But drip irrigation doesn’t scale
because you would need to lay + connect + pressurize + maintain hundreds of miles of hoses. It’s high-CapEx.
A “watering robot”, meanwhile, can just do what a human gardener does to water a garden, “at scale.”
Picture a carrot harvester-alike machine — something whose main body sits on a dirt track between narrow-packed row-groups, with a gantry over the row-group supported by narrow inter-row wheels. Except instead of picker arms above the rows, this machine would have hoses hanging down between each row (or hoses running down the gantry wheels, depending on placement) with little electronic valve-boxes on the ends of the hoses, and side-facing jet nozzles on the sides of the valve boxes. The hoses stay always-fully-pressurized (from a tank + compressor attached to the main body); the valves get triggered to open at a set rate and pulse-width, to feed the right amount of water directly to the soil.
“But isn’t the ‘drip’ part of drip irrigation important?” Not really, no! (They just do it because constant passive input is lazy and predictable and lower-maintenance.) Actual rain is very bursty, so most plants (incl. crops) aren’t bothered at all by having their soil periodically drenched and then allowed to dry out again, getting almost bone dry before the next drenching. In fact, everything other than wetland crops like rice prefer this; and the dry-out cycles decrease the growth rates for things like parasitic fungi.
As a bonus, the exact same platform could perform other functions at the same time. In fact, look at it the other way around: a “watering robot” is just an extension of existing precision weeding robots (i.e. the machines designed to reduce reliance on pesticides by precision-targeting pesticide, or clipping/picking weeds, or burning/layering weeds away, or etc.) Any robot that can “get in there” at ground level between rows to do that, can also be made to water the soil while it’s down there.