The interesting part of building a robot mower is not the design. It is the eighteen months between a design that works on a bench and a machine you would leave alone in someone's garden in the rain.
LYMOW started as a concept in August 2022. Here is roughly what happens in between.
Concept, and the decision that shapes everything else
Early on you commit to a drivetrain, and everything downstream inherits it. Choosing tracks over wheels bought traction and low ground pressure, and in exchange it made the machine heavier, more complex to build and much more dependent on software to turn without scuffing turf. Most of the work that followed was paying off that trade.
Prototypes: making it work once
Through 2023 the job is a machine that does the thing at all — drives, navigates, cuts, docks. Prototypes at this stage are ugly, over-instrumented and full of wiring. They exist to answer questions, not to be admired.
The redesign nobody plans for
In January 2024 the cutting system was reworked. That is the part of a development timeline that never appears in the marketing: a subsystem that tested acceptably turned out not to be good enough in real, dense, overgrown grass, and going back cost months.
The result became the current airflow-assisted deck — lift the grass upright before the blades reach it, recut the clipping inside the housing, discharge heavier material to the side rather than letting it pack. It is a better answer than the one it replaced, which is the only justification a redesign ever has.
Validation: breaking it on purpose
Engineering validation is where a design stops being a design. The machine gets driven at slopes past its rating until it fails, so the rated figure has margin underneath it. Water ingress is tested to the IPX6 standard with pressurised jets rather than a watering can. Frames get loaded and drop-tested. Battery packs are cycled and run at temperature extremes.
None of this is glamorous and most of it produces no story — the good outcome is a report that says nothing happened.
Field testing: the part that cannot be simulated
Test rigs cannot reproduce a garden. Real lawns have moles, sprinklers, gravel that migrates onto the grass, children's toys, tree roots that break the surface, hedgehogs at dusk and a hose that someone left out. Field units go into ordinary gardens across different climates and grass types, and the logs come back with the situations nobody thought to design for.
Most of what looks like intelligence in a shipping robot mower is really a long list of specific situations that a field unit hit first.
Production, and the discipline of not changing anything
By the Kickstarter launch in October 2024 the design was frozen. Pilot runs check that a machine built on a production line behaves like the one built by hand, and after that the discipline is to change nothing quietly. A component substitution that looks equivalent on paper is exactly how a reliable product becomes an unreliable one.
Why any of this matters to a buyer
Because a robot mower is one of the few appliances that operates unsupervised, outdoors, near people and pets, for years. The specification tells you what it does on a good day. The validation programme is what decides how it behaves on a bad one — and that is the part you will actually live with.
