What is a collaborative robot (cobot)? A plain guide for SMEs

When most people hear the word robot, they still picture the huge arms working inside cages at car plants. In recent years, however, a different kind of robot has arrived on the shop floor: smaller, easier to program and, when set up correctly, able to share the same space with people. These are collaborative robots, or cobots for short.

In this guide we explain in plain language what a collaborative robot is, which jobs it really pays off in, how safety should be handled, and how a small or medium-sized factory can start its first project.

What is a collaborative robot?

A collaborative robot is an industrial robot designed to work alongside people in a shared workspace under defined safety conditions. Traditional industrial robots usually run at high speed and force, so they are fenced off or caged. Cobots use features such as power and force limiting and contact detection, which means they can be used without a protective cage when the risk assessment allows it.

There is an international technical specification for the safety requirements of collaborative applications: ISO/TS 15066. It describes which limits must be respected when a robot may come into contact with a person. In other words, the cobot label alone does not make a setup safe; what decides safety is how the robot is used and for which job.

Why is it on the agenda now?

According to the World Robotics 2026 report published by the International Federation of Robotics (IFR) in September 2026, the number of industrial robots operating in factories rose by 9% in 2025 to a record 5 million units. In the same year, more than 600,000 new robots were installed worldwide, and annual installations grew by 11%.

Cobots are growing too. According to the IFR 2025 press presentation, collaborative robot installations rose by 12% in 2024. The IFR 2026 report also notes that advances in artificial intelligence, machine vision and sensing are increasing robot capabilities, while easier programming and system integration are reducing deployment costs and opening automation to new customer groups.

An observation from the IFR 2025 editorial matters for smaller manufacturers as well: companies are looking less for abstract automation strategies and more for user-friendly, ready-to-use solutions that teams without deep technical backgrounds can deploy quickly. Cobots are one of the tools that answer exactly this need.

Which jobs does a cobot pay off in?

Collaborative robots create the most value in repetitive, tedious and physically tiring tasks. Common examples on the shop floor include:

• Machine tending: loading and unloading parts on a CNC machine or a press, so one operator can look after several machines at once.
• Packing and palletizing: boxing products at the end of the line and stacking boxes on pallets.
• Screwdriving and assembly: repetitive assembly steps that must be done with the same torque in the same order.
• Quality inspection: working with a camera to check parts and sort out the faulty ones.
• Gluing, dispensing and welding: tasks that need a steady speed and consistent quality.

The common thread is simple: people focus on the work that needs judgment and skill, while the robot takes over the part that repeats the same motion all day.

Which jobs are not a good fit?

Not every task is a good candidate for a cobot. Very heavy loads, lines that demand very high speed, or processes involving sharp and dangerous tools make working side by side with people difficult. In those cases a traditional industrial robot behind a safety fence may be the better choice.

Also, if the position and shape of the part change a lot every time, vision systems and extra tuning come into play. That is not impossible, but it is not an ideal first project either.

Safety: no cage, but clear rules

Safety is the most misunderstood side of cobots. People tend to assume that when the cage goes, the rules go too. The opposite is true:

1. A risk assessment is required. It covers the whole application, not just the robot: the gripper, the part being carried, the motion path and the people nearby.
2. The gripper and the part change the risk. A robot carrying a soft box is not the same as one carrying a sharp sheet metal part.
3. Speed and force limits must be set. ISO/TS 15066 provides the framework for these limits.
4. The team must be trained. Everyone working next to the robot should know what it will do and how to stop it.

If these steps are skipped, the flexibility a cobot offers can quickly turn into a risk.

How should you think about cost and payback?

Looking only at the price of the robot arm is misleading. The total cost of a cobot project includes:

• The robot arm and its controller
• The gripper, cameras and sensors
• The mounting table, part feeding and safety equipment
• Programming, commissioning and training
• Maintenance and spare parts

When you calculate payback, do not look only at labor savings. Output per shift, scrap rate, lower risk of workplace injuries and moving operators to more valuable work should all be part of the picture. The most reliable way to get these numbers is to measure your current situation on your own shop floor.

How do you start your first cobot project?

A small, measurable start always beats a big, ambitious plan. Here is the path we recommend:

1. Pick a single task. It should be repetitive, use standard parts and be easy to measure. Machine tending or packing are good candidates.
2. Measure today's baseline. Record hourly output, scrap rate and downtime for a few weeks.
3. Define success up front. Set a clear goal, such as keeping a machine running through the night shift or freeing one packing operator for other work.
4. Run a trial. Many suppliers and integrators offer trials with your real parts before you buy. Always ask for one.
5. Involve your operators. The people who will work with the robot most know best which steps are hard and which are easy.
6. Measure the results, then scale. If the first application hits its goal, move on to similar tasks with the same approach.

Frequently asked questions

Will a cobot take the operator's job? Cobots usually take over the repetitive part of a task, not the whole job. Operators can move to more valuable steps such as setup, checks and quality. Planning this shift openly with your staff makes it much easier for the project to be accepted on the floor.

Do you need a software specialist to program one? Many cobots can be programmed by guiding the arm by hand and through visual interfaces. A short training may be enough for simple tasks, while jobs that involve cameras, many product variants or talking to other machines go faster with an experienced integrator.

Can the robot be moved to another task? Yes. Thanks to their light build, cobots can often be moved to different stations and reprogrammed. A new risk assessment is needed for every new use, though.

Conclusion

Collaborative robots are making automation accessible not only to large plants but to small and medium-sized manufacturers as well. Success, however, does not come from the robot brand. It comes from choosing the right task, taking safety seriously and measuring the results.

If you would like to assess together whether there is a task on your shop floor that suits a cobot, the AtomAxis team can support you from process analysis all the way to commissioning.