Robot Safety Standards: ISO 10218, TS 15066 and R15.06

Limited reference overview — not technically reviewed. No qualified robotics or safety professional has reviewed this page. It is written from publicly available catalogue records and is published as a starting point for your own reading, not as safety guidance. It contains no clause numbers, no force or pressure limits, no performance levels and no safety integrity levels, and it is not a compliance document. Conformity is determined by a competent assessor working from the standards themselves.

Read this first

This page explains structure and intent. It is not a compliance document, it reproduces no requirements, and it is not a substitute for the standards themselves. No clause reference, force limit, pressure limit or separation distance appears anywhere on this page. Anyone making a safety decision needs the actual standard text, and in most cases a competent safety engineer, not a web page.

It is written because the relationship between these documents is genuinely confusing, the landscape changed significantly in 2025, and a clear map is useful before you spend money on the documents themselves.

The documents, and what each is for

ISO 12100 is the foundation. It sets out general principles for machine safety and, critically, the risk assessment and risk reduction methodology that everything else assumes you are following. If you read one document in this list, this is the one that changes how you think.
ISO 10218 is the robot-specific standard, in two parts. Part 1 addresses the robot itself, as supplied by its manufacturer. Part 2 addresses the robot system and its integration into an application — the cell, the guarding, the peripherals, the actual installed thing.
ISO/TS 15066 was a technical specification covering collaborative applications, where people and robots share space without fixed guarding. Its content has been folded into the 2025 ISO 10218 revision.
ANSI/A3 R15.06 is the United States national adoption of the ISO 10218 series. The 2025 edition replaces the 2012 edition, which is being withdrawn.
IEC 61508 and ISO 13849 govern functional safety — whether a safety function is reliable enough to be depended on. These are where performance levels and safety integrity levels come from.

The division that matters most: Part 1 is the manufacturer's problem, Part 2 is the integrator's and operator's problem. A robot can be fully compliant as a product and installed into an application that is not safe, and that distinction is where a great deal of real-world confusion and liability sits.

What the 2025 revision changed

ISO 10218 was revised in 2025, its first substantial update since 2011. The changes both sources consulted agree on:

ISO/TS 15066 was absorbed. Collaborative application requirements, including power and force limiting, now live inside the ISO 10218 series rather than in a separate technical specification.
Robot classification was introduced in Part 1, with functional safety requirements and test methods attached to the classes. Commentary describes a reduced-requirement class for robots not presenting a significant hazard and a general class covering the rest, with the latter expected to be the large majority.
Cybersecurity requirements appeared, scoped to their effect on safety rather than as general security guidance.
Functional safety requirements were clarified.
Terminology shifted toward the application. The emphasis moved from the robot as an object to the robot application as the thing being assessed. Both sources note this; they describe the specific wording differently.
ANSI/A3 R15.06-2025 followed as the US adoption. The publication month is not stated here because it was not confirmed from a primary record; take it from the A3 catalogue entry rather than from a summary.

What the primary records say

An earlier version of this page reported that two industry sources disagreed about dates and terminology, and left the disagreement standing. That was the wrong call. The catalogue records settle most of it, and publishing avoidable uncertainty about a safety standard is worse than doing the work.

ISO 10218-1:2025 is published. It cancels and replaces the second edition, ISO 10218-1:2011, which was technically revised, and it incorporates safety requirements for industrial robots used in collaborative applications.
ISO/TS 15066:2016 is still published. Its catalogue status is an International Standard to be revised, the revision opened in June 2025, and a successor numbered ISO/AWI 15066-1 is under development.
Incorporation is not withdrawal. A newer standard absorbing a topic does not by itself withdraw the older document. The two statements above are not in conflict, and a source that treats them as interchangeable is not reliable on this subject.

What this page still does not do is interpret clauses, reproduce thresholds, or tell you which edition applies to your application in your jurisdiction. Those need the documents themselves and a qualified assessor. A precise month of publication for the 2025 edition is not stated here because it was not confirmed from a primary record; take it from the catalogue entry rather than from a summary.

Checked 14 November 2026.

The four collaborative techniques, in substance

Whatever the current naming, the underlying safeguarding approaches for shared-space working are stable and worth understanding. They are not alternatives to risk assessment; they are outcomes of one.

Monitored stop or standstill. The robot holds position while a person is in the shared space, with the stop itself monitored to a required reliability, and resumes when they leave. The person and the moving machine are separated in time rather than in space.
Hand guiding. An operator moves the robot directly through a device that is itself safety-rated, under enabling controls. Motion happens only while the person is deliberately commanding it.
Speed and separation monitoring. The system tracks the distance between person and robot and reduces speed or stops as that distance closes, with the required separation depending on speeds, stopping performance and detection latency. This is where sensing quality and reaction time become safety-critical rather than performance concerns.
Power and force limiting. The robot is designed and configured so that contact, if it happens, stays within limits derived from what the human body can tolerate at the body region involved. This is the only technique that contemplates contact, and it is the most frequently misunderstood.

The most common misconception

There is no such thing as a safe robot. Safety is a property of an application, not of a product. A robot marketed as collaborative is a robot with features that can support a collaborative application, under conditions established by assessment. Fit it with a sharp tool, run it at speed near a face, or place it where someone can be trapped against a fixture, and those features do not make the installation safe.

The 2025 revision's shift in emphasis toward the application is, as much as anything, a response to exactly this misreading. Nobody should be reassured by a product label.

What this means for software

For anyone building a capability layer rather than a robot, the division of responsibility is the whole point. Safety authority stays with the machine and with the party responsible for operating it. A third-party software layer does not get to overrule an interlock, relax an envelope, or decide that a condition is acceptable.

What software can and must do is declare the conditions under which a capability must refuse to act, and refuse. A capability without a well-specified refusal behaviour is not deployable regardless of how well it performs on average, and no amount of average performance substitutes for that.

Getting the actual documents

These standards are not free. They are sold by ISO and by national standards bodies, and for an organisation actually deploying robots they are an ordinary cost of doing it properly. Official text is available from ISO and from national members; in the United States the R15.06 adoption comes through A3.

Robot safety standards