Robot cells live at the intersection of three regulatory worlds. OSHA writes the rules that get cited on inspection. ANSI publishes what the US robotics industry agrees is good practice. ISO publishes what the international community agrees is good practice. They overlap heavily. They are not identical. And in a West Michigan plant getting an unannounced visit from an OSHA compliance officer, the difference between aligned and not aligned shows up fast.
This post is the engineer's view of how the three standards actually fit together on a real industrial robot cell, what each one demands, where most manufacturers fall short, and the documentation stack that holds up to scrutiny.
| Standard | Status | Scope | Latest revision |
|---|---|---|---|
| OSHA 1910.147 | US federal regulation, legally enforceable | Energy control for the servicing and maintenance of machines and equipment | 1989 with 2008 amendments; current |
| ANSI/RIA R15.06 | US consensus standard, referenced by OSHA | Industrial robot safety design, integration, installation, and use in the US | R15.06-2025 (replaces R15.06-2012) |
| ISO 10218-1 | International consensus standard | Safety requirements for industrial robots themselves (the manipulator) | ISO 10218-1:2025 |
| ISO 10218-2 | International consensus standard | Safety requirements for industrial robot system integration and the robot cell | ISO 10218-2:2025 |
| ISO/TS 15066 | International technical specification | Collaborative robots: power and force limited operation, biomechanical limits | ISO/TS 15066:2016 |
OSHA cites you. ANSI and ISO show what the engineering profession considers acceptable. In a real OSHA inspection of a Grand Rapids automotive supplier or food processor cell, the inspector references both: 1910.147 for the citation, R15.06 and ISO 10218 to define what the corrective action should look like.
1910.147 (the Control of Hazardous Energy or LOTO standard) does not mention robots once. It applies to "machines and equipment" generally. When that equipment is a robot cell, the standard's six core requirements still apply, and inspectors enforce them aggressively because robot cells are higher risk than most general machinery.
For an inspection of a multi cell automotive welding line in Grand Rapids or a packaging line in Holland, the OSHA officer wants to see all six in writing within 15 minutes of asking. If any one is missing, citations follow. For deeper detail on writing compliant procedures, see our robot cell LOTO procedures service page.
R15.06-2025 is the current US robot safety standard. It harmonizes with ISO 10218-1:2025 and ISO 10218-2:2025 with limited US specific edits. Where 1910.147 is procedure focused, R15.06 is design and integration focused.
An OSHA inspector who sees a cell with only a perimeter fence, a pushbutton e stop, and a lockable disconnect will ask whether the integrator did a R15.06 compliant risk assessment. If the answer is no, the cell is operating below industry standard. That can become a General Duty Clause citation even if 1910.147 was technically met.
The 2025 revisions of ISO 10218-1 and ISO 10218-2 are the most current global view. R15.06-2025 adopts most of it. The pieces that matter most in a US plant where R15.06 leaves room for interpretation:
Collaborative robots are now common in West Michigan plants. UR robots in Grand Rapids assembly lines, FANUC CRX in Holland packaging cells, ABB GoFa in Kalamazoo machine tending. ISO 10218 only briefly addresses collaborative operation. ISO/TS 15066:2016 fills the gap.
The document specifies four collaborative operation methods: safety rated monitored stop, hand guiding, speed and separation monitoring, and power and force limited operation. The last one is what most cobot deployments rely on. ISO/TS 15066 publishes biomechanical limit values (peak transient force, quasi static pressure, and energy limits) per body region. Validating a cobot application against these limits is the difference between a real collaborative cell and one that is just a marketed claim.
If your facility runs power and force limited cobot applications and you cannot point to a written 15066 validation, the cell is operating outside accepted practice. We see this on roughly 60 percent of the cobot installations we audit on first visit.
Take a typical example: a FANUC R-2000iC industrial robot in a Grand Rapids tier 1 automotive supplier, running spot welding on body panels, fenced cell with a light curtain entry, single pendant teach mode. Here is how the three standards stack:
| Cell element | What 1910.147 requires | What R15.06 / ISO 10218 add |
|---|---|---|
| Energy isolation | Lockable disconnects on each energy source; written procedure; verification step | Disconnect placement and labeling; lockout device standardization across the integrator's installs |
| Light curtain entry | Light curtain itself does not satisfy 1910.147 isolation; LOTO still required for service | PLd or PLe rated light curtain on Cat 3 circuit; muting and override rules |
| Pendant teach | Not directly addressed | Reduced speed, enabling device, sole control, and span of control rules |
| Stored energy in welder caps | Must be addressed in the LOTO procedure; release and verification before work | Bleed circuits and discharge time specifications |
| End of arm tooling change | LOTO required if power/pneumatic isolation is needed | Tool change procedure rated to the same safety level as the cell circuit |
| Recovery from fault | If recovery requires entry, full LOTO; minor adjustment exception is narrow | Recovery mode with reduced speed, enabling device, and clear restart conditions |
| Annual review | 1910.147(c)(6) inspection of the procedure | Periodic safety circuit integrity test per ISO 13849 / IEC 62061 |
Notice how often "1910.147 does not address it but R15.06/ISO 10218 do." That is where most enforcement gap shows up. An OSHA inspector reads R15.06 too, and the General Duty Clause covers what the specific standard misses.
From roughly 40 robot cell audits across West Michigan in the past 12 months, the same five gaps appear over and over.
"FANUC R-2000 LOTO Procedure" hanging on the wall. The specific cell has a different conveyor, different fixtures, different pneumatic bleed locations than the generic procedure describes. Fails 1910.147(c)(4) machine specificity.
The integrator did one verbally during installation. Nothing was written down. R15.06 and ISO 10218-2 both require a documented task based risk assessment that gets reviewed when changes occur. Without one, the safeguarding selections cannot be defended.
1910.147(c)(6) requires an annual review of every procedure. The records exist but stop in 2023. Inspector sees this within five minutes of pulling the file.
Servo drives store energy in DC bus capacitors that take 3 to 7 minutes to bleed. Hydraulic accumulators hold pressure indefinitely. Welding caps hold significant charge. The procedure says nothing about waiting, bleeding, or verifying. 1910.147(d)(5) violation.
Universal Robots installed two years ago. Programmer turned on PFL mode. Nobody validated the application against ISO/TS 15066 limits for the actual workpiece, end of arm tool weight, and operator interaction zones. Liability and OSHA risk both real.
Each of these has the same root cause: the cell was installed without standards driven documentation and nobody owned the gap after the integrator left. Our robotics gap analysis service and annual LOTO audit exist specifically to find and close these.
A West Michigan robot cell that satisfies all three standards has the following documentation on file, accessible at the cell or in a controlled binder near it:
That stack is what a senior OSHA compliance officer expects to see on a planned inspection of a robotic cell, and it is what produces a no citation outcome on a complaint based or unprogrammed inspection too.
Most West Michigan manufacturers have parts of this stack and are missing pieces. The gaps are usually not visible until either an inspector finds them or an incident exposes them. A no obligation gap assessment walks the cells, reviews the documentation, and produces a prioritized closure plan with effort estimates.
ANSI/RIA R15.06-2012 was a national adoption of ISO 10218-1 and ISO 10218-2 with US specific edits. The current revision is ANSI/RIA R15.06-2025, which aligns more closely with ISO 10218-1:2025 and ISO 10218-2:2025. They cover the same subject (industrial robot safety requirements for the robot itself and for integration), but the documents are not byte for byte identical, and US enforcement under OSHA references ANSI by adoption.
Yes for almost every cell. 1910.147(c)(4) requires documented machine specific energy control procedures unless a single very narrow exception applies (single energy source, no stored or residual energy, no reset required, no possibility of multiple employees, and several other criteria). Most robot cells have at least pneumatic plus electrical energy, so the exception does not apply, and a written cell specific procedure is required.
OSHA 1910.147 is the legally enforceable energy control standard in the US. ANSI R15.06 is a consensus standard for robot specific safety design, integration, and use. OSHA inspectors cite 1910.147 for procedure or hardware violations, but they reference R15.06 to define what good practice looks like for robotic systems specifically. Aligning to R15.06 makes 1910.147 compliance easier and reduces general duty clause exposure.
ISO/TS 15066:2016 specifies safety requirements for collaborative robots, including biomechanical limit values for power and force limited operation. It supplements ISO 10218-1 and ISO 10218-2, which only briefly address collaborative operation. If your facility runs cobots (UR, FANUC CRX, ABB GoFa, KUKA LBR, Yaskawa HC) for power and force limited tasks alongside humans, ISO/TS 15066 is the document you need to validate the application against.
Five recurring patterns: missing or generic written procedures (1910.147(c)(4)), inadequate periodic inspection records (1910.147(c)(6)), insufficient employee training and authorization documentation (1910.147(c)(7)), no formal verification of de energization before work begins (1910.147(d)(6)), and stored energy not addressed in the procedure for servo drives, hydraulics, or pneumatics (1910.147(d)(5)). All five typically appear together when one is found.
Yes, in almost all cases. 1910.147(c)(4) requires the procedure to be machine specific. Identical cells (same robot model, same end of arm tooling, same fixtures, same energy isolation locations) can sometimes share a procedure with the cell IDs listed in scope. Cells that look similar but have different conveyors, different safety circuits, or different energy isolations need separate procedures. When in doubt, write separate.