Universal Robots Review: The Lineup Is Two Ladders, Not One

Universal Robots did not invent the collaborative arm, but it made one that ordinary factories could buy and ordinary engineers could program, and the UR5e is plausibly the most deployed cobot in the world. If you are automating a bench task and you ask three integrators what to use, at least two will say UR.

Where buyers go wrong is the model choice, and they go wrong in a specific, expensive way: they read the lineup as a single ladder from small to large. It is not. It is two ladders that cross, and the payload number on the datasheet is not the payload you get.

Companion reading: how to choose a cobot, collaborative robots (cobots), how to choose an industrial robot arm, end-effectors and grippers, and robot safety and functional safety.

Table of contents

1. Key takeaways
2. The lineup, and why it is not a ladder
3. The payload number is not the payload
4. What the arm costs versus what the project costs
5. Why UR still wins on software
6. How it compares
7. Who should buy it
8. Verdict
9. Frequently asked questions
10. Changelog

The lineup, and why it is not a ladder

Here is the range as of 2026, ordered by payload rather than by name:

Read the reach column and the problem appears. The UR16e carries more than the UR10e but reaches 400 mm shorter. The UR30 carries 50% more than the UR20 and reaches 450 mm shorter. The lineup branches: one path optimises for load at close quarters, the other for envelope.

The practical selection rule is to fix the harder constraint first. Measure the furthest point the tool must reach in the real cell, including the approach posture, then find the smallest model that covers it, then check payload against the full tool stack below. Choosing on payload alone is how a cell ends up with an arm that can lift the part and cannot reach the far side of the fixture.

The UR3e deserves a separate note: at 3 kg and 500 mm it is a desktop-scale machine, light enough to mount on a mobile base, and it is a different kind of purchase from the rest of the range.

The payload number is not the payload

This is the mistake that costs the most rework, and it is not specific to UR. It is just most visible here because so many UR cells are specified by people automating for the first time.

The rated payload covers everything past the flange. The gripper, its mounting plate, any tool changer, the cable and hose bundle the arm carries, and the part. A 5 kg rating with a 2 kg gripper and a 1 kg tool plate leaves 2 kg for the workpiece before anything else.

It also derates with geometry. Effective payload falls as the arm extends and as the tool orientation moves away from the favourable case. Centre of gravity and inertia matter as much as mass, particularly on fast moves or when the tool sticks far out from the flange. The clearest evidence that the headline number is posture-dependent: some variants publish a higher rating specifically for wrist-down applications like palletising.

So the honest sizing method is: total the whole tool stack, add the part, apply the arm's derating at your worst-case reach and orientation, and only then compare to the datasheet. Our how to choose a robotic gripper guide covers the tool side, and end-effectors and grippers covers what those masses actually look like.

What the arm costs versus what the project costs

UR publishes list prices through its distributors, and the range across the lineup spans roughly a factor of three from the smallest to the largest arm. Those numbers move with configuration and region, so treat them as a band and get a current quote.

The number that does not move is the ratio: accessories, integration and training can roughly double the base arm price. Grippers, tool changers, vision, safety scanners, fixturing, the pedestal or rail, the integrator's time, and the training that lets your own people modify the program next year.

A business case built on the arm price alone is understating the project by around half. Build it the other way around: cost the cell, then check that the arm is the cheap part.

Why UR still wins on software

The hardware is good and it is no longer unique. Competitors match or beat the specs and often undercut the price. UR's durable advantage is that a maintenance technician can be taught to program one.

The teach pendant, the waypoint model and the accessory ecosystem mean a cell can be re-tasked by the people who work next to it rather than by a contractor on a callout. On a shop floor that changes what automation costs over five years far more than the purchase price does. Our how to program a robot arm guide covers where that model sits against offline programming and ROS 2.

The trade-off is the ceiling. If your application needs 1 kHz torque control, custom impedance behaviour or a learned policy, the friendly abstraction becomes the constraint, and a research-grade arm like the Franka Research 3 is the better tool.

How it compares

Who should buy it

Buy UR if you are automating real production tasks, you want your own staff to own the programming, and you value the accessory ecosystem and the integrator network. That is the majority of cobot buyers, which is why UR has the install base it does.

Look elsewhere if you need research-grade low-level control, if your payload or reach requirement sits outside the range once you have derated it honestly, or if price-performance dominates and you are willing to trade ecosystem depth for it.

Verdict

Universal Robots remains the default cobot for good reasons, and none of them are on the spec sheet: approachable programming, the deepest accessory catalogue in the category, and an integrator network that has seen your application before.

Get two things right and it is hard to regret. Pick the model against reach and payload as separate constraints rather than reading the range as one ladder, and size the payload against the entire tool stack at your worst-case posture rather than the datasheet headline. Then budget the cell rather than the arm, because the arm is about half of it.

Frequently asked questions

Which Universal Robots model should I buy? Fix reach first, then payload. The lineup is not a single ladder: the UR16e carries more than the UR10e but reaches 400 mm shorter, and the UR30 carries more than the UR20 but reaches 450 mm shorter. Measure the furthest point the tool must reach in the real cell, choose the smallest arm that covers it, then verify payload against your full tool stack.

Does the rated payload include the gripper? Yes, and more than the gripper. The rating covers everything beyond the flange: gripper, mounting plate, tool changer, the cabling and services the arm carries, and the part. Centre of gravity and inertia count too, especially on fast moves.

Does payload change with reach? Yes. Effective payload falls as the arm extends and as tool orientation moves away from the favourable case. Some variants publish a higher rating specifically for wrist-down work such as palletising, which is direct evidence that the headline number depends on posture.

How much does a Universal Robots cobot cost? List prices span roughly a factor of three from the smallest arm to the largest, and move with configuration and region, so get a current quote. The more useful planning number is that accessories, integration and training can roughly double the base arm price.

What is the difference between the UR20 and the UR30? Payload and reach trade against each other. The UR30 carries 30 kg to the UR20's 20 kg, and reaches 1300 mm to the UR20's 1750 mm. Pick by which constraint binds in your cell.

Is UR still worth it when competitors are cheaper? Often, yes, and the reason is not hardware. The teach-pendant programming model lets your own technicians re-task the cell, and the accessory ecosystem is the largest in the category. Over five years that usually matters more than the purchase price.

Can I do research on a UR arm? Up to a point. For 1 kHz torque control, custom impedance behaviour or learned policies, a research-grade arm is the right tool.