Forge Robotics in RoboticsTomorrow: how to read a quadruped spec sheet, and what changes on a UK site

On 11 September 2026 RoboticsTomorrow published Consumer platform or industrial patrol robot? How to read a quadruped spec sheet, a contributed article by our founder, Simon Bumford. RoboticsTomorrow is an established US-based online trade title covering robotics, advanced manufacturing and factory automation for industrial readers, and it takes contributed pieces on one condition: they have to teach something. Product cut sheets and promotional copy are turned away.

That condition suited us. Forge Robotics works across manufacturers and does not make robots of its own, so we can put five quadrupeds side by side on their makers’ own published figures without a reason to steer anyone towards one badge. The article does exactly that with the Unitree Go2 and B2, the DEEP Robotics X30, Boston Dynamics Spot and ANYbotics ANYmal, and it is the kind of independent reading a trade audience can use whether they ever speak to us or not.

We are glad to see it in print, and glad of what it says about the approach. The rest of this page is the short version of the argument, followed by the one thing a title with a largely North American readership could not spend its space on: what the same four lines mean when the robot has to work through a British winter.

Quadruped robots have converged on a single silhouette. Watch three promotional videos back to back and you see the same four legs, the same confident trot across gravel, the same recovery after a shove. What the footage does not show is that two machines which look identical can be an order of magnitude apart in what they will survive on a working site.

The market splits into two families that happen to share a body plan. The first is the consumer, education and research platform: affordable, capable, quick to get moving, built so that people can develop against a legged robot without a six-figure budget. The second is the industrial platform, specified for a job that runs on a schedule in weather that does not cooperate. Almost every disappointed deployment we have seen traces back to buying the first and expecting the second.

The specification sheet is where that difference actually lives, if you know which lines to read. The article walks through them in order. This diagram is the short form.

Read in this order, each line either rules a machine out or sends you to the next one.

The article goes on to perception, tiered SDK access and the three things that never appear on the sheet at all: certifications, spares and repair turnaround, and integration into the control room you already run. Read the full guide on RoboticsTomorrow.

None of the figures change. What changes is which of them bites first, and on a British site the order is not the one a buyer in a hot, dry climate would assume.

Temperature rules machines out before ingress does

In much of the United States, ingress protection is the first filter because heat and dust are the problem. In the UK it is the lower temperature bound. A published floor of 5 degrees Celsius rules a machine out of a large part of the British winter before you have considered rain, payload or price, and it does so on the manufacturer’s own words rather than on any test of ours. The conditional wording matters here too: a robot that must be powered on at zero degrees or above is a real constraint for a machine living in an unheated compound in January.

Rain is not weather here, it is the default

An IP54 rating is a genuine outdoor rating, and it is enough for a great deal of work. It is not a wash-down or heavy-rain rating, and a UK security or inspection patrol that runs every night, all year, will meet heavy rain as a matter of routine rather than exception. That is why, on a British site, the honest split is not IP-rated versus unrated but IP67 versus everything else, and why the absence of a published rating should end the conversation for outdoor use.

The stairs are older and the nights are longer

Much of the UK’s industrial and commercial estate predates the robots by decades, and its stairs, thresholds and kerbs were never specified with a machine in mind. A step rating of 15 or 16 centimetres will pass a purpose-built test course and can fail the plant stairs on the first night. Open grated steel walkways, common on UK process and utility sites, are a separate problem again, because the robot’s perception has to cope with seeing through the tread, and most specification sheets do not mention them at all.

Endurance is also a different question at this latitude. A December night in Britain is around sixteen hours of darkness. A patrol robot that cannot return to a dock and recharge on its own is not covering that, however good its headline runtime, which is why the recharge story is the endurance figure we read first for UK security work.

Everything above narrows the field. None of it tells you whether the machine works on your site, because published figures are laboratory and best-case numbers by nature. They are honest and useful, and they are not a test result from your yard in November.

The reliable way to close that gap is a bounded pilot: one route, run repeatedly, with the measures agreed before the robot arrives. Coverage against the intended round, false alarm rate and what each false alarm costs in operator time, uptime including charging behaviour, and the hours actually saved. Include an exit condition. A pilot that concludes a robot is wrong for the site has saved considerably more than one that flatters it. That is how we structure a pilot programme, and the full UK comparison behind the article, with the manufacturers’ own videos and a side-by-side table, is in our guide to security robot dogs in the UK.

RoboticsTomorrow is an online trade publication for the robotics and industrial automation industry, one of a family of sector titles from the same publisher that also covers manufacturing. Its readers are the people who specify, buy and integrate automation, and its editorial line is that contributed articles must inform rather than sell. That is a discipline we share, and it is the reason we offered them a piece that compares platforms on the manufacturers’ own figures and names no winner.

Our thanks to the RoboticsTomorrow editorial team for running the guide, and for running it with the figures intact. The article was an unpaid editorial contribution; we have no commercial relationship with the publication and receive nothing from it. We link to them here for the same reason they linked to us: because the piece is worth reading in full.