By the EVST Applications Engineering Team · Last updated 1 June 2026 · Reviewed by EVST robotics integration engineering
A collaborative robot can run fenceless and safely — but only after an ISO/TS 15066 risk assessment caps its force and speed for the specific task. “Fenceless” is a property of the task, not the robot: within those limits the cobot stops on contact and shares a bench with operators; outside them it isn’t compliant. This guide explains what ISO/TS 15066 requires, the two safety layers that make it work, and what changes when you swap the part, tool or speed.
Key takeaways
- Fenceless does not mean “inherently safe” — it means the robot and this task passed a risk assessment.
- ISO/TS 15066 sets force and pressure limits by body region; the robot’s speed and power are configured from those.
- Two layers do the work: power-and-force limiting (runs within capped speed/torque) and collision detection (stops on contact).
- Change the part, end tool or takt and the assessment must be redone — never copy-pasted.
- The payoff: drop the safety cage, shrink the station, and let people and robot share one bench.
This article is for safety and QA leads, manufacturing engineers and plant owners deploying collaborative robots. It covers collaborative (fenceless) operation under ISO/TS 15066; it does not cover guarded industrial-robot cells under ISO 10218 beyond contrast.
What “fenceless” really means
A guarded industrial robot keeps people out of its envelope with a fence or light curtain (per ISO 10218). A collaborative robot instead limits how hard and fast it can move, so contact with a person stays below an injury threshold. That is the whole idea — and it is why fenceless safety is engineered and assessed, not assumed. The robot is only as safe as the limits set for the exact task it is doing.
The standard: ISO/TS 15066
ISO/TS 15066 is the technical specification for collaborative-robot operation. Written specifically for human-robot collaboration, it defines force and pressure limits by body region — a contact to the hand is allowed more than a contact to the face — and from those limits the integrator configures the robot’s allowed speed and power for the task. It sits alongside ISO 10218 (the industrial-robot safety standard); collaborative applications use both.
EVST approaches every fenceless deployment with what our engineers call the Task-Risk-First method: run the ISO/TS 15066 assessment for the specific task — part, tool, speed, contact geometry — before committing to a fenceless layout, because the layout is only valid within the limits the assessment produces.
The two layers that make it safe
- Power- and force-limiting. The robot already runs within capped speed and torque, so even uninterrupted motion stays below the injury threshold for that task.
- Collision detection. The moment contact exceeds a threshold, the robot stops — it does not keep pressing. This handles the unexpected: a hand where it shouldn’t be.
Supporting design reduces harm if contact happens at all: rounded surfaces, pinch-free joints, and limits on how sharp the end tool can be. A cobot holding a sharp blade or a hot tool is not “collaborative” just because the arm is — the task has to be assessed, tool included.
Guarded vs collaborative, side by side
| Aspect | Guarded industrial robot | Fenceless cobot |
|---|---|---|
| Standard | ISO 10218 | ISO/TS 15066 (with ISO 10218) |
| How safety is achieved | Keep people out (fence/curtain) | Limit force and speed; stop on contact |
| Footprint | Larger guarded cell | Compact shared bench |
| People nearby while moving | No | Yes, within assessed limits |
| Re-assessment trigger | Layout change | Part, tool or speed change |
What changes when the task changes
This is the most-missed point. The risk assessment is task-specific. Swap to a heavier part, a sharper tool, a faster cycle or a different contact geometry, and the force/pressure picture changes — so the assessment is redone and the limits reset. A fenceless cell that was compliant for one job is not automatically compliant for the next. EVST addresses this by tying the assessment to the task definition, so a changeover that alters tool or speed flags a re-assessment rather than silently running outside the original limits.
The payoff: smaller, more flexible stations
Done right, fenceless operation is concrete value: drop the safety fence and the station shrinks, the layout gets flexible, and people and robot share one bench — loading, assembly and inspection flow together instead of being separated by guarding. For high-mix work where stations change often, removing the cage is a real space and flexibility gain, not just a safety checkbox.
Where this applies across industries
- Automotive powertrain — operators and cobots sharing fastening and light-assembly benches.
- 3C and appliances — compact, frequently-reconfigured assembly stations.
- Precision and medical components — light, human-adjacent handling where guarding would waste space.
The same logic applies wherever a light task sits beside a person. Looking ahead, as higher-payload cobots arrive, the assessment discipline matters more, not less — bigger forces mean the ISO/TS 15066 limits do more of the safety work.
FAQ
Is a fenceless cobot actually safe to work next to? Yes, within the limits an ISO/TS 15066 risk assessment sets for that task; it runs at capped force and speed and stops on contact. Outside those limits it is not compliant.
Does ISO/TS 15066 replace ISO 10218? No — they work together. ISO 10218 covers industrial-robot safety; ISO/TS 15066 adds the collaborative-operation limits for fenceless work.
Can any cobot run without a fence? Only after a task-specific assessment, and only for tasks whose forces stay within the limits. A sharp or hot end tool, or a high-speed task, may not qualify.
What happens if someone bumps the robot? Collision detection stops it the moment contact exceeds the threshold; power-and-force limiting keeps the contact below the injury threshold in the first place.
Do I need a new assessment for every product? You need one whenever the part, tool, speed or contact geometry changes enough to alter the force/pressure picture — not for trivial changes, but never assumed unchanged.
Deploying it safely
A cobot’s safety is assessed and engineered, not assumed — it rests on an ISO/TS 15066 task assessment, power-and-force limiting, and collision detection, and it must be re-checked when the task changes. Get that right and you trade a safety cage for a compact, flexible, shared station. EVST deploys fenceless cells with the Task-Risk-First method, tying each assessment to the task so changeovers stay compliant — see our guides to cobot vs industrial robot selection, robot machine tending and flexible quick-change, or talk to EVST about a compliant human-robot station.
About the author — The EVST Applications Engineering Team designs and integrates collaborative-robot cells for manufacturers across automotive, electronics, appliance and precision-component industries, running ISO/TS 15066 task risk assessments as a standard step using the Task-Risk-First method described above. Reviewed by EVST robotics integration engineering for technical accuracy. This article is general guidance, not a substitute for a site-specific risk assessment by a qualified safety engineer. Corrections and updates: see the Last Updated date.