Start With What Industrial Robots and Cobots Actually Are
An industrial robot is a fenced, high-speed manipulator built for repeatable throughput, typically from vendors such as FANUC, ABB Robotics, KUKA, and Yaskawa Motoman. A collaborative robot, or cobot, is a force- and speed-limited arm designed to share space with people under a per-application risk assessment, from vendors such as Universal Robots, Techman Robot, and Doosan Robotics.
The safety framework is different, not optional. ISO 10218-1:2025 and ISO 10218-2:2025 cover robot and robot-system safety generally, while ISO/TS 15066:2016 supplements those standards with the force and pressure limits that make a collaborative application possible. OSHA notes that most robot incidents happen during setup, programming, and maintenance rather than production, which is a reminder that neither category is "safe by default."
The Bot Scout framing: an industrial robot is a machine that assumes people are outside a fence, and a cobot is a machine that assumes people might be inside one. Everything else — payload, reach, cost, footprint — follows from that single assumption.
Compare the Twelve Attributes That Decide the Buy
The chart below is the honest side-by-side most vendor decks skip. Numbers are typical published ranges for current families in each category; check the specific model against the specific job before signing.
Two attributes matter more than the rest: the safety story (what actually keeps a person out of a hazard) and the true installed cost (arm plus tooling, fixtures, guarding, integration, and lost production during commissioning).
| Attribute | Industrial robot | Cobot |
|---|---|---|
| Safety approach | Fixed guarding, light curtains, safety-rated stops per ISO 10218 | Power- and force-limited operation with a per-application risk assessment per ISO/TS 15066 |
| Payload | Typically 3–2,300 kg (heavy-duty families from FANUC, KUKA, ABB, Yaskawa) | Typically 3–35 kg (UR20, Doosan H-Series, Techman TM25S at the top end) |
| Speed | Up to ~4 m/s tool-center speed in production | Reduced to meet force/pressure limits when a person is present; near-industrial speeds only when separated |
| Reach | Commonly 500 mm to 3,500+ mm | Commonly 500 mm to ~1,900 mm |
| Repeatability | Often ±0.02–0.05 mm on precision families | Often ±0.03–0.1 mm depending on model |
| Cost of the arm | ~$25k–$150k+ for the manipulator and controller | ~$20k–$75k for the manipulator and controller |
| Total installed cost | Arm often a minority of the cell cost after guarding, fixtures, integration | Arm a larger share of the cell cost; guarding may be lighter but not zero |
| Footprint | Arm plus fenced cell, safety mats, light curtains | Arm plus a risk-assessed work zone; can be smaller but is not zero footprint |
| Install time | Weeks to months for a fully integrated cell | Days to weeks for a typical machine-tending or pick-and-place deployment |
| Programming complexity | Vendor teach pendant and language (KRL, RAPID, KAREL, INFORM); often needs an integrator | Tablet-style teaching with hand-guiding on most brands; still needs safety validation |
| Typical use cases | High-throughput welding, painting, heavy palletizing, press tending, body-in-white | Machine tending, light assembly, inspection, packaging, small-part pick-and-place |
| Governing standard family | ISO 10218-1 and -2:2025 | ISO 10218-1 and -2:2025 plus ISO/TS 15066:2016 |
Use the Decision Tree Before You Request a Quote
The Bot Scout decision tree is a series of gates. Fail one gate and the answer flips. It replaces the vendor demo that ends in "a cobot could do this" with a checklist that produces a defensible choice.
Compare this framework against the broader machine-type view in the manufacturing robots guide and the collaborative-specific view in the cobot guide before committing.
- Is the required payload above about 35 kg, or the reach above about 1.9 m? If yes, go industrial — most cobot families cannot serve the job.
- Does the target cycle time need continuous full-speed motion? If yes, go industrial — a cobot operating in collaborative mode will not hit it.
- Will a person routinely be within the robot's reach during production? If yes, a cobot is a candidate — but only after a risk assessment against ISO/TS 15066, including the tool and the part.
- Is the end effector sharp, hot, or otherwise hazardous regardless of arm speed? If yes, the collaborative story collapses; plan for guarding whether the arm is called "cobot" or not.
- Is floor space or a dedicated cell footprint a hard constraint? If yes, a cobot deployment usually wins, provided the first four gates allow it.
- Is the team going to program and maintain the cell in-house, without a full-time integrator on call? If yes, favour the cobot programming model unless throughput forces the other way.
Budget the Whole Cell, Not the Arm
Both categories share the same uncomfortable pattern: the arm is the small number and the cell is the big one. End-of-arm tooling, fixtures, feeders, safety assessment, guarding, controls integration, operator training, and the production time lost during commissioning routinely exceed the manipulator price.
Cobot deployments compress the schedule but do not zero out safety cost. A power- and force-limited arm carrying a sharp gripper or a hot torch is not collaborative in that configuration, and the risk assessment will require guarding, presence sensing, or speed-and-separation monitoring to close the gap.
The line most budgets miss is year two: re-teaching after a part change, replacing wear items, and retraining after turnover. An industrial cell absorbs those changes through an integrator visit; a cobot cell absorbs them through in-house engineering time. Either way, a year-one-only budget is a purchase, not a plan.
When the Answer Is Both
Plenty of real production floors run both. A guarded industrial cell handles the high-throughput anchor operation while cobots pick up machine tending, kitting, or inspection tasks that would otherwise stay manual. The two categories are complements more often than substitutes, and treating them as an either/or forces bad compromises.
The tell that a plant needs both is a mix of one high-volume, high-cycle-time operation and several low-volume, mixed-SKU tasks. The industrial arm carries the anchor; the cobots fill the gaps. That pattern shows up across the same vendor families — FANUC and ABB both sell into industrial and collaborative lines, and Universal Robots parents sit alongside industrial peers in many integrator portfolios.
The wrong version of "both" is a cobot bought as a cheaper industrial robot and then fenced in permanently. If the answer is a fence, the honest buy is an industrial arm sized for the job.
Bottom Line
Industrial robots and cobots are not tiers of the same product — they are answers to different questions about who stands next to the machine. Pick the industrial arm when payload, reach, or cycle time forces the fence; pick the cobot when the job is light, mixed, and shared with people; pick both when the floor has one anchor operation and a long tail of smaller ones.
Run the six-gate decision tree against your actual job before you accept a demo, and price the full cell — arm, tooling, guarding, integration, and year-two ownership — before you compare quotes.
FAQs
Is a cobot always safer than an industrial robot?
No. Safety is assessed per application under ISO 10218 and ISO/TS 15066, including the tool and the part. A cobot carrying a sharp, hot, or pinch-hazard end effector can present more risk than a fenced industrial arm running the same motion.
What payload can a cobot handle compared to an industrial robot?
Most cobot families top out around 20–35 kg (for example UR20, Doosan H-Series, Techman TM25S), while industrial families from FANUC, KUKA, ABB, and Yaskawa run from a few kilograms to well over 2,000 kg. Above roughly 35 kg or 1.9 m of reach, the cobot category usually cannot serve the job.
Do cobots need fences?
Sometimes. A cobot operating in a truly collaborative mode meets ISO/TS 15066 force and pressure limits without fencing, but if the end effector or the part introduces a hazard the arm's own limits do not cover, the risk assessment will require guarding, presence sensing, or speed-and-separation monitoring.
Which is cheaper to install: an industrial robot or a cobot?
The cobot arm is usually cheaper and the cell is faster to stand up, but neither category is cheap once you add tooling, fixtures, safety, integration, and commissioning. Compare total installed cost and year-two ownership, not the arm price.
Can one plant use both industrial robots and cobots?
Yes, and many do. A common pattern is a guarded industrial cell for the high-throughput anchor operation and cobots for mixed-SKU machine tending, kitting, or inspection. Treat the two categories as complements, not substitutes.