The practical point. A robot does not make a job human-centred. The design of the work does. The 2025 edition of the standard for industrial robot applications makes this explicit: it no longer uses the term “collaborative robot” and treats collaboration as a property of the whole application [1]. Plants that design the task, the cell and the skills together get safer work and better results than plants that buy a robot and then look for a place to put it.
The robot is not the unit of design
When people talk about human-centred automation, the conversation usually starts with the machine: which collaborative robot, what payload, how fast. I understand the appeal. The machine is the visible part, and it is the part with a price tag. But in many years of automation work, I have found that the hard problems usually sit somewhere else. They sit in the task, the layout, the hand-offs between people and equipment, and the skills of the people who keep it all running.
The standards bodies have reached the same conclusion. In February 2025, the International Organization for Standardization (ISO) published new editions of both parts of ISO 10218, the standard for industrial robot safety [1], [2]. Part 1 covers the robot itself and is written for robot manufacturers. Part 2 covers robot applications and robot cells and is written for system integrators [3]. The Association for Advancing Automation (A3), which announced the revision in North America, noted that it followed nearly eight years of work [3].
The most important line in Part 2 is easy to miss. The standard says that the term “collaborative robot” is not used in the document, because only the application “can be developed, verified, and validated as a collaborative application” [1]. That is exactly right. A robot that is safe beside a person while placing a light plastic part may be unsafe beside the same person while carrying a sharp metal blank. Safety belongs to the whole application: the robot, the gripper or tool on its arm, the workpiece, the task program and the people nearby.
What changed in 2025
The second edition of ISO 10218-2 cancels and replaces the 2011 edition. Its foreword lists four main changes [1]:
- An emphasis on the robot application rather than the robot system, because the application includes the workpieces, the task program and the supporting equipment
- Safety requirements for collaborative applications, formerly in the technical specification ISO/TS 15066, now inside the standard
- Clearer requirements for functional safety
- Requirements for cybersecurity, to the extent they bear on safety
ISO/TS 15066, published in 2016, gave the industry detailed guidance on collaborative operation [4]. It described four ways in which people and robots can share a workspace, and only the last of them allows contact between a moving robot and a person. For that case, the specification set limits on force and pressure for different regions of the human body [4]. Most of its requirements now sit in ISO 10218-2 itself [1].
| Method | How it protects people | Typical use |
|---|---|---|
| Safety-rated monitored stop | Robot halts while a person is in the shared space | Occasional loading |
| Hand guiding | Person moves the robot directly | Teaching, positioning |
| Speed and separation monitoring | Robot slows or stops as people approach | Shared work areas |
| Power and force limiting | Contact allowed within body limits | Light, close work |
Most robots still work apart from people
It is worth keeping the scale in view. The International Federation of Robotics (IFR) counted 4,281,585 industrial robots operating in factories worldwide in 2023, with 541,302 new installations that year [5]. Collaborative robots made up about 10% of those installations, and their installations fell for the first time, by about 2% [6]. Most robots are still traditional industrial robots, separated from people by guarding or other safeguards, in cells that people enter for loading, changeover or maintenance.
That last point deserves more attention than it gets. A study by the United States National Institute for Occupational Safety and Health (NIOSH) identified 41 robot-related workplace fatalities in the United States between 1992 and 2017 [7]. Most cases involved stationary robots (83%). Most involved a robot striking the worker while operating under its own power (78%), and many of those incidents happened while maintenance was being performed on the robot [7]. The people most exposed to a robot are often not the operators who work beside it every day. They are the maintenance crews who reach inside the cell when something has gone wrong.
Forty-one deaths over 26 years is a small number, and the industry’s safety record has been earned through careful standards work. But each of those cases was a person who went to work and did not come home. Human-centred design has to include the maintenance technician, the setter and the cleaner, not only the operator at the station.
Design the work first
The European Commission’s 2021 framing of Industry 5.0 asks that technology adapt the production process to the needs of the worker, rather than the worker continually adapting to the technology, and that workers be closely involved in the design and deployment of robotics and artificial intelligence [8]. In plant terms, I would put it simply. Before choosing a robot, describe the work.
- What does the person do now, step by step? Include the awkward reaches, the heavy lifts, the inspections done by eye and the judgement calls.
- Which steps should the person keep? Judgement, inspection of unusual parts and recovery from faults usually belong with people.
- Which steps hurt or exhaust people? Repetitive lifting, awkward postures and monotonous placing are good candidates for automation.
- Who touches the cell besides the operator? Setters, maintenance crews, quality technicians and cleaners all need safe ways in and out.
- What happens when it stops? Fault recovery is where people and machines meet most closely, and under the most time pressure.
ISO 10218-2 builds its requirements around a risk assessment of the application [1]. That assessment is far better when the people who do the work help write it.
Skills are part of the design
The skills question is not optional. The World Economic Forum’s Future of Jobs Report 2025, based on a survey of employers, found that workers can expect, on average, 39% of their existing skill sets to be transformed or become outdated between 2025 and 2030 [9]. In the United States, a 2024 study by Deloitte and The Manufacturing Institute estimated that manufacturing could need as many as 3.8 million new workers between 2024 and 2033, and that 1.9 million of those jobs could go unfilled if skills and applicant gaps are not addressed [10].
A robot cell that only one outside specialist understands is not resilient. The operators who run it, the technicians who maintain it and the engineers who change it all need training, and that training needs time on the plant schedule. The best investment is often the simplest one: let the people who will run the cell help commission it.
Practical advice for plant and operations leaders
- Start every robot project with a task analysis, not a catalogue. Write down the work, the people involved and the hazards before you look at robots.
- Specify the application, not the robot. Ask your integrator to show the risk assessment for the complete task, including the tooling and the workpiece.
- Design for maintenance access. Treat setters and maintenance technicians as primary users of the cell, with safe procedures for entry, lockout and recovery.
- Know where existing cells stand. ISO 10218-1:2025 does not apply to robots manufactured before its publication [2], but you should still know how your older cells compare on functional safety and cybersecurity.
- Budget training as part of the project. Include operator, maintenance and engineering training in the capital request, and schedule it before go-live.
Human-centred automation is not a product category. It is a way of designing work so that machines take the strain, the repetition and the danger, and people keep the judgement, the skill and the pride in a job well done.
References
- ISO 10218-2:2025, “Robotics: Safety requirements, Part 2: Industrial robot applications and robot cells,” 2nd ed., International Organization for Standardization, Geneva, Feb. 2025.
- ISO 10218-1:2025, “Robotics: Safety requirements, Part 1: Industrial robots,” 3rd ed., International Organization for Standardization, Geneva, Feb. 2025.
- Association for Advancing Automation, “Updated ISO 10218: Major advancements in industrial robot safety standards now available,” press release, Ann Arbor, MI, Feb. 17, 2025.
- ISO/TS 15066:2016, “Robots and robotic devices: Collaborative robots,” International Organization for Standardization, Geneva, 2016.
- International Federation of Robotics, “Record of 4 million robots working in factories worldwide,” press release, Sept. 24, 2024.
- International Federation of Robotics, “World Robotics 2024,” press conference presentation, Sept. 2024.
- L. A. Layne, “Robot-related fatalities at work in the United States, 1992-2017,” American Journal of Industrial Medicine, vol. 66, no. 6, pp. 454-461, 2023.
- M. Breque, L. De Nul, and A. Petridis, “Industry 5.0: Towards a sustainable, human-centric and resilient European industry,” Policy Brief, European Commission, Directorate-General for Research and Innovation, Jan. 2021.
- World Economic Forum, “The Future of Jobs Report 2025,” Geneva, Jan. 2025.
- Deloitte and The Manufacturing Institute, “Taking charge: Manufacturers support growth with active workforce strategies,” Apr. 2024.