a community asset. Any "ROS Certified Developer" or similar offering is a third-party training vendor's credential: unofficial, unrecognized by the ROS maintainers, and worth exactly what that vendor's reputation is worth. When a job posting asks for ROS, it wants demonstrated project work (navigation stacks, MoveIt, ros2_control, real hardware), not a certificate.

3. Standards mastery (the real "license" in robotics)

What each standard demands of the integrator (checklist form)

Exams and audits don't ask "what is ISO 10218" — they ask what you must do. Per standard:

• ISO 12100: produce a documented risk assessment covering the whole lifecycle (transport, installation, commissioning, operation, maintenance, decommissioning, reasonably foreseeable misuse). Apply the hierarchy in order and document residual risk with a justification for acceptance.
• ISO 10218-2:2025: define the robot system's limits (space, use, time); perform the cell risk assessment; implement safeguarding (fences, interlocked guards, presence-sensing devices); validate every safety function; supply the information-for-use package (manuals, markings, training requirements). The integrator owns the system, not just the arm.
• ISO 13849-1: for each safety function, determine the required PLr from the risk assessment, design the safety-related control system to achieve it (category, MTTFd, DCavg, CCF), and validate — calculation plus functional testing. Keep the SISTEMA (or equivalent) calculation files; auditors ask for them.
• IEC 62061 / 61508: where the SIL route is used instead of PL, assign SIL targets from the risk assessment, manage the safety lifecycle (plan → design → verify → validate → modify), and control systematic failures (competence, procedures, change control).
• ANSI/A3 R15.06-2025 Part 3: the end-user's duties — but the integrator's handover package must enable them (residual-risk information, maintenance safety procedures, lockout points).
• R15.08: for mobile robots — define the operating environment, validate navigation safety (detection of persons, speed control in shared zones), manage fleet/traffic rules, and handle the handover between the AMR and the fixed cell (the interface is where incidents happen).

The four collaborative modes (exam-grade detail)

ISO/TS 15066 defined them; ISO 10218:2025 carries them forward. Know all four and when each applies:

1. Safety-rated monitored stop — the robot stops when a person enters the collaborative workspace and may resume automatically when they leave. The robot is not collaborating while moving; it's time-sharing the space.
2. Hand guiding — the operator moves the robot by hand (enabling device + reduced speed, typically ≤ 250 mm/s) for teaching or guiding tasks like hand-guided assembly assist.
3. Speed and separation monitoring — robot and operator move simultaneously in a shared space; the system continuously maintains a computed protective separation distance and slows/stops as it shrinks. Needs reliable person-detection (scanners, 3D cameras) with validated response times.
4. Power and force limiting — contact is allowed by design: the robot's force, pressure, speed, and energy are limited so transient and quasi-static contact stays below biomechanical thresholds (the 29-body-area tables). This is the mode people mean by "cobot" — and it requires validating the application (tooling edges, workpiece hazards), not just buying a "collaborative" arm.

The exam trap: a "collaborative robot" running mode 1 behind a scanner is not doing power-and-force-limiting — the mode and its validated limits define safety, not the robot's marketing category.

PL vs. SIL (when each appears)

Rough correspondence used in practice: PL d ≈ SIL 2, PL e ≈ SIL 3. The audit doesn't demand you convert — it demands you pick one route consistently and validate against it. Mixing PL and SIL claims for the same safety function is a finding.

Vendor academy comparison (one table)

Rule: none is portable across brands. The brand-neutral signals are RICP (organization) and TÜV Functional Safety Engineer (individual, safety-specific).

ISO 10218-1:2025 and ISO 10218-2:2025 (ISO TC 299)

The international robot-safety core, revised February 2025 — first revision since 2011. Part 1 covers robot manufacturers (20+ required safety functions, new cybersecurity requirements); Part 2 covers integrators and robot systems (it grew from 72 to 223 pages) — Part 2 is the standard an integrator is audited against. The 2025 editions absorbed the collaborative-robot content of ISO/TS 15066, including the power-and-force-limiting biomechanical limits, and added cybersecurity requirements for safety-related control systems. Know: Part 2 = your standard; 2025 = the current edition; cybersecurity = now in scope.

ISO/TS 15066:2016 — collaborative robots

The dedicated cobot technical specification: four collaborative operating modes (safety-rated monitored stop, hand guiding, speed and separation monitoring, power and force limiting) and force/pressure limits across 29 body areas in 12 body regions. Status trap: its content was absorbed into ISO 10218:2025, but it was not withdrawn — ISO still lists it as published (confirmed 2022; a successor, ISO/AWI 15066-1, is in development). Rule: existing installations validated under it remain valid; new installations cite ISO 10218:2025.

ANSI/A3 R15.06-2025 (US national adoption)

Published 2025 (403 pages total), replacing ANSI/RIA R15.06-2012. Parts 1–2 are technically equivalent to ISO 10218-1/-2:2025; Part 3 (end-user use of robot cells) has no ISO equivalent. This is the primary US robot-safety standard — and the document OSHA inspectors treat as recognized industry practice.

ANSI/RIA R15.08 — industrial mobile robots

Covers manufacturers and integrators of AGVs and AMRs, with formal definitions distinguishing the two. (Edition/part details were thin in the catalogue — verify current parts at automate.org; a 2024 update was reported by trade sources but not confirmed.) Know it exists and when it applies: the moment the robot drives itself through the facility, R15.08 joins 10218 in the compliance stack.

ISO 13482:2014 — personal care robots

Safety requirements for service robots that interact with people (mobility aids, domestic robots). A revision is reported underway — treat 2014 as current until verified otherwise. Relevant when the robot leaves the factory floor for human environments.

IEC 61508:2010 — functional safety (umbrella)

The E/E/PE functional-safety foundation: safety integrity levels (SIL 1–4), lifecycle, systematic vs. random failures. Robot safety controllers are certified against it (e.g., SIL 3-rated safety PLCs). Sector standards IEC 62061 and ISO 13849 derive from it — know the lineage for audit interviews.

ISO 12100:2010 — machinery risk assessment

The Type-A base standard: hazard identification → risk estimation → risk reduction hierarchy (inherently safe design → safeguarding → information for use). Every robot-cell risk assessment under ISO 10218-2 is performed per ISO 12100 — it's the methodology, not an add-on.

ISO 13849-1 — safety-related control systems

Defines Performance Levels PL a–e from architecture category (B, 1–4), MTTFd, diagnostic coverage (DCavg), and common-cause failure (CCF) measures. The integrator's validation duty: the achieved PL of each safety function must meet the required PLr from the risk assessment — component datasheets don't self-certify the system. (TÜV Rheinland's Functional Safety Engineer machinery track — see the shared guide — is the practitioner credential for this work.)

UL 4600 — autonomous products

UL's standard for evaluating the safety of autonomous products — increasingly cited by insurers and deployment partners for autonomous mobile robots. Catalogue sourcing was thin; confirm the current edition at ul.com before citing it contractually.

4. Regulations mastery

OSHA robotics enforcement (United States)

OSHA has no robot-specific standard. Enforcement runs through:

• The General Duty Clause, Section 5(a)(1) — employers must keep workplaces free of recognized hazards likely to cause death or serious physical harm. Robot hazards (struck-by, caught-between, unexpected motion) are "recognized" via ANSI/RIA R15.06, which OSHA uses as recognized industry practice in citations — R15.06 is not formally adopted by reference; it's the benchmark for what the employer should have known.
• 29 CFR 1910.212 (machine guarding) and 1910.147 (lockout/tagout) — the specific standards most robot citations attach to.
• OSHA Technical Manual, Section IV, Chapter 4 ("Industrial Robot Systems and Industrial Robot System Safety") — guidance for OSHA staff, not a rule; it doesn't create new duties.
• Scale of the problem: NIOSH research identified 61 robot-related deaths in the US from 1992–2015 — the statistic that justifies the entire safety apparatus above.

EU Machinery Regulation (EU) 2023/1230

Adopted 14 June 2023, in force 19 July 2023, applies from 20 January 2027, repealing the Machinery Directive 2006/42/EC. Robotics impacts to know cold:

• AI safety components (self-evolving machine learning in safety functions) are a named high-risk category requiring Notified Body assessment.
• Cybersecurity of safety functions is mandatory (Annex III).
• Substantial modification — physical or digital — makes the modifier the manufacturer, with full CE duties (conformity assessment, technical file, EU declaration). Integrators retrofitting or reprogramming cells for EU customers inherit manufacturer liability the moment their change is "substantial."

State AV and delivery-device rules (US)

Scope clarification: state autonomous-vehicle regulations (e.g., California DMV AV testing/deployment permits) cover on-road autonomous vehicles, not warehouse AMRs/AGVs. Many states separately permit sidewalk personal delivery devices (PDDs) under size/speed limits. Catalogue sourcing here was thin — check the state DMV's AV program and PDD statutes case by case. And the bottom line stands: no US state issues a robot-engineer license.

5. Rules of thumb

• Sizing: check wrist moment (N·m), not just payload (kg) — moment ratings fail first on long EOAT.
• Repeatability (returning to a taught point) is typically 10× better than accuracy (hitting a commanded coordinate) — teach critical points, don't trust the model.
• Singularities: never program through θ2 = 0°/180° (elbow), aligned wrists, or wrist-over-base — offset the cell instead.
• Safety validation: measure stopping time/distance on your cell; catalog values are for the quote, not the validation report.
• Separation distance: S ≈ (robot speed + human speed) × stopping time + intrusion allowance — recompute whenever speed, payload, or layout changes.
• PL/SIL: the system achieves the rating, not the components — validate every safety function against its PLr.
• Collaborative ≠ inherently safe. "Cobot" is a set of operating modes with validated limits, not a property of the arm — the application risk assessment decides.
• Cybersecurity is safety scope now (10218:2025): a networked safety controller needs a security concept.
• OEE: Availability × Performance × Quality — write the run-off criteria into the contract before FAT.
• Documentation: the integrator with dated validation reports, commented programs, and signed FAT/SAT protocols wins every dispute.

6. Veteran knowledge

• What the audit/tests don't cover: the year of preparation for RICP is mostly process archaeology — writing down the risk-assessment procedure you've been doing informally, filing the validation reports you measured but never formatted, turning "Dave knows PLCs" into a training matrix. The technical bar is real, but the documentation bar is what fails companies.
• Costly beginner mistakes: (1) Sizing the robot to the part and forgetting the EOAT — the wrist moment always bites. (2) Teaching paths at 10% speed and discovering singularities at 100%. (3) Treating the safety scanner's catalog range as the protective field — the field is computed from your stopping data. (4) Quoting cycle time from the robot's max speed instead of a simulated path with accel/decel and process time.
• When to call a specialist: functional-safety validation beyond your PL calculation comfort (TÜV or equivalent assessor); CE marking for EU shipments (Notified Body where the Machinery Regulation requires it); anything involving AI in a safety function after January 2027 (high-risk category — get the assessment scoped early); personal-injury incidents (attorney before the incident report leaves the building).
• The standard is the license. In a field with no practitioner license, "we followed ISO 10218-2:2025 and validated to the required PL" is the sentence that satisfies customers, insurers, auditors, and — via the General Duty Clause — OSHA. Learn to say it with documentation behind it.

Credential map for this discipline

Original ETS reference writing, verified against public sources on 2026-09-30. Thin sourcing is flagged inline. Not legal advice — confirm current editions with the issuing body.