Industrial robot safety after the 2025 standards update
What industrial robot safety means after 2025
Industrial robot safety is no longer just a matter of placing a fence around a fast-moving arm. The 2025 update to ISO 10218 makes a clearer distinction between the robot as a machine and the complete robot application or cell. For manufacturers, integrators and plant safety teams, that distinction changes the focus of the safety review. The assessment has to cover the task, end effector, workpiece, access points, maintenance activities and reasonably foreseeable misuse.
A robot described as collaborative, enclosed, slow or modern is not automatically safe. A well-controlled robot cell is one where hazards are identified, protective measures are selected, safety functions are validated and workers understand when production safeguards are sufficient and when full energy control is required.
The scale of robot deployment makes this more than a standards issue. The International Federation of Robotics reported in its World Robotics 2025 industrial robot release that 542,000 industrial robots were installed globally in 2024 and that the operational stock reached about 4.664 million units. More robots usually mean more routine interaction, more troubleshooting and more pressure to keep automated lines running. For more workplace risk-control topics, see our industrial safety section.
Why robot hazards are often found outside normal production
Many robot safety discussions begin with the robot arm, but serious exposure often occurs when the process is no longer in its normal automatic cycle. OSHA’s robotics materials have long emphasized that many robot accidents occur during non-routine conditions such as programming, maintenance, testing, setup and adjustment. That pattern is consistent with a NIOSH analysis of U.S. robot-related workplace fatalities from 1992 through 2017.
The NIOSH study identified 41 confirmed robot-related fatalities over that 26-year period. Manufacturing accounted for 78 percent of those cases. Contact with objects and equipment was the leading event category, and maintenance of the robot was mentioned in 58.5 percent of fatalities. The study also found that 78 percent of the cases included narrative information describing direct contact between the moving robot and the worker.
These figures should not be treated as a current injury rate because they are based on fatality surveillance data and a keyword-review method. Their practical value is in showing where robot safety plans often break down. A production operator may be protected by fencing, interlocked gates and light curtains during normal operation, while a technician clearing a jam may be exposed to stored pneumatic energy, gravity-loaded tooling, unexpected restart, a suspended part, a conveyor zone or another machine inside the same cell.
The standards map that teams should review
Robot safety standards do not replace local legal obligations, but they give manufacturers, integrators and users a common technical language. The central standards change to review is the 2025 revision of the ISO 10218 series. ISO 10218-1:2025 addresses the industrial robot itself as partly completed machinery, including inherently safe design, risk reduction and information for use. ISO 10218-2:2025 addresses robot applications and robot cells, including integration, commissioning, operation, maintenance, decommissioning and disposal.
For the United States, ANSI/A3 R15.06-2025 is especially important. The ANSI Webstore describes it as a national adoption of ISO 10218-1:2025 and ISO 10218-2:2025, presented in their entirety. It updates and replaces ANSI/RIA R15.06-2012, which is being withdrawn. This does not mean every older cell must be rebuilt immediately, but new projects, major modifications and safety reviews should be compared with the newer framework.
| Safety area | Reference to review | Practical question for the project |
|---|---|---|
| Robot design | ISO 10218-1:2025 | Has the robot manufacturer provided the safety information, limits and functions needed for integration? |
| Robot cell integration | ISO 10218-2:2025 | Have hazards from the complete application, including tooling and surrounding machines, been assessed? |
| U.S. robot safety practice | ANSI/A3 R15.06-2025 | Does the project align with the current U.S. adoption of the ISO 10218 requirements? |
| Collaborative operation | ISO/TS 15066:2016 and related robot standards | Have force, pressure, speed, separation and contact conditions been validated for the real task? |
| Workplace compliance | OSHA machine guarding and lockout/tagout rules | Are employees protected from machine hazards and unexpected energization during servicing? |
OSHA’s Robotics overview states that there are no OSHA standards specific to the robotics industry. Robot cells, however, are still subject to applicable workplace safety requirements. In general industry, 29 CFR 1910.212 covers machine guarding, and 29 CFR 1910.147 covers the control of hazardous energy during servicing and maintenance. Depending on the application, electrical, welding, material handling, confined-space or process-specific rules may also apply.
Start the risk assessment with tasks, not equipment names
A useful robot risk assessment starts with tasks, not device names. The same six-axis robot can present very different risks when it is welding inside a fully enclosed cell, tending a press, palletizing near forklifts or being jogged by a programmer inside a large work envelope.
At minimum, the assessment should cover:
- Automatic production, including normal loading, unloading and part transfer.
- Manual loading or unloading where hands approach fixtures, clamps or tooling.
- Teaching, jogging, programming and recovery after faults.
- Jam clearing, sensor cleaning, tool change and quality checks.
- Preventive maintenance, repair, troubleshooting and contractor access.
- Startup, restart after interruption and shift handover.
- Cleaning, housekeeping and removal of scrap inside or near the cell.
- Interaction with conveyors, presses, turntables, welders, fixtures, grippers and automated vehicles.
Each task should identify who is exposed, how often exposure occurs, the severity of possible injury and whether the person can avoid the hazard. The team can then decide whether risk reduction should come from elimination, substitution, engineering controls, administrative controls, personal protective equipment or a combination. In robot cells, engineering controls usually carry the most weight because they do not rely solely on attention, memory or production discipline.
Safeguarding layers for industrial robot cells
Industrial robot safeguarding is strongest when several layers work together and are validated as a system. No single device should be treated as a complete safety program.
Physical barriers and controlled access
Fixed perimeter fencing, interlocked gates and locked access doors remain common because they separate workers from the robot’s restricted space during automatic operation. They must be designed around the full reach of the robot, end effector, workpiece and any auxiliary axes. Gaps, overreach, underreach and service panels can become weak points if they allow entry without stopping hazardous motion.
Presence-sensing devices
Light curtains, laser scanners, pressure-sensitive mats and vision-based safety systems can reduce risk where physical barriers would prevent necessary work. They require safety-distance calculations based on approach speed, stopping time and system response time. If the robot carries sharp, hot, heavy or unstable parts, the presence-sensing design must address the hazard created by the payload as well as the arm. See also: automation and controls.
Safety-rated control functions
Modern robot systems may include safety-rated monitored stop, safe speed, safe position, limited axis range and enabling-device functions. These controls are useful only when they are configured, tested and documented for the application. A setting stored in a controller is not a safeguard unless the complete safety function, including sensors, logic, actuators and reset behavior, has been validated.
Emergency stops and resets
Emergency stops are important, but they are not a substitute for preventing access to hazardous motion. Reset devices should be outside the hazard zone where the operator has a clear view, or the design should otherwise prevent reset while a person remains inside the cell. Unexpected restart after a gate closes or a fault clears is a common design concern and should be addressed directly.
Collaborative robots still need engineering controls
Collaborative robots are often marketed as easier to deploy, but collaborative operation is a method of risk reduction, not a blanket exemption from safeguarding. ISO/TS 15066:2016 was created to supplement the ISO 10218 framework for collaborative robot systems and work environments. It is especially relevant where people and robots intentionally share space.
The key question is not whether the robot is called a cobot. The key question is which collaborative mode is being used and whether it is safe for the actual process. Power and force limiting may be suitable for some tasks, but it can become inadequate when the robot carries a pointed tool, hot part, sharp sheet-metal edge, abrasive wheel or heavy payload. Speed and separation monitoring may work well in a controlled area, but only if detection zones, stop times and approach paths are validated. Hand-guiding can reduce some hazards during teaching, but nearby fixtures, pinch points and stored energy still need controls.
A practical collaborative robot review should include contact-force validation where applicable, assessment of body trapping points, tool and workpiece hazards, floor conditions, operator posture, reach distances and foreseeable workarounds. If the only safety argument is that the robot moves slowly, the assessment is incomplete.
A practical checklist for new and existing robot cells
For new projects, safety should be built into the purchase specification rather than added after installation. For existing cells, a review is especially useful after tooling changes, speed increases, product changes, guarding damage, repeated faults or near misses. The following checklist can help structure the discussion:
- Define the operating modes. Separate automatic production, manual intervention, teaching, maintenance and recovery modes.
- Map all hazards. Include the robot, end effector, workpiece, fixtures, conveyors, presses, welders, fluids, gravity and stored energy.
- Confirm applicable standards. Review ISO 10218-1:2025, ISO 10218-2:2025, ANSI/A3 R15.06-2025 and any application-specific machinery standards.
- Design access control. Decide where people may enter, how hazardous motion stops and how restart is prevented.
- Validate safety functions. Test interlocks, sensors, stop times, safety-rated limits, reset logic and fault behavior.
- Write task-specific procedures. Distinguish short production interventions from servicing that requires lockout/tagout.
- Train by role. Operators, maintenance technicians, programmers, contractors and supervisors need different levels of instruction.
- Review changes. Treat new grippers, higher speeds, altered paths, software changes and layout changes as possible triggers for reassessment.
Strong safety programs also track smaller warning signs: bypassed interlocks, blocked scanners, repeated nuisance trips, informal shortcuts and unclear lockout boundaries. These issues often reveal a mismatch between the designed safeguard and the way work is actually performed. Correcting that mismatch can improve both safety and uptime.
Frequently asked questions
Does OSHA have a robot-specific safety standard?
OSHA’s Robotics overview states that there are currently no OSHA standards specific to the robotics industry. However, robot applications may still be subject to OSHA requirements such as machine guarding under 29 CFR 1910.212 and lockout/tagout under 29 CFR 1910.147, along with other rules depending on the hazards present.
Is a collaborative robot safe without fencing?
Not automatically. A collaborative robot may operate safely without fencing only if the risk assessment supports that design and the selected safety measures are validated for the real task, tooling, payload, speed, contact conditions and surrounding equipment.
Should older robot cells be updated to the 2025 standards?
Older cells should be reviewed when they are modified, moved, repurposed or involved in near misses or repeated interventions. The 2025 standards provide a strong benchmark for review, but the exact action should be based on the cell’s hazards, local legal obligations and a competent risk assessment.
What is the most overlooked industrial robot safety issue?
Non-routine access is often the weak point. Clearing jams, cleaning sensors, teaching paths, recovering faults and performing maintenance can expose workers to hazards that are not present during normal automatic production. These tasks need specific controls, not just general awareness training.
