Factory automation / Article

Cobot vs Industrial Robot: Which One Your Station Needs

A cobot is an arm built so it can share a workspace with a person under defined safety conditions. A conventional industrial robot is built for speed, payload and reach, and is separated from people by fencing or safety devices. Choosing between them is not a preference, it is a trade: collaborative operation buys you floor space and access, and it costs you cycle time and payload. The station decides. If a person genuinely needs to stand there, and the part is light, a cobot usually wins. If the cell will end up fenced anyway, you are paying for capability you cannot use. This article gives you the trade in numbers, four worked station examples, and an honest account of where cobots stop making sense.

The numbers at a glance

542,000
Industrial robots installed worldwide in 2024, more than double ten years earlier
~12%
Cobot share of new installations, over 64,500 units, growing 13% year on year
177
Robots per 10,000 manufacturing employees, global average density
Apr 2025
ISO 10218-1 and 10218-2:2025 in force, absorbing ISO/TS 15066

01 / The definition

Collaborative describes the application, not the arm

This is the single most useful thing to understand before you shortlist anything. A collaborative robot is an arm designed with the features that make collaborative operation possible: force sensing, rounded geometry, safety-rated control, low mass. Whether your cell is actually collaborative depends on what you bolt to the end of it and what it handles. The safety framework is explicit about this. ISO 10218-1:2025 and ISO 10218-2:2025 came into force on 1 April 2025, the first update to the core robot safety standard since 2011, and folded the collaborative-application requirements that ISO/TS 15066:2016 previously held separately, including power and force limiting, directly into the revised standard.

Those requirements recognise four ways of working safely alongside a robot: power and force limiting, speed and separation monitoring, hand guiding, and safety-rated monitored stop. Most real cells combine them. A cell might use speed and separation monitoring while the operator is in the general zone, then drop into power and force limiting when hands come inside the reach of the tool.

A conventional industrial robot uses none of that. It is fast, strong and blind to your presence, so it is separated from people by physical guarding, light curtains, scanners and interlocks. Nothing about that is old-fashioned. It is the correct engineering answer whenever the station does not need a person inside it.

Where the confusion costs money

Buyers regularly assume that choosing a collaborative arm removes the safety scope from the project. It reduces it. It does not remove it. The risk assessment still has to be done, on the finished cell, and if the application relies on power and force limiting there is a measurement obligation attached to it. Budgeting as though safety engineering disappeared is how a cobot project ends up costing what a fenced cell would have cost, only slower.

02 / The market

Nine in ten new arms are still conventional

It is worth knowing the proportions before you assume the industry has moved on from fenced robots. The IFR World Robotics 2025 report records 542,000 industrial robots installed worldwide in 2024, more than double the number of ten years earlier, with Asia accounting for 74 percent of new deployments and a global operational stock of 4,664,000 units.

Within that total, more than 64,500 cobots were sold in 2024, close to a 12 percent share of new installations, growing 13 percent year on year, against an average manufacturing robot density of 177 robots per 10,000 employees globally and 204 in Asia.

Read that honestly and it says two things at once. Collaborative robots are now an ordinary part of the market rather than a novelty, and roughly nine of every ten new arms installed are still conventional. That is not conservatism. It is the shape of production work. Most stations are about cycle time, payload and repeatability, and at those stations a fence is cheaper than a compromise.

The useful conclusion for a Malaysian plant is that both categories are mainstream, both are supportable, and the decision is a station-level engineering question rather than a company-level strategy question. You will very likely end up with both on the same floor.

03 / The trade

Safety near a person is paid for in speed

Here is the mechanism, because it explains every specification difference. Power and force limiting keeps any contact between the robot and a person inside defined biomechanical force and pressure limits for each body region. The practical way to stay inside those limits, with a given mass moving toward a person, is to move slowly. That is why a collaborative arm running in collaborative mode is slower than the same class of conventional arm behind a fence. It is physics, not a firmware limitation, and no supplier can sell you out of it.

The second thing the datasheet will not tell you is that the published figures describe the bare arm. Robot manufacturers publish force and speed figures for the bare arm moving in free space, while validating power and force limits means physically measuring transient and quasi-static contact force and pressure at the actual end-effector and workpiece, at real operating speed, on the finished cell rather than the bare arm.

Your gripper changes the answer. A vacuum cup on a carton is one thing. A three-finger mechanical gripper holding a machined part with a sharp edge is another, and the part itself may be the hazard even when the arm is gentle. This is where a lot of optimistic cobot specifications fall apart, and it is the reason we do the safety work against the finished cell rather than against the brochure.

Payload is smaller than it looks

Collaborative arms are commonly offered up to roughly 20 to 30 kilograms of rating. The usable figure is lower, because the gripper eats into it. As one buyer guide puts it plainly, a 20 kg payload robot carrying a 5 kg gripper has only 15 kg of usable product capacity, and the robot's rated speed must include acceleration, deceleration and end-of-arm-tool exchange time rather than just the theoretical maximum. Size against the worst case part, not the nominal one, or the cell will be marginal on the day the heavy SKU runs.

04 / Side by side

The same job, two different machines

DimensionCollaborative armConventional industrial robotWhat it means for you
SpeedReduced in collaborative modeFull rated speed behind guardingAggressive cycle times settle this row first
PayloadCommonly up to 20 to 30 kg ratedTens to hundreds of kgSubtract the gripper before comparing
GuardingOften reduced, never assumedFencing, scanners, interlocksGuarding is decided by risk assessment, not category
Floor spaceSmall, fits an existing benchCell footprint plus safe clearanceTight floors push toward collaborative
RedeploymentEasy to move and reteachFixed installation, reprogrammedHigh-mix short runs favour collaborative
Process fitHandling, tending, light assemblyWelding, cutting, heavy handling, high speedHazardous processes are fenced regardless
Where cost sitsArm, tooling, validationArm, tooling, guarding, floorCompare cells, never arm prices

Scroll the table sideways on a phone

Notice that only two rows are really about the robot. The rest are about the station: the space, the process, the product mix and the person. That is the argument of this whole article in one table.

05 / Worked stations

Four stations, four different answers

Abstract comparison never settles a purchase. These are the four station types we are asked about most often, with the reasoning that decides each one.

01CNC or press tending, small parts  Part under 5 kg, cycle governed by the machine rather than the arm, an operator nearby for setup and gauging. This is the strongest cobot case on the floor. The arm is idle for most of the machine cycle, so its lower speed costs nothing, and the operator keeps access to the door.
02End-of-line palletizing, 12 to 15 kg cases  Genuinely borderline, and it turns on rate. At a modest case rate a collaborative arm on a lift column is a clean answer with a small footprint. Push the rate up and the conventional arm wins on both cycle time and cost per case.
03High-speed pick and place  Small parts at high rate from a moving conveyor. Not a collaborative application. The whole value is in speed, and speed is exactly what collaborative mode removes. Fence it, and use a delta or a fast six-axis arm with vision guidance.
04Welding, cutting, hot or sharp parts  The process is the hazard, not the arm. Arc, spatter, fume and heat mean guarding regardless of what is holding the torch, so the collaborative premium buys nothing.
05Inspection load and unload  A shared bench where an operator presents parts and the arm feeds an inspection station. Collaborative, light, low rate, and space constrained. It also connects the cell to machine vision, where the camera decides pass or fail and the arm sorts on the result.

Two patterns come out of these. Where the arm waits on a machine, collaborative speed is free. Where the arm sets the beat of the line, collaborative speed is expensive. Work out which of those two your station is, and the shortlist usually writes itself.

06 / Safety

The risk assessment is on the cell, not the arm

Our guarding approach starts from the same place every time: a risk assessment of the finished cell, with the real gripper, the real part and the real layout. The robot category is an output of that work, not an input to it. Sometimes the assessment says an open collaborative bench is fine. Sometimes it says light curtains on the operator side and a fixed guard on the conveyor side. Sometimes it says fence the whole thing and put the operator outside.

Under the current framework the documentation obligation is real. ISO 10218-2:2025 requires that the collaborative operation mode is selected and documented with rationale, that any installation using power and force limiting has measured force and pressure values for each contact scenario, that stopping distances and safety device response times are verified, and that validation testing is completed at maximum speed, full payload and worst-case geometry.

Read that list again and notice that most of it is about your cell rather than about a robot brand. This is why we treat the safety scope as part of the cell design from day one rather than as a compliance step at the end. Retrofitting guarding onto a cell that was sold as needing none is the most expensive way to arrive at the same place.

Guidance on collaborative deployment makes the same point about budget: the risk assessment is done in accordance with ISO 10218, and there is a substantial gap between a collaborative robot and an industrial one, which requires fencing, external safety devices and more floor space. That gap is a real cost difference. It is also a real capability difference, and the mistake is to price only one side of it.

07 / Cost

Compare cells. Never compare arms

An arm price tells you almost nothing about a cell price. The arm is typically the smaller half of the number. Around it sit the gripper or end-of-arm tooling, part presentation, the safety scope, controls and integration, commissioning on your floor and operator training. Change the category of arm and you move cost between those lines rather than removing it.

A collaborative cell usually saves on guarding and floor space and spends more on validation and on the tooling needed to work at lower speed without losing rate, sometimes by handling two parts per cycle instead of one. A conventional cell spends on guarding and floor and saves on cycle time, which shows up as fewer cells for the same output. Neither is inherently cheaper. The station decides which set of costs is smaller.

If you want the full build-up of what sits inside a cell price and how payback is actually calculated, we have written that separately in what a factory robot really costs and when it pays back. The short version for this article: get quotes for a working station, with tooling and safety included, and refuse to compare line items across suppliers who have scoped different things.

One more cost that rarely appears on a quotation. High product mix has a price in both categories, paid in fixtures and changeover rather than in the arm. If your product changes often, ask how a changeover is done, who does it, and how long the line is down while it happens. That answer moves the total cost more than the robot brand does.

08 / On your floor

We size the arm to the station

CODETRACE builds robotic arm cells for handling, machine tending, packing and end-of-line work, and we specify the arm last. First we take the station: the cycle time you need, the mass and geometry of the part including the gripper, the reach and the footprint you have, whether a person genuinely needs to be inside that space, and how often the product changes. Those five answers narrow the choice to one sensible category, and usually to one or two arm sizes.

Then the cell gets built and run on your parts in our workshop before it ships, because a cell that has only ever handled a sample part in a simulation is an assumption rather than a machine. We would rather find the gripping problem, the presentation problem or the reach problem in Shah Alam than on your production floor with your line stopped.

The control layer is part of the same job. A robot cell that cannot hand off cleanly to the machine before it and the conveyor after it is a fast island in a slow line, which is covered in our article on what a PLC does and why the control layer decides whether your line scales.

CODETRACE integrates on site from Shah Alam in Selangor and Batu Kawan in Penang, and is a member of the NVIDIA Inception program. Bring us the station, the part and the cycle time you need. We will tell you which arm belongs there, including when the answer is that this station should not be automated yet.

If a person needs to stand there, go collaborative. If not, fence it and go fast.

FAQ / Cobot vs industrial robot

Questions, answered.

01

What is the difference between a cobot and an industrial robot?

A cobot is an arm built so that it can be used in a collaborative application, meaning it can share a workspace with a person under defined safety conditions. A conventional industrial robot is built for speed, payload and reach, and is separated from people by fencing or safety devices. The important point is that collaborative describes the application, not the robot. An arm sold as collaborative can still need guarding once you put a sharp gripper, a hot part or a heavy load on it, because the hazard is the whole cell rather than the arm alone.

02

Are cobots slower than industrial robots?

Yes, when they are running in a collaborative mode. Power and force limiting works by keeping any contact with a person inside biomechanical force and pressure limits, and the practical way to stay inside those limits is to move slowly near the person. A conventional arm behind a fence has no such constraint and can run at its full rated speed. If your station needs an aggressive cycle time, that constraint decides the answer before payload or price does.

03

Do cobots still need a risk assessment and guarding?

Always a risk assessment, and sometimes guarding. ISO 10218-1:2025 and ISO 10218-2:2025 came into force on 1 April 2025 and absorbed the collaborative-application requirements that ISO/TS 15066 previously held separately. Under that framework the risk assessment is done on the finished cell, and any application using power and force limiting has to demonstrate measured contact force and pressure values for each contact scenario. A robot manufacturer's certification of the bare arm does not cover your gripper, your part or your layout.

04

What payload can a cobot handle?

Collaborative arms are commonly available in the range up to roughly 20 to 30 kilograms, while conventional industrial robots go far beyond that. The number that matters is not the arm rating but the usable payload after the gripper. A 20 kg arm carrying a 5 kg gripper has 15 kg left for the product, and that margin shrinks further at full reach and at high acceleration. Size the arm against the heaviest part you will ever run, not the average one.

05

How common are cobots compared with industrial robots?

Cobots are the faster growing segment but still the smaller one. IFR figures for 2024 show 542,000 industrial robots installed worldwide, of which more than 64,500 were collaborative, close to a 12 percent share of new installations with 13 percent year on year growth. Read that two ways. Collaborative robots are now a normal part of the market rather than a novelty, and roughly nine in ten new installations are still conventional arms, because most production stations are about cycle time and payload rather than about sharing space with people.

06

When is a cobot the wrong choice?

When the cycle time is aggressive, when the payload is heavy, when the reach needed is long, when the process itself is hazardous regardless of the arm, such as welding, cutting or handling hot parts, and when nobody actually works at the station. If the cell will end up fenced anyway, you are paying a collaborative premium for capability you cannot use. In that case a conventional arm behind a fence is usually faster and cheaper per part.

07

How do I decide for my own station?

Work in this order: required cycle time, mass and geometry of the part including the gripper, reach and footprint available, whether a person genuinely needs to be at that station, and how often the product changes. Only after those five answers does the cobot or conventional question have a correct answer. CODETRACE sizes the arm to the station rather than the other way around, proves the cell on your parts in our workshop, and carries out the risk assessment on the finished cell before it ships.

Sources / Every figure in this article

Where the numbers come from

Installation and density figures are IFR World Robotics data, either from the IFR release or from trade reporting of it. Safety requirements are described from published summaries of the 2025 ISO 10218 revision rather than from the paid standard text, so read them as an accurate account of scope and not as a substitute for the standard. Vendor guidance is labelled as vendor guidance.

01International Federation of Robotics, World Robotics 2025  542,000 industrial robots installed in 2024, 4,664,000 operational stock, Asia 74% of new deployments.Industry body
02Trade press reporting IFR 2024 data  Over 64,500 cobots sold, close to 12% of new installations, 13% growth; density of 177 per 10,000 employees globally, 204 in Asia.Trade, secondary
03Guide to the 2025 revision of ISO 10218  In force 1 April 2025; ISO/TS 15066 collaborative requirements folded into ISO 10218-1 and 10218-2.Vendor guide
04Power and force limit validation guide  Published figures describe the bare arm; contact force and pressure must be measured on the finished cell at real operating speed.Vendor guide
05Collaborative robot safety standards reference  Documentation requirements under ISO 10218-2:2025, including mode rationale, measured forces and validation at worst case.Vendor guide
06Summary of collaborative safety standards  The four recognised collaborative techniques and the relationship between ISO 10218 and ISO/TS 15066.Vendor guide
07Palletizing robot buyer guide  Usable payload after gripper mass; rated speed must include acceleration, deceleration and tool exchange.Vendor
08CODETRACE first-party: robotic arm cells built and run on customer parts in our workshop before shipping, with the risk assessment carried out on the finished cell.First-party

Send us the part and the cycle time. We will size the arm.

Talk to our engineers
Scroll to Top