IP66 vs IP67 for outdoor robots: what the rating actually tests, and what a UK site needs

IP66 means the enclosure is dust-tight and withstands powerful water jets: a 12.5 mm nozzle at 100 litres a minute from three metres, for at least three minutes. IP67 means the enclosure is dust-tight and withstands temporary immersion: 30 minutes with the lowest point one metre below the surface. The first is a hose test; the second is a bath test. A robot rated IP67 has been shown to survive being submerged, not to survive being jetted, and under IEC 60529 the two tests are separate: the water ratings are not cumulative beyond IPX6, so a device compliant with IPX7 is not necessarily compliant with IPX5 or IPX6. A maker who has tested both expresses it as a dual rating, IP66/IP67, and a maker who publishes only IP67 has told you about immersion and nothing about jets.

For a UK site the distinction is not academic. A yard that floods after heavy rain is an immersion question. A wash-down bay, a food plant, a pressure-washed forecourt or a robot that will be hosed clean after a muddy patrol is a jet question. Most outdoor security and inspection work is neither: it is sustained rain and spray, which IPX5 and IPX6 cover and which IPX7 alone does not formally address.

An IP code has two digits. The first is solids and dust, the second is water. “X” in either position means that digit was not tested, not that it passed. These are the levels that matter for an outdoor robot, with the test conditions the standard specifies.

It is the immersion duration in the IPX7 test: the enclosure sits with its lowest point one metre under water for 30 minutes and no harmful quantity of water gets in. It is a test of a still bath at a shallow depth, at the temperature the standard specifies, with the enclosure as the maker submitted it. It is not thirty minutes of hosing, thirty minutes of driving rain, or thirty minutes under a pressure washer; those are the IPX5, IPX6 and IPX9K tests, with their own nozzles and flows. So when a specification says a robot “must withstand 30 minutes of wetting”, the first question to ask is which wetting: a puddle, a hose or a jet wash. The answer chooses the rating, and a maker who publishes IP67 alone has not answered the hose.

The second question is what was in the bath. The IP rating applies to the enclosure as tested: body, covers, seals and connectors in the state the maker submitted them. A payload bolted on afterwards, an open connector cap, a charging port, a swapped battery or a cracked cover is outside the test. On a quadruped the most common weak points are exactly those: the payload interface on the back, the leg joints, and the charging contacts underneath.

The makers’ published ratings, read on 22 September 2026, with what each actually tells you. Where a maker publishes nothing, it is not making a claim, and a buyer should not make one on its behalf.

The pattern in that table is worth stating plainly. The industrial quadrupeds publish IP67, which is the immersion test, and none of them publishes the jet tests alongside it. That is not a criticism; it is a gap in what a buyer can know from the page, and it is why a UK wash-down or driving-rain requirement should be put to the maker as a direct question, with the answer in writing, rather than inferred from a 67.

Decide from the water the robot will meet, not from the highest number on offer.

And in every case the published rating narrows the field rather than closing the question. Ratings are laboratory tests on a new enclosure; a robot that has walked a gravel route for six months has seals and covers that the test never saw. The reliable way to know is a bounded pilot that runs the route through the wettest week available, which is how we structure a pilot programme. The wider reading of a spec sheet, temperature, step, endurance and recharge alongside ingress, is in our guide to reading a quadruped specification, and the full platform comparison is in security robot dogs in the UK.