Collaborative Robots in Manufacturing: Where Cobots Fit and Where They Don't

Collaborative robots — cobots — are the best fit for manufacturing tasks that are repetitive, low-to-medium payload, and located on the factory floor next to people: machine tending, palletizing, screwing, dispensing, and pick-and-place assembly. They fall short whenever a job calls for raw industrial-robot speed and payload behind a guarded cell, or where the work is too variable, too heavy, or too safety-critical for a power-and-force-limited arm to handle without slowing down to a crawl. That split — not “cobots versus robots” as a winner-take-all contest — is the practical reality on shop floors right now, and it shows up clearly in how fast each category is actually being bought.

What exactly makes a robot “collaborative”?

A collaborative robot is defined less by what it looks like and more by how it is allowed to operate: it’s built and certified to work in the same physical space as people without a fence, using power-and-force limiting, speed-and-separation monitoring, or hand-guiding instead of a perimeter cage. A traditional industrial robot arm can be just as precise or even faster, but it’s designed to run at full torque and speed, which is exactly why it needs a guarded cell to keep humans out while it’s moving.

That safety architecture used to live in a separate technical specification, ISO/TS 15066:2016, bolted onto the main industrial-robot safety standard. As of April 1, 2025, that changed: the revised ISO 10218-1:2025 and ISO 10218-2:2025 folded the collaborative-application requirements directly into the core standard, replacing the 2011 editions, and added cybersecurity requirements for robot systems for the first time. In practice, this means “collaborative” is no longer a bolt-on afterthought in the safety literature — it’s now written into the same document that governs every industrial robot, cobot or not.

How big is the cobot segment in manufacturing today?

Cobots are still a minority of new industrial robot installations worldwide, but a growing one. According to the IFR World Robotics 2025 report, roughly 64,500 cobots were installed globally in 2024, representing 11.9% of all new industrial robot installations — up from 10.6% in 2023. (Secondary coverage of that same IFR dataset differs slightly on the exact growth figure, with some outlets citing 12% year-over-year growth and others 13%, and unit counts reported as either about 64,500 or 64,542 — a reminder that even well-sourced industry statistics carry some rounding noise between publications.)

North America shows the same trend, but it has accelerated even faster lately. The Association for Advancing Automation (A3) reported 7,212 cobot units ordered in North America in 2025, worth $241 million — 19.6% of all robots ordered that year and 10.7% of total order revenue, the first full year A3 tracked cobots as a distinct order category. That momentum carried into 2026: Q1 2026 saw 1,637 collaborative robots ordered in North America, worth $69.8 million, up 55.6% in units and 78.2% in revenue year-over-year, and accounting for 18.1% of all robot units ordered in the quarter. Cobots are growing faster than the broader robot market, even while remaining a smaller slice of total volume.

On the vendor side, Universal Robots — the company most closely associated with popularizing the category — passed 100,000 cumulative cobots sold in early 2025, a milestone reported in May 2025 and confirmed on the company’s own history page. Kim Povlsen has led Universal Robots as president since March 1, 2021.

Where do cobots fit best on the factory floor?

Machine tending. Loading and unloading CNC machines, injection-molding presses, or stamping equipment is one of the most common cobot deployments because the task is repetitive, the payloads are usually modest, and a human operator often still needs to work nearby for setup, inspection, or handling exceptions.

Palletizing and de-palletizing. Stacking boxes or totes onto pallets is physically taxing and repetitive — a strong fit for a cobot arm that can run near-continuously at a moderate pace while an operator manages the rest of the line.

Assembly and pick-and-place. Screwdriving, dispensing adhesive, inserting components, and light assembly are common cobot tasks precisely because they benefit from close proximity to a human co-worker who handles the parts of the job that still need manual dexterity or judgment.

Quality inspection and light finishing. Cobots equipped with cameras or sensors can present parts for inspection, or run polishing and deburring passes, in workcells shared with people doing other steps of the process.

Universal Robots’ own newest product illustrates where the category is stretching. The UR15, opened for pre-order in May 2025 and shipping from June 2025, carries a 15 kg payload (17.5 kg in a wrist-down orientation), a maximum TCP speed of 5 m/s, and a 1,510 mm reach — marketed as UR’s fastest cobot yet, aimed squarely at machine tending, palletizing, assembly, and pick-and-place. It’s a good proxy for the ceiling cobots are pushing toward: heavier, faster, but still built to operate without a cage.

Where do cobots fall short?

Very heavy, very fast, or very large-scale work. Full-size automotive body welding, large-part handling, and high-throughput lines built around cycle times measured in fractions of a second are still the domain of traditional caged industrial robots, which can move at full speed and full payload without power-and-force limits slowing them down.

Highly variable, low-volume, high-complexity assembly. Tasks that change constantly, or that need a level of dexterity and judgment closer to human hands, often aren’t worth automating with a cobot arm at all — the setup and programming overhead can exceed the labor it replaces.

Long-distance material movement. Moving parts or totes across a facility is generally a job for an automated guided vehicle or autonomous mobile robot, not a stationary cobot arm bolted to a workbench; the two categories are complementary rather than substitutes, a distinction covered in more detail in our look at warehouse robotics ROI.

Environments where raw throughput per dollar invested is what matters most. Where floor space is cheap, guarding is already in place, and volumes are high and stable, a traditional caged robot is frequently the better economic choice — a cobot’s safety-by-design comes at a cost in speed that a fenced cell doesn’t have to pay.

Cobots vs. traditional industrial robots vs. manual labor

Dimension	Collaborative robot (cobot)	Traditional industrial robot	Manual labor
Typical payload	Light to medium (UR15: 15 kg, up to 17.5 kg)	Medium to very heavy	Limited by ergonomics
Speed	Moderate, safety-limited without guarding (UR15: up to 5 m/s TCP)	High, full speed behind a guard	Variable, fatigue-limited
Safety approach	Power-and-force limiting, speed-and-separation monitoring, hand-guiding	Perimeter guarding, light curtains, e-stops	Human judgment, PPE
Governing standard	ISO 10218-1:2025 / ISO 10218-2:2025 (collaborative requirements now integrated)	ISO 10218-1:2025 / ISO 10218-2:2025	Workplace safety regulations
Best fit	Machine tending, palletizing, light assembly, inspection	High-volume, high-speed, heavy-payload lines	Highly variable, low-volume, judgment-heavy tasks
Deployment flexibility	Redeployable across workcells with less re-guarding	Fixed, guarded installation	Fully flexible, but scarce and costly

What changed with the 2025 safety standard overhaul, and why does it matter?

The move to ISO 10218-1:2025 and ISO 10218-2:2025 is more than a paperwork update. Folding the collaborative power-and-force-limiting requirements out of the standalone ISO/TS 15066:2016 and into the main standard signals that regulators now treat collaborative operation as a first-class mode of industrial robot use, not a niche exception. The addition of cybersecurity requirements — a first for this standard — also reflects a broader shift: as cobots increasingly connect to plant networks, MES systems, and remote monitoring tools, their attack surface has become a safety consideration in its own right, alongside physical force limits. For manufacturers planning new cobot deployments, this means safety compliance work now has to account for both mechanical risk and network exposure from day one.

What does Teradyne’s new US manufacturing hub signal about where this is heading?

In December 2025, Teradyne Robotics announced a new US Operations Hub in Wixom, in Metro Detroit, Michigan, opening in 2026 to manufacture Universal Robots industrial cobots, with future potential to also build MiR autonomous mobile robots at the same site. The hub is projected to create more than 200 jobs over the coming years and will double as a customer training, service, and visitor-experience center. Locating cobot manufacturing closer to North American customers — the same region where A3 tracked cobot order growth of over 55% in units in Q1 2026 alone — suggests vendors are betting that demand for collaborative automation in general manufacturing will keep outpacing the broader robot market, not just enjoying a passing trend.

How should a manufacturer decide between a cobot and a traditional robot?

The decision usually comes down to three questions: How much does the task need to move, how fast, and how close to a person? If the answer is “moderate payload, moderate speed, right next to an operator who’s still needed for part of the job,” a cobot is very likely the right tool. If the answer is “heavy, fast, high-volume, and the cell can be guarded,” a traditional industrial robot is usually the more economical choice. And if the task is too variable or judgment-heavy to automate at all yet, neither category may be worth the investment — a question worth asking before any automation project, covered more broadly in our overview of Physical AI use cases and ROI. Once the hardware decision is made, getting a cobot rollout right operationally increasingly depends on the same fleet-management discipline used for other autonomous systems — see our explainer on RobotOps for how that plays out in practice.

The bottom line

Cobots have carved out a durable, fast-growing niche in manufacturing rather than replacing traditional industrial robots outright. The data backs this up on both sides of the Atlantic and across vendor announcements: a rising share of new installations globally, accelerating order growth in North America into 2026, a major safety-standard overhaul that formally integrates collaborative operation into the core rulebook, and vendor investment in dedicated manufacturing capacity to keep up with demand. The practical takeaway for anyone planning a factory-floor automation project is to match the tool to the task — light, variable, human-adjacent work tends to suit a cobot; heavy, fast, high-volume work still tends to suit a guarded industrial robot — rather than treating either category as a universal answer. For a wider view of how these deployment patterns compare across industries, our hub on where Physical AI works today tracks the broader picture, and readers new to the underlying terminology can check the glossary for quick definitions of terms like power-and-force limiting and speed-and-separation monitoring.