Industrial Robots
First cobots fail in the same few ways: the task was vague, the gripper was forgotten, the cycle time was fantasy, or nobody owned safety. The brands on a shortlist — KUKA, ABB, FANUC and others — will not rescue a station that was never defined.
This is not a rematch of cobot versus fenced industrial arm. If you still need that split, read which one fits the line. This is the checklist for a shop that has picked “start small, beside a person, one job”.
ⓘ Quick Answer
Automate one stable, repetitive station with a known part, a measured takt time, and a payload that includes gripper and fixture. Budget integration, guarding if the tool is hazardous, training, and service in your country. Do not buy an arm because a competitor posted a video. Prove the cell on that one job before you talk about a line.
Pick a boring first job
Good first jobs are machine tending, simple pick-and-place, or palletising of a consistent pack. Bad first jobs are “whatever is behind this week” and anything that requires human judgement every cycle. If the part changes shape monthly, you are buying reprogramming as a lifestyle.
The seven checks
- Cycle time is measured, not hoped. Time the manual station. If a cobot cannot hit takt even in a vendor’s optimistic demo, stop. Speed is the usual reason to go fenced and traditional instead.
- Payload includes the gripper. Add tool, part, and any offset. Undersizing here is the most expensive amateur mistake.
- Reach and mounting are drawn, not guessed. Include cable dress, the fence you might still need, and operator access.
- The application gets a risk assessment. A cobot is not a free pass. Sharp tools, heavy parts, and pinch points still need controls.
- Someone will program and maintain it. Name them. A robot with no internal owner becomes a tarpaulin.
- Integrator and spare parts exist near you. Global brands help only if a local partner will show up. KUKA, ABB, and FANUC all sell through networks — quote the network, not the logo.
- Success is one metric. Units per hour, scrap, or overtime on that station. “Innovation” is not a metric.
💡 What to Check
Compare complete cell quotes. Gripper, vision, fixturing, safety, programming, and training often exceed the arm. A cheap cobot with a vague integrator is not cheap.
Safety is the application, not the brochure
Force-limited arms still throw parts, still hold knives, still surprise people who walk behind them. Do the assessment. Train the people who share the space. Write what happens when it stops mid-cycle. If your culture is “we will be careful”, you are not ready.
How to use a brand shortlist
KUKA, ABB, and FANUC are mature industrial portfolios. Your first cobot may be from those families or from a specialist cobot line the same groups support. Ask for a similar cell in your region, not a flagship plant tour. Ask who services it in your language and your voltage.
If cycle time and payload push you out of collaborative speeds, that is a successful check — it means you should be looking at a traditional cell, not forcing a cobot to pretend.
Pilot like grown-ups
One station, written success measure, time-boxed. Do not retask the arm every week “to get value”. You will never finish commissioning. After it hits the metric for a month, then consider a second job with a tool changer you have actually priced.
Buyer checklist
Before you issue a purchase order
- Is the part stable enough for three months of the same motion?
- Have we timed takt and weighed payload with the gripper?
- Is there a risk assessment owner and a service number that answers?
- Does the quote include integration, not only the robot?
- What happens to production when the cobot is down?
Final Thoughts
A first cobot is a process project wearing an arm. People, parts, and a written takt time matter more than which logo is on the controller. If two vendors both pass the seven checks, pick the partner who has already commissioned a similar gripper on a similar part in your region. A factory tour of a different industry will not teach you your changeover. Compare industrial platforms on Robot.help, then make the station boring enough to succeed.
Write down what you will not automate yet. That list is as useful as the station you chose. It stops the cobot from becoming a roaming intern that never finishes a job.
Frequently Asked Questions
Do we still need an integrator for a cobot?
Usually yes, unless you already have people who have commissioned cells. The easy programming demo is not the gripper, the fixture, or the safety validation.
Can a cobot run with no cage?
Sometimes, after a risk assessment of the whole application. Never assume the word collaborative means unguarded with any tool.
Which brand is best for a first cell?
The one with a competent local partner for your application. Portfolio width matters less than who will stand on your floor.
Should we start with used equipment?
Only with known history, support, and a spare-parts path. A cheap unknown arm with no integrator is a training accident.
How long until we see a return?
It depends on labour recovered on that one station, not on a factory-wide story. If you cannot sketch it on a page, you are not ready to buy.